State of Women's Education in Developing Countries, Idlustratingthe Extent of the Gender Gap in Education in Those Countries

I''"~~~~~~~~~~~~~~~~~~I

PHREE Background Paper Series Public Disclosure Authorized

FLoECOPY| Document No. PHREE/91/40

Women's Education in Developing Countries Public Disclosure Authorized Barriers, Benefits and Policy"

Elizabeth M. King and M. Anne Hill Public Disclosure Authorized Editors

Education and EmploymentDivision Populationand Human Resources Department The World Bank

Public Disclosure Authorized September 1991

Mhzpublication series senes as an outlet for background produictsfront the ongoir.g w-orkprogramn of policy .-esearch and analylsis of the Eduication and Employment Ditision in the Population and lluntan ResouircesDepartment of the World Bank. Tke lviewsexpressed are those of the author(s), and should not be attributed to the World Bank. Copyright, World Bank, 1991.

This volume has been accepted for publication by the World Bank. It is forthcomingfrom the Johns Hopkins University Press. Preface

Several noteworthy volum. s piblished in the 1980s have dealt with the topic of women's education in the Third World. These include the volumesby A.C. Smock, Women's Education in DevelopingCountries: Opportunities and Outcomes, New York: Praeger, 1981,and G.P. Kellyand C.M. Elliott (eds.) Women's Educationin the ThirdWorldt ComparativePerspectives, Albany. State University of New York Press, 1982, as well as regional studies by UNESCO. Smock reviewedlessons from existingliterature and presents selectivedata on five countries (namely, Mexico, Ghana, Kenya, Pakistan, and the Philippines) with the aim of identifyingthe f.actors affecting women's opportunities to participate in formal education and the effects of education on women's marriage, labor supply, and fertility behavior. Her work attempts to shed light and understanding on the complex relationships pertaining to this topic, and the conclusions drawn, though "tentative",provide a springboard for future studies. Kelly and Elliott assembled a collectionof previouslypublished papers on the participation of women in education, the factors affectingwomen's education in selected countries--somefocusing on specificfactors such as religion and textbooks, and some on aspects of women's education--and the influencesof female education on fertility and labor market outcomes. Together with Smock's volume, these papers reflect the knowledge and understanding about these topics at the beginning of the 1980s. How does our volume contribute to the literature? First, this collection of papers compiles more recent data on the status of women's education in the developing world, organized by region, and links these information to development indicators, including income per capita, mortality rates, life expectancy,and fertility decline. Second, it revisits questions pertaining to the factors that have sloweddown educational progress for women. It reviewsJhe findings of more recent social science research, with a focus on available quantitative studies, and identifies gaps in our knowledge. Descriptive and exploratorystudies often raise more questions than they answer partly because their analytical approach is usually limited. Many do not reveal the potential impact of specific policies or interventions on female participation and performance in school, and, hence, cannot guide clearly the practioner on the field. Future researmhin this area will have to consider methodologicalissues more seriously. Third, the volume examines the gains from women's education for family welfare and development. In addition to a review of the literature on the impact of women's education on child and maternal health, fertility,investments in education of their children, and on their own labor supply and earnings, the volume investigates some of these relationships using country-leveldata. The estimates reveal that how much education women receive is a significant factor determining economic growth and the qualityof life of their families,and that gender disparitiesin educatiou imply significant losses in national welfare. The gender gap in education is worthy of attention not only because of equity considerations but also because of its consequences for development. Lastly,the volume reviewspolicy reforms and interventionsby governments,donor agencies and non-governmental organizations (NGOs) aimed at improvingfemales' education in various parts of the world. The last chapter covers a broader set of specific program and project experiences than other studies have done previously. Acknowledgements

This volume is a collectionof papers commissionedby the Educationand EmploymentDivision and the Women in Development"ivision of the Populationand Human ResourcesDepartment, World Bank, for a project aimed at examiningthe problemsof girls' and women's educationin the developingworld. The papers were first prepared for an inter-agencyconference at the World Bank in June 1989. The project was undertaken under the general directionof Ann 0. Hamilton. The volume benefits from the ideas and advise of many. Ann Duncan managedthe project togetherwith ElizabethKing and helped to conceptualizethe volume. Helen Abadzi, Ila Patel and Maigenet Shifferraw wrote back,jound papers for the project on nonformaleducation and the participationof women in it. We are also gralefuI to Jere Behrman, Rosemary Bellew, Birger Fredriksen, Ann Hamilton, Barbara Herz, Lynel Long, Chloe O'G2ra, and AdriaanVerspoor for their commentsand support at various stages of the study. Four anonymc reviewers gave us excellent suggestionsfor improving the volume. CharleneSemer and H%lenWhitney Watriss edited the first drafts of chapters 3 through 7. Marianna Ohe took over f-:m them; with pati-nce, good humor, and imagination,she further transformedeach chapter. Althea Skeete-Comedyprovided superb secretarialsupport throughoutthe preparationof the volume; her excellent textprocessing,organizational skills. and industrywere critical to its completion. We are heavily in her debt. We also thank Cynthia Cristobalwho patientlyand competentlytyped in the final revisions. Lastly, we thank the many others who contributedto the individualchapters; they are acknowledgedin each chapter. Table of Contents

Page

Chapter 1. Women's Education in the Third World: An Overview ...... I1 M. Anne Hill and Elizabeth M. King

Chapter 2. Returns to Women's Education .43 T. Paul Schultz

Chapter 3. Sub-Saharan Africa .79 Karin A.L. Hyde

Chapter 4. Middle East and North Africa .111 Nagat El-Sanabary

Chapter 5. Latin America .143 Ines Bustillo

Chapter 6. South Asia .179 Shahrukh R. Khan

Chapter 7. East Asia .209 Jandhyala B.G. Tilak

Chapter 8. Educating Women: Lessons from the Past .251 Rosemary T. Bellew and Elizabeth M. King Overview 1 Chapter 1. Women's Education in the Third World: An Overview

ALAnne Hl and Elizabeth Flmg*

Improvingand widening accessto education has been a major goal of education policyin most Third World countries in the past two decades. This reflects the broad recognition that education contributes to development. Evidence is overwhelmingthat education improves health and productivity, and thait the poorest people benefit the most. However,this evidencealso indicates that when schoolsopen their doors wider to girls and women, the benefits multiply. While educating both men and women is essential to the process of development, failing to invest in women's education can even reduce the potential benefi.s that educating men can have on measures of social well-being. Indeed, failure to raise women's education to equality vith men's exacts a high development wst--in lost opportunities to raise productivityand income, and to improve the quality of life. Yet, female education is still much lower than for men in most developing countries, and many women and girls do not receive the type and level of education that will allow them to develop or to utilize their skills fully.

This chapter is an overviewof the state of women's education in developing countries, iDlustratingthe extent of the gender gap in education in those countries. It examines the linkages between the level of development and welfare of these counmriesand women's education, using extensivedata for 152countries over the time period 1960through 198f. Evidence from many countries (described in detail in Chapter 2) points to the strong relationship betwee:; the education of girls and women and national development. While most analysishas focused on the leiel of women's education, the implications of the gender gap in education on development have not been fully explored. Our empirical research begins to assess the considerable effects that gender disparities iP educational attainment impart for economic ayr! social weil being. Given the negative consequences ol gender differentials in education, the chapter proceeds to consider the environment within which educational decisions are made. In low income countries, much of the cost of investingin education is borne privately,yet many benefits to education are public. And gender differentials in the extent .o which costs and benefits are public lead to underinvestment in women's education. This chapter concluLdeswith a description of the remaining chapters in this volume.

State of Women's Education

Several indicators illustrate important patterns and trends in women's education in developing countries. Each of these indicators leads to the same conclusions: the level of female e.'ucation is low in poor countries, with just a handful of exceptions;and by any measure, the gender gap is largest in poor countries.

Adult literacy Rates

Consider first two common indicators, namely, adult literacy rates and school enrollment rates. Literacy is one of the principal goals of educational systems around the world, and the ability to read and write is

We are grateful to Joseph DeStefano for his excellent research assistance. 2 Overview

considered almost a basic human right. Yet low literacy rates prevail for women (figure 1.1).1 In 51 developing countries for which school data or estimates are available in the 1980s,female adult literacy is less than 20 percent in fourteen countries; in no country is male literacy rate as low. In particular, in Burkina Faso, Nepal, Somalia, Afghanistan, and Sudan, less than ten percent of adult women are literate. In these countries, the male literacy rates are three to four timet larger. Among those countries with male literacy rates of greater than 70 percent, the gender difference is notably large in Libya (30 percentage points), China (28), Zaire (26), Botswana (21), and Turkey (23). In the Dominican Republic, the Philippines and Colombia, the literacy rates for men and women are about equal.

Primaiy School Enrollment Low adult literacy rates are a result of past underinvestment in the education of women and thus do not necessarily reflect recent progress. Consider now gross school enrollment rates at the primary level, and the difference in the participation rates of males and females over time.2 Figure 1.2 illustrates the trend since 1950 for countries grouped accordingto GNP per capita; figure 1.3 shows the patterns and trend by geographic regiGn. Without question, s. hool -ollment rates at all levels have been rising in the developingworld, both for females and males. Howev..r, this expansion has not diminished substantiallvthe initial gender disparities. The enrollment rates of girls remain much lower than those of boys, the widest gap being evident in the lowest income countries. In the 40 lowest inco.ne countries that have GNP per capita below $500, the gap in primary school enrollment between boys and girls averages 20 percentage points. This gap has persisted in large part since 1960. Both enrolment rates and the gender gap in enrollment differ dramatically by region. Except for South Asia and Sub-Saharan Africa, all regions have achieved uear universal primary enrollment for males. However, only in Latin America and the Caribbean, and East Asia have the enrolment ratios of girls approached similar levels. Enrolment rates for girls continue to lag behind in the other regions, most notably in South Asia. In South Asia, the gender gap in enrollment has widened over the twenty-eightyear period as school expansion policies improved accessfor boys but not equally for girls.3 In a later section, we describe enrolment trends at the secondaryand tertiary levels. As countries achieve primary education for all, the gender gap becomes more apparent at the higher levels.

1 By definition, literacy rates indicate the proportion of the adult population who can read or write. These literacy data are obtained from censuses undertaken after 1980,and from UNESCO estimates for 1985for countries with no data for the eighties. (See footnote in Figure 1.1.) On the method used for estimation, UNESCO notes that the 'methodology used is basically an extrapolation and an interpolation of the illiteracy rates observed in the past. They do not, for example, take into account the impact of literacy campaignswhich are currently taking place ... in several countries. The use of literacy rates to measure the population's educational level has been criticized for several reasons. The data are usually self-reported, and thus, less accurate; and literacy is sometimes defined only with respect to a selected major national language(s). Nonetheless,better indicators that reveal the result of a country's past school enrollment rates on education are very costly to collect for an entire population.

2 Gross enrollment rates are computed as the ratio (expressedin percentage) of total enrollment in primary education to total youth population in the appropriate age group. Due to intake from later or earlier age groups into the primary grades, or due to grade repetition. gross enrollment rates can exceed 100 percent. In spite of this shortcoming,gross enrollment ratta ire a favored measure of educational progress because they reflect the intake capacity of the system.

3 Country-specificdata for each region are discussed in chapters 3-7.

1 1~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Overview 3 Figure 1.1 Women's UIteracy Rates Lag Far Behind Men's in Many Developing Co, mtries

Burkina Faso - Benin*- Somalia- Malil Nepal 1981 Niger' Chad- Pakistan 1981 - Haiti 1982 - Afghanistan' - Sudan - Bangladesh 1981 - Mozambique 1980 - Central African - Togo 1981 - Yemen, PDR - Burundi - Morocco' - India 1981 - Nigeria - Uganda - Rwanda' - Egypt - Tunisia 1984 Cameroon* Cote d lvo're- Algeria 1987 Ghana- Dominican Rep. 1981 Zimbabwe - Libya' - El Salvador 1980 - Brazil 1980 - Zambia - Kenya - Indonesia 1980 - China 1982 - Zaire - Malaysia 1980 - Boliviao - Botswana - Philippines 1980 - Myanmar 1983 Colombia 1981 Ecuador 1982 - Legend: Turkey 1985 - I _ _ Viet Nam1 - Female literacy Male literacy Peru 1981- Paraguay 1982 - Sri Lanka 1981 - Thailand 1980 -

0 20 40 60 80 100 Literacy Rates

Sn', .s: (1) D;.¢ derived from UNESCO, 1990. Whilb the critetia to determine whether a person is literate or not can differ amon,t eotuit-ies, the definition used her- .s whether a person can "with understanding both read and write a short, simple statemc'lt on his everyday life.' ( ) Countries marked by an asterisk (0) do not have census-based literacy data since 1980. Most have data going back to the sixties. We report UNESCO estimates for 1985. For all other countries, we report the latest survey- based data. 4 Overview Figure 12 The Gap Between Male and Fema: llEnrollmentRates i5 Widest in the Poorest Countries

LowIncome Countries Lower-MiddleIncome Countries

I l l l l l l l l l 1201 1201

Primarylevel ( so-80 s

M4c 60- 4 60

Secodary level 40 o d

O 20 w 20

90 C ______1960 1965 1970 1975 1910 1985 19 8 195 1970 1975 1900 1985 Year Year

Upper-MiddleIncome Countries

120

03 80.

a 60 this may eayyarfom986throgh989LEGEND rz ~ ~~~~~~~~~~~AMale Primary Level 0 0~~~~~~~~~~~~~FemalePrimary Level ui~ 20-~ ~ ~ ~ ~ ~~~~~)Male SecondaryLevel o Female Secondary Level

0 1S;0 19f ~16~0 1965 Isi ~ O Year,

Notes: (1) Gross enrollment data are derived from UNESCO; GNP per capita data are derived from the World Bank database. Although the latest year for which data are available after 1985 is noted as 1988, for some countries this may be any year from 1986 through 1989. (2) The grouping by income is based on 1988 levels of GNP per capita, and follows the World Bank classification. Overview 5 Figure 1.3 Gross Enrollmentin Primary and SecondaryEducation, By Region, 1960-85

EAST ASIA SOUTHASIA 1201 120 r

too- Primary 0oo so 80 _B.

60 60 Secondary _0 40 - ~40--

20 - 20-

I9k0 19'65 19'70 1YEA5 1980 1985 1988 19 Ug YEAR19'7 YEAA YEAA

MIOOLEEAST ANONORTH AFRICA SUB-SAHAAANAFRICA

120 - ,20 - 100 o L 80 - s

60 .I60

40 -40

20 20 '- --

0 ______190 195 1970 197 190 18 9816 5 170 1975 1980 YEAR YEAR

LATIN AMERICAAND THE CARIBBEAN

120

100- . s ~ LEGEND Males la sO- Females o

60 - Source: UNESCO Database

40 -

20 1-

90 1950 1965 19'70 19'75 1980 19.85 1se88 YEAR 6 Overview

Retention, Promotion and Completed Levels Gross enrojUmentrates, whichare usuallyreported for all primarygrades or all secondaryclasses, tend to mask somevery importantaspects of educationalprogress--the rate at whichstudents who enter school remainin school,how; many of them are promotedto the nextgrade, how manyrepeat grades,how much learningtakes place,and how many graduate. If high dropout rates prevaileven at the lower primary grades, then there exists a serious question of whether those who enter primary school ever achieve functionalliteracy. If so, muchof the recentgrowth in enrollmentratcs may not be reflectedin future adult literacyrates, and will ove,estimate educational progress. For this reasonit is importantto examinealso survivalor retentionrates and how they di!.'erfor boysand girls. Despite high gross enrollmentrates irz ..velopingcountries, fewer than 60 percent of those who enter schoolin the low-incomecountries, sno unlyabout 70 percentiv the lowermiddle-income countries reach the terminalyear of the primarycycle; moreover, primary school completion rates have declinedover the pas+decade in the poorestcountries (Lockheed and Verspoor1990).

Althoughdropout rates varyconsidc * fro_. _ountryto country,on average9.6 percentof girlsin low- incomecountries drop out of primary&- 31comp-red to 8.2 percentof boys;in Africa,dropout rates for girls average8.6 percent comparedwith 73 pe _nt for boys;in North Africa and the Middle East 6.0 percentof girlsdrop out comparedwith 43 percenmof boys(Table 1.1). In contrast,in LatinAmerican and Caribbeancountries, and in Lesotho,Madagascar, the Congo,and the Philippines,primary school girls are less likelyto drop out of schoolthan are boys (Bellewand King,1991).

Table 1.1 Primary School Dropout Rates, 1988

Girls Boys

Low-incomecountries 9.6 8.2 Lowermiddle-income countries 6.1 5.9 Upper middle-income countries§/ 6.2 63 Sub-SaharanAfrica 8.6 7.3 SouthAsia (Sri Lanka) 1.5 1.5 East Asia (Philippines) 6.6 6.7 North Africa& MiddleEast 6.0 4.3 LatinAmerica and the Caribbean 7.8 8.8

/ Data are availableon onlyfive countries in this group. Four of the fiveare African,North African and MiddleEastern countries. Source: UNESCOdata Overview 7 To illus,ratethe adverseeffect of dropoutson completedschooling, consider the expectedhighest schooling levelsimplied by age-specificenrollment rates in selected countries(see tabie 1.2)8 These levelsare estimatedusing a "syntheticcohort" approach--that is, summingup the age-specificenrollment rates for males and femalesobserved in a calendaryear to predictthe likelyeducational attainment of the entering cohort for that year. One obviousshortcoming of the approachis that it assumesthat the enrollmentrates of, say, today'sseven-year-olds will be the same when they rwch fifteenas those of today's15-year-olds. Thus, the method ignoresimprovements across youthcohorts in age-spec:ncenrollment rates. Despite the large increasesin enrollmentratios in most countries,expected attainment levels in low- incomecountries are low, especiallyfor females(with the exceptionof Lesotho,in Africa). For example, in Nepalin 1985,a six-yearold girl who enters schoolcouid expect to achieveot ly 3.1 yearsof schooling by the timeshe reachesage 18;and in BurkinaFaso, a six-yearold girl couldbe expectedto attain only1.5 years. In Nepal,the enrollmentrate of girls at the primarylevel has increasedfivefold since 1965;and in BurkinaFaso, by almostthree times. However,of girlsenrolled in Nepal,44.5 percent were in grade one, and only 9.6 percent in grade five;in BurkinaFaso, 26.1 percent were in grade one, and 13 percent in grade five. In bcth countries,the retentionreaes for boyswere higher. The differencein expectedyears of schoolingby gendervaries by incomecategory. The expectedyears of schoolingrange in 1985from means of 3.6 and 5.1 years for femalesand males in nine low-income countries,respectively, to averagesof 10.2 and 10.5 years in eleven upper middle-incomecountries, respectively.In the low-incomecountrie. the expectedyears of schoolingfor males exceedsthat for femaleswith the exceptionof Lesothowhere girls will attain 2.4years more schoolingthan boys, on average. In Nepal and Benin, the expectedattainment levels for girls are 4.4 and 3.5 years lower than fvr boys, respectively.In the middle-incomecountries, besides Bolivia, only the Middle Ea,e'and North African countrieshave significantlyhigher expected years of schoolingfor boysthan girls. l thosecountries, girls can expectto completeabout 2 to 3 yearsless schoolingthan can boys. Althoughgrade repetitioncan haveadverse consequences for girls' completedschooling levels, we did not find significantgender differencesin grade repetitionrates at the primarylevel from cross-countrydata. However,it is easyto misinterpretthis evidenceas proof that girls can perform at least as well as boys under prevailingschool conditions, since these rates can be estimatedonly for thosegirls who are able to enter and remainin school. It is entirelypossible that onlya verysmall, select group of girls (perhapsthe most well-offor the most able) enter school,and they performbetter than the averageboy in school. Becausethis indicatoris conditionalon accessto school,it must be used with care in characterizingthe state of femaleeducation.

4 These countrieswere selectedbecause of availabilityof data on age-specificenrollment in 1985in the UNESCOdatabase. 8 Table 12 Avera in Selected Coun

Country

Low-Income Countries Burkina Faso Nepal Bangladesh Burundi Benin Somalia Rwanda Lesotho Guinea

Lower Middle-Income Cc Philippines Morocco Bolivia Honduras Nicaragua El Salvador Botswana Paraguay Tunisia Costa Rica Syrian Arab Republic Upper Middle-IncomeCs

Hungary Poland Portugal Panama Algeria Venezuela Greece Ireland Trinidad and Tobago Hong Kong Iraq Overview 9

Secondary Education

Progress at the secondary level since 1960 has been dramatic in parts of the developing world, as figures 1.2 and 1.3 indicate. Gross enrollment rates for females have increased from an average of 12 percent in 1960 to 44 percent in 1988 in lower middle-income countries, and from 25 percent to 70 percent in upper middle-incomecountries. Enrollment of females over this period rose at a faster pace than that of males. A few low-income countries experienced a major setback in enrollment in the early 1980s, with a five percentage poilt decline for females and eight percentage point fall for males in China, for example. On the whole, however, the average female enrollment rate for this group of countries has quadrupled since 1960.s In 1960, the average enrollment ratios in all five regions did not exceed 25 percent, with Sub-Saharan Africa's regional average being 3 percent. After two and one-half decades of expansion however, the regional pattern in enrollment rates is more diverse. Tn 1988, the African and South Asian average enrollment rates lag behind East Asia by about 40 percentage points. Indeed, the averages for Sub- Saharan Africa are still below the ratios that East Asia achieved25 years earlier. But the regional diversity has been greater for females. While the gender difference in secondary enrollment has narrowed in East Asia, it has widened in Sub-Saharan Africa, North Africa and the Middle East, and South Asia (since 1965).

At the secondary level, most students can choose to follow either the general academic track or the vocationalstream, where the latter includes also teacher training programs. On average, female secondary school enrollment is heavilyweighted towards general education, accounting for four-fifths to nine-tenths of all girls enro2ed. Hence, the female share in vocational-technicaleducation (VTE) has generally been well below 50 percent in developing countries,except in Latin America where women make up one-half or more of total VTE enrollment (UNESCO, 1983). The compositionof non-general enrollment has changed over time, with the share of teacher training programs declininggreatly for Asia and the North African and Middle-Eastern countries since 1970 relative to other VTE programs (see table 1.3). The small and decreasing proportion of female secondary school students going into teacher training in all regions except Africa partly accounts for the shortage of female teachers in low-incomecountries. Much importance is accorded to the supply of female teachers in discussionsof improving the education of girls and women. Indeed, although thc relationship between having a female teacher and girls enrollment and school performance has not been adequately examined,reported anecdotal evidenceappears to support the view that female teachers are a positive force for raising female education. Yet, In low-income countries, the share of females in the teaching force at all levels of education is about one-half that in higher incon,#.groups. At the primary level, in 1985,just about one-third of teachers are women; at the secondary level, less than one-fourth; and at the tertiary level, just over one-tenth. In lower middle-income countries, there has been little change over time at the primary and secondary levels. Female teachers comprise about one-third of the teaching force at the primary level and one-third at the secondary level. At post-secondary levels the female share rose from 17 to 24 percent. In upper middle-incomecountries, the majority of primary school teachers are women, making up two-thirds of the teaching force. In those countries, the largest increase for female teachers has occurred at the secondary and tertiary levels. The female share at the secondary level is approaching one-half and at the tertiary level has exceeded one-fourth.

5 Since nearly all countries saw an increase in their enrollment rates over the last two and one-half decades, those that experienced a decline, especially among the low-income countries, deserve some mention. Besides China where enrollment fell in the 1980s,enrollment also fell in Bangladesh (by 1 pe-centage point for males); Nepal (by 1 and 6 percentage points for females and males, respectively);Ghana (by 2 and 3 percentage points, respectively);Guinea (by 2 percentage points for both males and females); and Togo (by 6 and 18 percentage points for females and males, respectively). 10 Overview Table 13 Percentage Distrbution of Female Enrollmentat the SecondaryLevel By TIpe of Education

1970 Most Recent YewW Re8ion General Teacher Vocational General Teacher Vocational training training Sub-Saharan 85.2 6.4 8.4 83.1 6.1 10.8 Africa LatinAmerica 79.7 4.7 15.9 78.0 13 20.7 & Canbbean Asia 85.7 7.4 6.9 92.1 0.8 6.9 North Africa 86.6 8.7 4.7 91.1 2.7 6.2 & MiddleEast II The most recentyear for whichdata are availableusualy pertain to early to mid-1980s. Source: UNESCO

Higher Education In tertiary education,the broad trends and patterns amongregions and betweenthe sexesare similarto thoseat the secondarylevel (table 1.4). Enrollmentrates for malesand femalesin Sub-SaharanAfrica and SouthAsia are also far belowthose in the other regions. The averagemale enrolment ratio in East Asia is higher than in any other region,but LatinAmerican and the Caribbeanled in femaleenrollment. In 1988,the averagemale enrolment rate in East Asiawas 2.5 percentagepointc higher than in LatinAmerica (at 18.9versus 16A percent), but the femaleenrollment rate slightlylower (14.6 versus 16.2 percent). The MiddleEast and North Africa averages are closer to Latin America'sfor males, but as much as 4.4 percentagepoints lower for females.

Chapters3 through7 presentthe distributionof femaleenrollment in highereducation according to fields of study. It is not surprisingthat they found female enrollmentto be heavilyskewed towards the humanities,home economicsand the arts. But womenhave gainedground in businessand administration studies,and are expandingtheir numbersin science-relatedfields and medicine. With increasesin economic growthand greater opennessin the labor markettowards women, continued progress can be expectedin women'sparticipation in highereducation. Over%iew 11 Table 1.4 Enrollment Rates at the Tertiary Level, 1965 and 1988

CounDy Groups Females Mates 1965 1988 1965 1988 Low-Income 0.2 1.1 0.7 2.8 Lower-middle-income 2.1 103 3.6 14.1 Upper-middle-income 3.1 15.6 5.9 18.3 Sub-SaharanAfrica 0.1 1.1 0.6 3.3 North Africa & MiddleEast 2.1 11.8 5.6 16.2 SouthAsia 0.6 2.0 2.0 5.2 East Asia 4.1 14.6 6.7 189 Latin Americaand Caribbean 3.1 16.2 5.7 16.4

Notes: Althoughthe latest year for whichdata are availableis noted as 1988,the most recentyear for some countriesis anyyear from 1985through 1989. Source: UNESCO data

Table 1.5 Proportion of Females in the TeadcingForce, By Income Groups

Income groups 1965 1970 1975 1980 1985

PrimaryLevel Low Income 0.20 0.24 031 0.30 032 Lower MiddleIncome 0.47 0.50 0.385k 0.51 0.49 Upper MiddleIncome 0.63 0.58 0.62 0.61 0.65 SecondaryLevel Low Income 0.24 0.23 0.28 0.21 0.23 LowerMiddle Income 0.34 032 0.34 0.36 0.34 Upper MiddleIncome 037 0.41 0.42 0.46 0.47 HigherLevel Low Income 0.12 0.11 0.15 0.15 0.14 LowerMiddle Income 0.17 0.18 0.21 0.19 0.24 Upper MiddleIncome 0.18 0.19 0.23 0.26 0.28 l/The sharp drop for this year is due to missingdata for a countrywith a much higherfemale share in other years than other countries. 12 Overview

Economic Growth, Family Welfare, and Women's Education The measuresof educationalattainment discussed above each revealed clear gender disparities. Yet these genderdisparities are costly--indevelopment terms. Development,broadly defined, encompasses economic progress and improvementsin the overallquality of life. The followingsection assessesthe effectsof gender disparitieson a country'seconomic and socialwell-being, as measuredby the levelof GNP per capita as well as by severalsocial indicators of familywelfare in the aggregate. These indicatorsinclude life expectancy(at birth), infant and maternalmortality rates, and total fertilityrates. A principalquestion that faces policymakersand developmentspecialists alike is whether growth in aggregateincome and productivityare sufficientto ensure a better qualityof life for present and future generations.Are the poorestcountries neces&arily the least developed?Do increasesin incomeguarantee an improvedquality of life? What does femalee lucationcontribute to development?How is the gender gap in educationrelevant to this relationship? There is a growingbody of literaturewhich examines the benefitsof educatingwomen. Muchof thiswork is describedby T. Paul Schultzin the foliowingchapter. A more educatedmother raises a healthierfamily, sh can better applyimproved hygiene and nutrition practices. She has fewer and better educated children. She is more productiveat home and in the workplace. Educationcan substitutefor communityhealth programsamong womenwho are informed about healthcare and personalhygiene, and alsocomplement such programsvia an increasein incomeand in the recognitionof the value of these services.Studies in the Philippinesand Nigeriaconfirm the finding that mother'seducation affects child health, with largest benefits experienced by youngerchildren (Barrera 1990;Caldwell 1979). So importantis the influenceof mother'seducation on childmortality that education mitigatesthe impact of the absence of medicalfacilities in the community.A studyon India provides additionalsupport of this relationshipby demonstratingthat higherfemale education and wideravailability and utilizationof medicalservices are two crucialfactors associated with the lowerinfant mortality in the state of Kerala (Jain 1985). For Malaysia,mother's educationwas indeed foundto have a marked effect on infant mortalityrates, particularlyin the secondsix months of life (Butz 1982). In contrast,family income,though generally associated with infant mortality in cross-countryanalyses, was found to bear a less critical relationshipto infant mortalitythan was education; incomewas discoveredto be altogether unimportantafter taking into account mother's education. A study of the determinantsof chronic malnutritionamong children in the Philippinesfound that mother's schoolingand the availabilityof safe drindingwater explainedhealth differences among children, while hc *seholdincome did not. Moreover, mother's schoolingappeared to attenuate the aegativeeffect- of poor; onmunity sanitationand water supply(Barrera 1990). Studieson Nicaraguaand India repoi: similarfindings (Behrman and Wolfe1984, Behrmanand Deolalikar1987). Thesebenefits are also clearlyevidernt from aggregatedata. Figure1.4 displaysscatter plots of l_velsof GNP per cal-itaand socialindicators in 1985against female enrollment rates in primaryeducation in 1975. The plots illustratethe expectednegative correlation between education and fertilityand infant mortality, and the positiveassociation between education and incomeand life expectancy.They arguefor increasing female enrollmentrates at the primarylevel as part of any developmentstrategy. Overview 13 Figure 1.4 Higher Female Educ&donImproves Economic Productivityand Social Welfare

US1980 dollaas, In log values Years

10 0 in I wo o OD Wno o0t 0 o 80X '.4~~ ~ ~ ~ ~ ~ ~

e6 ° °4 / .EC 60 o % o

40 o o.t U w 0

6 0 0 00 a 00° ° 4-40 0 0

(0 0

4 ° 20- d 2A 4 6'0 eb toa io d 2b 4b fb 60 I a 120 PrimwyEnrollment Aate, 1975 PrimaryEnrollment Rate, 1975

Deethspe 1000live bIette Children , I , _ I , I , I in 200 - S 0a 1

to o~~~P 0o 0 0 0 0~~~00 0 0 10008 aQo 0 Cc 4

0 0 0 4J~~~~a.2

00~~~~~ C 0 ______.

d 2b 4b 6b Oh ida t~ d 2b 4b ab Oh ida ilo PrimaryEnrollment Rate, 1975 PrimaryEnrollmlent Rate, 1975

Notes: These graphs plot 1985 values of indicators against 1975 data on female primary enrollment rates. 14 Overview The Effect of the Education Gender Gap As willbe revealedin the analysisthat follows,not onlythe levelof women'seducation but also the gender disparitiesin educationalattainment matter. (Box 1.1 listscountries by size of gender differencein their enrollmentrates at the primaryor secondarylevel.) 6

Boi 1il Which Cointries Have the Largest Gender Gap?

Between1965 and 1985,considerable progress has been made in reducingthe gender disparity inprimary enrolments Althoughthere were 19 counttiesi which1965 female enrollment rates wereless than 42 percentthose ofmales, ony Chadand Yemenremained in that category in 1985. Similarly,the'number f count'ieswit femaleto mal ratios at ltetween42 and 75 percent fell from 37'in 1965to 17 in 1985. And s' is evidentfrom this table, the rhajorityof these countriesare low income. For the analysisiln'the rest of the chapter,we coibine informationon gender disparitiesin primaryenrollments with those for secondaryenrollments to yield one measure of the gender disparityi educationfor a givencountry at a point in tme. Thus, what we henceforthrefer to as t-hegender gap isthe rato of femaleto maleenrollment at the primaryor secondarylevl whicheveris smaller

Table box-1.1 Comtries With Large Gender Differences in Primaty Eirolment Ratios, by Levd of Femaleto Male Ratio and Year

. <-4e 0.42 to0.75 c: 042 tO,7s

Athanitan - ludria Malat --- d Afghnistan Biurian Angola Mali Yemen Benin Burundi -Bngladesh .hutan :MB .cntral AfricanlRep. Benin -'Burkblqun -lina Faso -C:haid : BaliMia-. . Niga . --.- E:--:::. CentralAfrican Rep. :omoros Bulla lPaso. :Nige : - tilopia lDemcraticYet~nsCawnetoon - apa WENew Oca .Gambia -iago .. Ci- . RWa GuineaBissau G@ambia.'',. . :'acote llvoite Senega Guinea ,uin'eBisu -. ' :'t . -SS-err Mali Liberiia E.. .i3quatorialGuinlea ...... Sudanl Mauritania.: -: ...... - . . libya -hana .,rocco oSyria AnM,uriia' Cuinea -a - . . Niger India t :Wa....:.... - -- ~~~~Istan: . :.:: :negal SaudiArabia Iraq Ua.', ' Soma Somalia lCamichea 'ir :''.Tog:',;0'.',ap Ib5- Keya : - Zimbab:w'':: :ee . . -oLeE PDR

6 The cut-offpoints of 0.42 and 0.75indicate the twenty-fifthand fiftiethpercentiles in the distributionof countriesranked from higherto lowerlevels of gender disparity. Overview 15

The numberof cnfrier dithlarge gender gaps.i dary enrollmentsexceeds that for the primarylevels. In 1965,there were:45 ountriesin wich femaleenrollments represented 42 percent or less the levelof male e ents and I an additional27 counties, the ratio was between 42 and 75 percent. However by 18t number of countrieswith wide gender dispaties in enrollme fe to 16 (lessthat 4 percentand 27 (between42 and 75 percent).:-. -. :.- .: .- ; ...

Table box 12 CountriesWith Lage6:Gnder Differencesin Secondaxy EiBnrmentRatios, by Levelof Femaletol Ratio and Year

1*5 2~~~~~~~~~~~~~985 v.0U.4 6. 42wto . ,. -'C0.42 0.42w 0.75

Afghanistan Nepal Albania Bangladesh Afghanistan Eangl.dech Niger Al%pria Benin BurkinaFaso Beajit Pakistan, Ang.a Bhutan Burundi Cameroo- Papua NewGuinea Botwna CentralAfican Rep. Cameroon CentralAfrican Rep. Rwanda BuridnaFaso Chad Comoros Cad . Sa9udiArabia Bumia Cote DIvoire Egflt mortos Senegal Bumundi Gambia Ethiopia Congo Siea Leone Denmark GuineaBissau Ghana cote DJoire Somalia GiCuinea India DemocaticYcmtn Sudan Gambia Malawi Iran .gAypt . .SyCaa Guat-mala Mauritania laq qatorwil Guinea iti KanaiKenyaaMoambique UtbioApi Iran:;:: -: lao PDR ;abon TunTiia Joa Pakstan Mali Ghana Turkey Ko-.a Togo Mauritania Gunea Upada tao #DR Yemen Onian 1 .ia SZair Lebanolt - Saudi Arabia Indonhalbia M ada : .Senegal Iraq . . : M*ii Somalia Raaip*bea . MauretIw: .:-Syria Keyf..Mexico . - ::...Tanzania LIbeia - Mogarbique .:: - -.Tunisia Uba :ff:--Nigeria Turkey Malawi Porttapl - 71:E. Di::tUganda .Mali . : . . . Spiain . .::V . 0n -. - X - :Zaire Maurita** . . ITbaiad Zambia Moron~ Zi....mbab.we0:...... f:.0 .. Zimbabwe

As a first look at the potentialeffects of the gendergap, consideragain the scatterplots of femaleprimary schoolenrollment rates againstfour developmentindicators. This time, figure 1.5 distinguishes the countries bythe sizeof the educationgender gap. To emphasizethe relationshipobserved, regression lines have been drawn for the two subsetsof plotted points. The first subsetrefers to countriesin whichthe gender gap is wide,with the ratio of femaleto male enrollmentrates in primaryeducation (F/M) being less than or equalto 0.75;the secondsubset pertains to countriescloser to parity,that is, in whichthe ratio exceeds0.75. 16 Overview Figure 1.5 Larger Gender Gap SlowsEconomic Growth and SocialDevelopment

US180 l1Iws6,in og Yom Yrj

10 0 02 A 0

B 0

.d20 40 6~0 2 0 4 D 8 LO0 2A0 I -J~~0

4frX90\ ''4 UD\i1^10 200- x 0 20 40 00 80 100 12 0 200 0 40 50 0 100 120 Priw'yEnollsnt Rate, 1975 Pria'yEnroIl.enl Rate, 1975

a 6s- t 8 A <^

Notespop 1000 live bi95 Caildna ( Ih I I I --I I h I I I I I I r 200 0 0 0 0 0

S p c6 hv a

0,I~~k" t4oo 4~i

50 a, 2~0 C~~~~

0o 20 4 0 8 d i0 2- 0 4 0 8102

petanst cutrles wih asalrgedrgp PviwkyEnroUsefl Rate, 1975 Pr1nrVErrollett Rate,1975

Notes: (1) T'hesegraphs plot 1985values of indicatorsagpinst 1975 data on femaleprirnary enrollntent rates. (2) The countriesrepresnted by circleshave a largereducation gender gap-that is, the fenmaleenrollment rate is lensthan 75 percentthat of males-whilethose represnted by triangleshave a smallergender gap-that Is, the femaleenrollment rate Is better than 75percent that of males. (3) The twolines ane regresion lines estimatedfrom the scatterplots of the twogroups of countries. The thickerline pertainsto countrieswith a smallergender gap. Overview 17

In each graph, the levels of fertility and infant mortality associated with a particular level of female education are much lower, and GNP per capita and life expectancy,much higher, in countries with greater equity between the sexes. Since these observations hold even for countries that have achievedgross female primary enrollment rates of 100 percent, the gender gap at the secondary level is apparently quite important too. Note also that in the graph for total fertility, the slope for countries with a larger gender gap is significantlyflatter than for countries with greater equity. This indicates that the widely known negative effect of education on fertility will be weaker when the education gender gap is larger. Therefore, in order to reduce total fertility rate in the future, a larger increase in female education levels will be required in those countries. These graphs stronglysuggest important benefits for development goals to improvingwomen's education. However, simple correlations can be misleadingbecause the level of female education itself is related to other factors such as national income which itself influences social welfare. To estimate correctly the magnitude of the education effect, the influence of all variables that determine income and social indicators must be considered concurrently. In the followingsection, the contribution of female education will be explored using multivariate analyses for more than one hundred developing countries.8 Moreover, to establish causality rather than simple correlation, it is important to use measures of past investments in female education.9

The Impact of Female Education on Development The full effect of female education on the development indicators can be broken down into indirect and direct effects. Education bears a direct relationship to social well-being.10 As discussed earlier, women's education influences infant mortality perhaps because educated mothers are more likelyto appreciate the importance of prenatal and neonatal care and, thus, more likely to be users of such health care services. But women's education also indirectly improves infant survival through higher market productivity and income for women, and with higher income, better living standards.11

7 Gujarati (1988, page 188) provides one of the clearest examples of the benefits of multivariate analysis by way of the followingagricultural example. Suppose that the simple correlation between crop yield and rainfall indicates that there is no relationship. However, both crop yield and rainfall are related to a third variable, namely temperature. By holding constant the effect of temperature on crop yield as well as the effect of temperature on rainfall, a multivariateregression of crop yield on rainfall and temperature reveals the expected positive net effect of rainfall.

8 Appendix IA describes in detail the data used for this analysis.

9 Past education investments,even those undertaken ten years back, could neverthelessbe correlated with current levels of investments. These investmentsfeed on themselves due to social and political pressure. However, the degree of correlation can be expected to be less and to differ across countries.

10 Chapter 2 describes these effects in detail.

11 Other indirect effects may work themselves out through variables other than the level of income. Consider the effect of total fertility rates on the level of infant mortality or of infant mortality rates on fertility behavior, for example. Research in this area (see the review by King 1987) has argued that these two variables are interrelated, and that the appropriate model for identifyingtheir determinants would be a simultaneous equations framework. As a result, the indirect effects of female education should also include its effect on fertility behavior (infant mortality) via its impact on infant mortality rates (fertility). We have simplifiedour empirical model, omitting the estimation of these other indirect effects, due to lack of variables satisfyingthe identification requirements of such a simultaneous equations model. 18 Ove. view

We estimate an empirical model which examines both the direct and the indirect effects of female education. The model includes two recursive equations for each social indicator considered.12 The first equation estimates a country'sgross national product (GNP), the determinants of wh.ch are modelled within a production function framework. A second equation specifiesthe factors that determine the level of each social indicator.13 Both equations include female education as an explanatoryvariable. The level of female education is measured by gross enrollment rates in primary and secondary education. While there are shortcomings to using these rates to index female education, they are commonlyused and widely available. An ideal measure would be one which reflects the actual stock of knowledge and skill in the female population; such a measure is not available for the countries considered.14 To represent the educational attainment of the current population rather than current investmentsin education, we use enrollment rates lagged by ten years for primary education and by five years for secondary education.15 Each equation also includes a measure of the gender gap in education as an explanatory variable. Based on the inferences which we may draw from fgure 1.5, we hypothesizethat at a given level of enrollment, the educational gap between men and women has an independent effect on social indicators. Yet the mechanismsby whichgender inequalityin education affectsdevelopment are not well understood. Amartya Sen (1984) illuminates the potential for conflict that men and women may have in distribution of family resources. It is possible that the gender gap in education affects the relative earnings potential of husband and wife and, thus, the division of labor between them. Within a family,a husband who is more highly educated than his wife is also likelyto command a higher wage in the labor market. To benefit most from this situation, the husband and wife willspecialize more in their respective familyresponsibilities, increasing the husband's hours of wnrk in the market at the expense of home work and the wife's home activities at the expense of her market work (see for example, King and Evenson 1983). As a result, the wife and the rest of the family can benefit from the returns to her husband's education through income transfers from him, but they cannot benefit as easilyfrom the non-monetary returns to his education, such as greater skills and technical knowledge. Consider the case of many Sub-Saharan African countries where men and women also maintain separate budgets, and where there are well-establishedconventions concerning which expenditures are to be met from each income. Although men are expected to support their wives, this support varies widely among demographic groups. This separation of budgets implies that women's expenditures--fortheir farms, their

12Except for a shift factor obtained using dummyvariables for detailed regions, these estimates restrict the coefficientsto be equal across all countries. Earlier empiricalwork allowedthe education variables to vary by region, however, test statistics revealed that the regional slopes were not all significantlydifferent in each equation.

13 In addition to the levels of primary and secondary education and measures of gender disparity in education, variables employed in the regression models include: accumulated gross domestic investment used to measure the stock of physicalcapital (to predict GNP); the size of the labor force (to predict GNP); the ratio of population to physicians;and percent of population with access to safe water. However, each model does not include all of the variables. In addition, all the models contained control variables for year and detailed region. Appendix tables A3 and A.4 display the estimation results for GNP and the social indicators, respectively.

14 Lau, Jamison and Louat (1991) have recently estimated education levelsfor the labor force in about sixy countries using enrollment data. Their data were not publiclyavailable while our analysiswas being done. Their estimates are based on annual school enrollments for the primary and secondary levels, which are themselves backward trend extrapolations of actual enrollment data.

15 Sensitivityanalyses supports this variable specification. The correlation between adult literacy rates, which is a stock measure, and primary enrollment rates improves significantlywhen lagged values of enrollment rates are used. The rank correlation coefficient between these stock and flow measures of education increases from 0.70 to 0.82 for a sample of 35 countries with 1980 data on both variables. Overview 19

children and themselves--can be severely limited by their own productivity and access to credit and technology, which are influenced by their own education, not their husbands'.

Another explanation for the gender gap effect is that the wife's role in decisionmakingmay be weaker than the husband's when the husband's education is much greater than the wife's. Little control by women over their reproductive outcomes and over the allocation of resources for child care could mean larger families and poorer health status of children. A large gender gap in education, thus, leads to significant welfare losses.

The empirical results from estimating our recursive models are described in detail in appendix I.A to this chapter; a brief overview is provided here. Our findingsare consistent with others in that we estimate the level of education to have a strong positiveeffect on GNP. (See,especially, Lau, Jamison and Louat, 1991.) On the gender disparity in education, our GNP equations indicate that for given levels of female education, the size of a country's labor force and its capital stock, those countries in which the ratio of female to male enrollments is less than 0.75 can expect levels of GNP that are roughly 25 percent lower than in those countries which are otherwise similar apart from the level of the gender gap. That is, large gender disparities in educational attainment actually appear to reduce GNP.

GNP is but one measure of a country's well being.16 Wu consider in addition the effects of women's education on male and female life expectancv, infant mortality rate, maternal mortality rate, and total fertility. Table 1.6 extractsthe estii..ated female education effects from the complete set of empiricalresults (see table A.4).

These results indicate that both the level of female enrollments and the gender disparity in enrollments influence social well being, even after accounting for intercountry differences in GNP. They confirm what was observed from the scatter plots: higher levels of primary and secondary enrollments are associatedwith longer life expectancy (with comparable benefits experienced by men and women), lower infant and maternal mortality and lower total fertility rates.17 These coefficients represent the net effect of the education variables on the social indicators,and can be interpreted as follows. Consider, for example,infant mortality. The coefficient of -0.41 implies that after taking into account the effects of GNP and other factors likely to be associated with infant mortality, an increase in female primary enrollment rate by 10 percentage points can be expected to reduce the infant mortality rate by 4.1 deaths per 1000 live births. If female secondary enrollment rose by the same amount, an additional reduction of 5.6 deaths per 1000 live births would be experienced.

These results further demonstrate that even after accountingfor the effectsof GNP per capita and the levels of female enrollments, a large gender disparity in educational attainment reduces social well-being. Consider first the life expectancy of males. Men living in those countries in which the female/male enrollment ratio is less than 0.42 would experience reductions of nearly 4 years in their average life expectancyrelative to men in those countries that were otherwise comparable apart from having a smaller gender gap in schooling. Similarly striking results hold for each social indicator. As countries approach

16 And GNP may be measured only partially, since most accountingignores the value of home production or production in the non-monetized (or informal) sector. According to a 1975survey of the coverage of nonmonetary activities reported by Goldschmidt-Clermont(1987), countries differ widely with respect to what activities are included in the national accounts. Among 70 developing countries, all claimed to fully cover own-account (or subsistence or non-cash) agricultural production and rental incomes of owner- occupied dwellings. About two-thirdscover other own-accountprimary production, such as fishing,forestry, and house-building. About one-half coverown-account food processingand handicrafts. Only six countries include water collection; three include crop storage; and two include domestic activities.

17 Data for maternal mortality are available for only nine countries in 1985, consequently, these results exclude1985, implying a more limited sample. Only primary enrollments bear statisticallysignificant effects on maternal mortality. 20 Overview

greater genderequality in enrollments, the detrimental effect of the differential diminishes,with a gender gap of between 0.75 and 0.96 imparting effects which are still negative, but not statisticallysignificant.

Table 1.6 Effects of Female Education and the Gender Gap in Education on Social Indicators

Female Life Male Life Infant Maternal Total Education Variables Expectancy Expectancy Mortality Mortality Fertility

Female Primary 0.10 0.09 -0.41 -3.31 -0.01 Enrollment

Female Secondary 0.12 0.11 -0.56 3.02nS -0.03 Enrollment Female/Male Enrollment -4.80 -3.85 21.16 99.82ns 0.72 Ratio <0.42 0.42-0.75Ratio -3.41 -2.75 11.37 111.84'W 0.73 0.75-0.95Ratio -O.69ns -O.52ns 1.88ns 82.56 ns 0.26ns

ns Not statisticallysignificant at a 90 percent or greater level. Note: The other variables included in these regression estimates are given in Table A.4 of this Chapter.

We illustrate the significanceof these results by conducting the followingpolicy simulations. We assume that gender disparities in education are reduced by raising the levelof female enrollment, then consider how this increase will influence national income and social well-being. Since our measure of the gender gap depends on whether the enrollment disparity is greater at the primary or secondary level, we increase the level of female primary enrollment in ten percentage point increments if the female-male enrollment ratio is lower (or the gender gap is larger) at the primary level. However, if the gap in secondary education is larger, closing the gender gap would be more costly because increasingsecondary school enrollment would entail larger resources. We examine the effects of raising female enrollments in secondary education by five, instead of ten, percentage point increments. Once a country reaches parity in both primary and secondary levels,female enrollments are raised no further. The results of the simulations are summarized in figure 1.6.18

18 Given the limited data on maternal mortality, we focus on infant mortality, fertility, and male life expectancy. Results for maternal mortality and female life expectancyare comparable.

I.~~~~~~~~~~~~~~~~~~~~~~~~~~ Overview 21 Figure 1.6 What Closing the Gender C-ap in Education Would Bring. Simulation Results

Infant Mortality Rate Deaths per 1,000live births 140 -

120

100

40-

20- 0 Low Lower middle Upper middle Upper Income Group

Baseline 3 10%and 6% L 20% and 10%

Male Life Expectancy Total Fertility Rate ChildrenpWr woman

80 -

4, 40-

*0- 2

o Low; Lowr middle Upper middle Uppir Low Lower middle Upper middle Upper Inoome Group Ircome Group

111111Bs eile 0e end 5S O 20%and _- SieellnIn 10%end 6% =D 20%end 10% 22 Overview

Qualitatively, the simulation results are the same for all income groups of countries. However, the magnitude of the effects differs, with gains usually being larger in poorer countries. To illustrate this, consider the simulation results for infant mortality rates. In 1985, the average infant mortality rate in our sample of low-incomecountries was 126deaths per 1,000live births, nearly 60 percent greater than the rate in lower middle-incomecountries and more than double that for high-income(i.e., mostly OECD) countries. Had the female enrollment ratio in low-incomecountries been 10 percentage points higher at the primary level in 1975 (or 5 percentage points higher at the secondary level in 1990), the average infant mortality rate in those countries would have been 7 percent lower than its actual 1985level. In comparison, if female enrollment ratios were that much higher in lower-middleand upper-middle income countries,their average infant mortality rates would be 6 and 9 percent lower than their actual 1985 levels, respectively. One important conclusion we draw from these analyses of country data is that both the level of female education and the gender gap in education are each important determinants of (aggregate) family well- being and economic growth. The benefits of improving female education go beyond increasing individual productivity and income. By decreasing desired fertility, population pressure eases; by improving the family's health, life expectancyincreases and the quality of life rises, not only for the family, but also for the community. Indeed, our results demonstrate that a country's failure to raise the education of women to levels equal to those of men imposes substantial cost for their development efforts. Unless the gender equity in education is improved, desired improvement in social indicators can be achieved only at much higher levels of economic growth.

Why Do Gender Differentials Persist?

If women's education is so important, why do women remain undereducated when compared to men? Why do gender differentials in education persist? The following section presents a conceptual approach for organizingthe factors that can explain the current state of women's education and for understanding the dynamicsunderlying educational outcomes. We propose that gender differentialsin education endure since those persons who bear the primary costs of investingin schooling fail to receive the full benefits of their investment. This is particularly true because of the broad social payoffs in educating women. There is an expanding body of empirical work on the determinants of school enrolLmentin developing countries. The contribution of this literature is that it disentangles the many and complex influences on school enrollment. The studies are based, implicitlyor explicitly,on a framework which treats education as a family or individualdecision entailing current costs and future benefits. The decisionmakerweighs the benefits, net of costs, from spending family resources on education against the net benefits of staying out of school. Costs broadly include direct financial costs, indirect or opportunity costs, and non-pecuniary costs, which are borne privatelyby the parents or the student in the absence of public education. Education yields returns both to the family (in higher earnings for the children, some share of which is remrned to the parents) and to society (by raising dimensions of the quality of life, as evidenced in the preceding sections). Costs are often measured by the availabilityof, or distance to, school for lack of better data on costs that each student or parent faces.19 A few studies have estimated the effect of the opportunity cost of schooling on enrollment or attainment in developingcountries and have found a negative relationship (for example, Rosenzweig and Evenson 1977). Unlike costs, benefits do not usa l-ly find their way into empirical studies

19The presence of primary and junior high schools in the communityhas been found to increase enrollment and years of completed schooling in Brazil (Birdsall 1985a),Indonesia (Chernichovskyand Meesook 1985), and Nepal (Moock and Leslie 1986). Distance to a primary school was found to be negativelyassociated with enrollment and schooling attainment in Egypt (Cochrane, Mehra, and Osheba 1985),the Philippines (King and Lillard 1983),Nepal (Shrestha and others 1983),and in Thailand (Cochrane and Jamison 1983). Similar effects have been found for secondary schools in Malaysia and the Philippines (King and Lillard 1987). Overview 23 of school enrollment,primarily because of measurementdifficulties.20 Improved productivityin the workplace,as measuredby expectedearnings in the labormarket corresponding to givenlevels of education, is easier to calculatethan other retuns to schooling,such as increasedfuture productivityin unpaidlabor, greater efficiencyin takingcare of childrenor of one's health,or enhancedability to deal with p.oolems or "disequilibria"in one's daily life (Schultz1975). But, howdo the costsand benefitsaffect the schoolingof men and womendifferently? What accountsfor gender differencesin the amountof schoolingparents are willingto investin their daughtersand sons?

A Model of Schooling with Gender Differences The retuns to schoolinggo first to the student. Yet, the decisionand the resourcesusually belong to parents especiallyin the early schoolyears. It is, thus, the perceptionof parents whichmay be the key factor. Parentsmay havedifferent preferences regarding their sons' and daughters'education. 21 Parents tend to favorsons in certainsocieties, not onlyin educationbut sometimesalso in the allocationof food at mealtimeor the distributionof inheritance(Greenhalgh 1985, Rosenzweig a:,d Schultz1982). These behaviorsmay not be discriminatoryin themselves.This is not to say that they are any less perniciousor that sex discriminationin the home does not exist,but that the unequaltreatment of sons and daughters maybe a rationalresponse by parents to constraintsimposed by povertyand to expectedreturns determined by labor marketconditions and tradition. Whenthe expectedreturns to sendingdaughters to schooldo not exceedthe costs . doingso, then female educationas an investmentbecomes unattractiveto parents. Daughterswill then be educatedto the extentthat parentsthink theyshould be givenlow economicreturns. SeeAppendix 1.B to thischapter for a formaleconomic model that incorporatesthe effectsof thesefactors.

Fbwnci4 Opponwty ad Tsychic Costs Even when educationis publicand tuitionis free, schoolattendance still entailscost outlaysfrom family resources,both at the primaryand postprimarylevels. Contributionsto the school,learning materials, transportation,and boardingfees are someof the non-tuitioncosts of sendingchildren to school. Indeed, parentaleducation expenditures can be quitelarge. In Malawi,for example,a studentat the secondarylevel paid 30 Malawiankwacha (K) in tuitionand K71in boardingcharges in 1982,or 38 percent of the total costs per student place(Psacharopoulos et aL 1986). The familiesalso incurredadditional expenses for uniformsand transport to school. In Korea, in the mid-sixties,about 70 percentof national education expenseswere paid for by studentsand parents. These contributionswere used for the constructionand operation of schools,and for expenseson books,school supplies, room and board, and transportation (McGinn et al. 1980). These miscellaneousexpenses accounted for 80 percent of total education expendituresby the family. For a varietyof reasons,these out-of-pocketexpenses may be differentfor boysand girls. For example, parents'greater reluctance to senddaughters to schoolwithout proper attire raisescost of schoolattendance of girls. Or parentalconcern for the physicaland mortalsafety of youngdaughters in some culturesdeters them from allowinggirls to attend distantschools requiring long traveldaily, thus necessitatingdifferent boardingand lodgingarrangements costs for them.

20Some of these issuesare discussedlater in this volume. 21 In an economicmodel, this can be shownby representing household utility as a functionof twodifferent commodities--thehuman capital stock of sonsand daughters(Rosenzweig and Evenson1977, Rosenzweig and Schultz1980). Appendix1.B to this chapterpresents a modelof demandfor educationthat does not rely on this assumption.Rather, investmentsin educationare drivenby market forces. 24 Overview

In addition, parents may not be able to afford the opportunity costs of educating children, which vary by sex and from country to country. Although in some countries (e.g. Botswana,Cote d'Ivoire, and some areas of the Philippines),boys perform a larger share of familylabor, herding livestockor plowingthe field, with few exceptions,girls do more home and market place work than boys. They cook, clean house, fetch water, and help their mothers care for younger children, especially those who are ill. In Nepal and Java, for example, most young girls spend at least one-third more hours per day working at home and market than boys of the same age, t.id in some age groups as much as 85 percent more hours (table 1.7). In Malaysian households, girls aged 5 to 6 who do home or market chores work as much as three-fourths more hours per week than boys of that age. In Chinese and Indian households in Malaysia, girls aged 7 to 9 work as much as 120-150percent more hours than boys. Clearly,girls who work more than their brothers will be less Likelyto attend school, or they willbe more overworkedif they do (causing them to perform less well). These examples provide the evidence that gender inequality exists event at early ages.

Table 17 InternationalDifferences in the Market and HouseholdTime Spent by Girls in ActivitiesRelative to Boys' Time

Author(s) and Definitionof Work Location of Study and Time Dimension Used Age Gruup

Nag and others (1980) Home and Market Production 6 to 8 9 to 11 12 to 14 Indonesia Hours per day Javanese Village 0.97 1.74 1.85 Nepalese Village 1.32 1.29 1.36 de Tray (1983) Home and Market Production S to 6 7 to 9 10 to 14 Peninsular Malaysia Annual Participation Rates Malays 1.82 1.17 1.10 Chinese 2.50 1.78 1.31 Indians 2.18 1.79 1.35 Average Weekly Hours if Working Malays 1.21 1.74 1.75 Chinese 1.00 2.19 1.35 Indians 1.76 2.49 1.80 King (1982) Participation Rates 7 to 1O 11 to 14 Philippines In SchoolYouths 0.60 0.29 Out of School Youths - 0.37 Cabafiero (1978) Average Annual Hours of Home 6 to 8 9 to 11 12 to 14 P'.ilippines and Market Production 0.93 1A9 1.02 Newman (1988) Participation Rates 7 to 14 C6te d'Ivoire 0.94 Mueller (1984) Proportion of Total Time in 7 to 9 10 to 14 Rural Botswana Home and Market Production 0.93 0.92 King and Bellew (1989) Participation Rates S to 7 8 to 10 11 to 13 Peru In School Youths 0.56 0.89 0.88 Out of School Youths 1.07 1.03 1.14

Besideslost work, parents may feel that girls are foregoingimportant childcare,household and craft training at home if they go to school. The relative importance of these foregone training opportunities will differ across countries depending, in particular, on the expected adult occupation. For example, if most women enter the informal labor market by continuing in a crafts tradition (or in agriculture), the skillsfor which Overview 25 are imparted by their mothers,then the cost of attendingformal schoolingmust includenot only the opportunitycost of current time, but also the lost alternativetraining. 2 2 Lastly,in additionto the financialcosts and opportunitycosts of schooling,educating girls may exactnon- pecuniaryor "psychic"costs as well. In certainsettings, socio-cultural factors (such as normsproscribing societal,economic, and familialroles of women)and religionstrongly influence the behaviorof parentsby imposinga heavycost on nonconformistbehavior. These may bear significantlyon schoolingdecisions. In countriesin whichfemales are usuallysecluded, for example,girls may attend onlyschools that do not admitboys or onlythose that employfemale teachers. These concerns are usuallystronger when girls reach the onset of puberty. A related concernis that parentsmay considereducation itself as a negativefactor becausethe suitability of more highlyeducated women to be goodwives is held in doubt. In manytraditional societies, education beyondthe acquisitionof literacyis contrary to the socialpressure for womento becomewives and mothers, and threatenswomen's possibilities for marriage. With economicdevelopment and correspondinglyexpanding work opportunitiesfor women,however, tension builds up between traditionalsocial norms and the family'sdesire to benefit from changing conditions.The questionsthat arise are, how much,when, and whichfamilies or individualswill respond to these environmentalshifts? Economictheory does not deal formallywith the impactof socio-cultural forceson individualbehavior, but it predictsbehavioral adjustments to unexpectedchanges in incomesand pricesthat resultfrom economicgrowth. For example,it predictsthat a rise in femalewages, which in turn increases returns to their education,would tend to increase parental investmentin their daughters' education. The magnitudeof the responseand the speedwith whichit is made dependson acquisitionof new informationand the price and incomeelasticities of their demand for education. Clearlyparental preferencesare both shapedby and shapethe economicenvironment.

Multiple Benefilt at Work and at Home But evenwhere the costsof educatinggirls and boys are identical,parents may still keep girls home to work ard send their sons to school. Althougheducating a girl benefitsnot onlyher parents,but the girl herself, her ownfuture family,as wellas societyat large,it maybe the expectedreturns to parents fromeducating sons relativeto those fromeducating daughters that determineswhich children parents will send to school. Educationenhances women's economic productivity in the farm and non-farmsectors. In a studyof the productivityof men and womenfarmers in Sub-SaharanAfrica, the gainin productivityfrom education was found to be higher for womenthan men. Studieson the determinantsof wageearnings have found the marginaleffect of educationto be about as largefor womenas for men oncelabor forceparticipation, work experience,and sector of employmenthave been taken into account. But discriminatoryemployment practicesagainst women have limitedtheir work opportunitiesand have reduced the earningsthey can expectto gain from education. Entry barriersagainst women, explicit or implicit,in certain occupations serveas obstaclesto education. Examplesare restrictionsagainst the hiring of marriedwomen in wage- payingjobs in the manufacturingor servicesectors. Someof the barriersbegin even at the primaryschool levelwith teachers and textbooksprojecting attitudes that discourageperformance of girls, or promoting stereotypesof girlsnot beingas capableas boysin learningtechnical subjects or mathematics.Some begin at the postprimaryeducation level with gender-specific admissions policies in certain areas of study.

22In some societies,still another opportunitycost of schoolingis the earlier use that the familycan make of the bridepricefor daughters. Delayingmarriage due to schoolingpostpones receiving the bride wealth and mayeven reduce its amountif there is greatervalue placed on younger,than on more educated,brides. 26 Overview Unless daughterstransfer part of their future incometo their parents,then parentswho mustbear some of the costsof educationmay not havesufficient incentives to do so.23 For example,the earliergirls marry and move into their husbands'families, the lessparents enjoythe benefitsof their daughters'education. In Bangladesh,75 percentof womenliving in rural areas who haveever been marriedwere married by the age of 17. In India,75 percentof this group weremarried by age 19. Someevidence suggests that when girls do not marry so early,but spend sometime workingin the labor force,parents are more willingto educatetheir daughters(Schultz, this volume; Acharya and Bennett1981). In HongKong, although custom dictatesthat sonstake responsibilityfor their parents,girls who marryat later ages and help their parents in the interimappear to attain higherlevels of schoolingthan others. Whilemore educatedwomen are generallybetter paid and more likelyto find employmentin the paid sector than less educatedwomen, married women are more likelyto withdrawfrom the labor market as their schoolingincreases from the primaryto the secondarylevel. Pregnancy, childbirth and childcare duties remove women from the work force for substantialperiods, or permanently,denying them paid work. Moreover,this labor supplypattern feeds back intoemployers' wage-setting decisions, causing them to place a lowervalue on womenworkers than men workers. But the withdrawalfrom the labor forceby married womenis also due partlyto the fact that educationincreases women's productivity in nonmarketactivities too, and that unlessbetter-paying jobs outsidethe home are availableto those withsecondary education, stayingat home is often a superioroption. In the home, women'seducation has a greater effecton familywelfare than men's education. Studiesin demography,economics, medicine, and anthropologyhave founda stronglink betweenmother's schooling and decreasesin the incidenceof mortalityamong her children--arelationship that appearsto be stronger in low-incomecountries. These results show that an addedyear of educationfor a mother is associatedwith a reductionof between5 and 10 percent;n childmortality. Greater schoolingof the motherappears to lead to betterhygiene, improved nutrition practices, and greater effectivenessin caringfor the family'shealth. Does educationsimply encourage the use of more health inputsand enablethe more effectiveuse of theseinputs, or does educationactually provide a mother with the capacityto cope with health risks and better manageher child'senvironment? In West Africa,the educationof the mother enabledher to exploitlocal public health more effectively(Caldwell 1979). In a studyon the Philippines,mother's schooling was foundto havea larger protectiveeffect on childhealth in communitieswithout piped water and water-sealedtoilets and in communitiesthat were further from outpatienthealth care facilitiesthan in areas that were better off (Barrera 1990). In general,schooling seemsto equipmothers with knowledgeneeded to be more effectivein their roles at home. Mother'sschooling improves also her ownhealth status. One reasonfor this is that more schoolingseems to accord her greater controlover the frequencyand spacingof childbearing,and to influenceher use of health servicesduring pregnancy and birth. Frequentpregnancies take their toll on the mother,resulting in what is termed "maiernaldepletion syndrome,* particularly in poorer areas where the higher dietary requirementsof pregnantor lactatingwomen often remain unfulfilled. Among the better-educatedwomen in LatinAmerica and Asia,through higher prevalence of contraception,fertlity rates havedeclined and are approachingdesired levels. Moreover,it may be of interest to populationplanners to know that the educationof the wifehas a strongernegative effect on fertility(by almostthree times)than does husband's education. Mother'seducation improves the educationalattainment of her children,particularly, that of daughters. In manycases, it has been found to have a larger impacton children'sschooling than father's education (eventhough father's education implies also an incomeeffect), and to exert a greatereffect on the schooling of daughtersthan sons. A studyof femalestudents enrolled in the publiceducation system in Cairo found

23 One attenuating factor may be that parents have preferences about the number of surviving grandchildren,and that more highlyeducated daughters produce a greaternumber of survivingoffspring than morehighly educated sons. Thiscould occur in the absenceof assortativemating, and in the presence of assortativemating if there is greatercertainty regarding the qualityof the daughter'schildren than those of the son's wife. Overview 27 that differencesin measuresof their self-confidencewere associated with the educationof the mother. The more formalschooling a mother had, the more she gave praiseand confidenceto her daughters,and the more differenther standardsand expectationswere for her daughtersfrom those of less educatedmothers. Lastly,educating women supports or enablesthe exerciseof their rightsand obligations.The rightto avail of creditor ownland is diminishedby not being able to read or understandcontracts, or performsimple arithmetic.The right to vote is meaninglessunless women can informthemselves of the issuesof the day and protectthemselves through du( processof law. Violenceagainst women in the home or on the streets have been associatednot just withpoverty but also with illiteracy,which prevents women from asserting their rights. These benefitsare just as importantto the livesof womenas are those discussedabove. Table 1.8 summarizesthe costs and benefitsassociated with women'seducation, whether market or nonmarket, as weli as the agents who gain and those who pay. As this table emphasizes,educating women yieldsdurable future benefitsto individualwomen, to their families,and to societyat large. Becausethe publicbenefits are large and potentiallyunmeasurable, but the burdenof costs oftenborne by parenis%the educationthat girlsare givenis likelyto be belowthe sociallyoptimal level. It is hardlysurprising then that in poorer countries,women remain undereducated.

Table 1.8 A Summaryof Costs and Benefits Associatedwith Educating Girls

Market Non-Market

Cuffent Costs

Parents Uniforms,supplies Opportunitycost of Tuition time in school TravelCosts Psychiccosts Student P/ Psychiccosts SWi Publicexpenditures for teachers Foregoneoutput of childrenwhen salaries,supplies, buildings in school

FutuneBenefits

Parents Higherfamily income Student Higherearnings Lowerlife expectancy Greater occupationalmobility Lowerinfant mortality Greater fertlity control Sgui HigherGNP and GNP per capita Reducedpopulation growth Increasedlabor productivity Healthierpopulation (lower infant mortality, Higher GNP growth longer life span) Higher taxeson eanings Better functioningpolitical processes

/ The assumptionthat the studentdoes not bear anycurrent "market" costs is realisticonly at the primary or secondarylevel of education.At higherlevels, the studentherself is likelyto bear the costsof tuition and travel,and to incur the opportunitycost of time spentin school. 28 Overview

The Chapters Ahead Thischapter began with an overviewof the state of women'seducation in developingcountries. It revealed gender disparitiesin each of the measuresof educationalattainment we considered. Usingextensive data for 152 countriesover the time period 1960through 1985,our empiricalresearch indicated that the level of femaleeducation and the gendergap in educationare each importantdeterminants of aggregatefamily well-beingand economicgrowth. To underpinthe discussionin the chaptersthat follow,we presenteda frameworkfor educationaldecisions within the family.This frameworksuggests that the degree to which costs and benefitsare publicaffects the levelof women'seducation. Chapter2 focuseson the multiplebenefits that arise fromincreasing the educationof womenand the issues regardingthe measurementof these benefits. The discussionemphasizes the need to understand the sourcesof genderdifferentials in schoolingand how these are affectedby the structureof aggregatedemand for labor, as well as by the costs of supplyingeducation to girls. The chapter concludesthat greater investmentsin the primaryand secondaryeducation of womenare warrantedon economicgrounds. Chapters3 through7 describethe regionalsettings within which schooling decisions are made as well as the currentstatus of women'seducation in thoseregions. A selectivesummary of their findingsis givenin the nextsection. Fmally,chapter 8 concludesby describingand analyzingvarious policy interventions that havebeen attemptedregarding women's educational attainment, and reviewing,in particular,examples of pastand ongoingeducation programs and interventionsto demonstratethe typesof schemesthat havebeen effective.

A SelectiveSummary of Findings:Determinants of the Gender Gap The regionalchapters illustrate the vastinternational differences in both the levelsof femaleeducation and the nature of the gendergap in the developingworld. These chaptersdepict the rich diversityin culture, institutions,and the levelof economicdevelopment which jointly give rise to distinctbarriers to education for women.The chaptersreview what published documents--academic studies, government reports, project assessments--haveto say about the factorsthat influencethe educationof girlsand women. Each regional chaptercorresponds to a group of countries,namely, Africa south of the Sahara,the MiddleEase and North Africa,Latin America and the Caribbean,South Asia, and East and SoutheastAsia. Each chapterbegins witha summaryof the statusof femaleeducation in the regionand then surveysthe availableliterature and discussesfindings. Each is limitedby the amountof availabledata and researchin the region. Interest in researchon the educationof womenvaries widely by region. For example,while there existsa strongand growingbody of internationalresearch on gender issuesin educationin SouthAsia, onlylimited empirical research has been carriedout for the MiddleEast and North Africawhere female enrollmentsrates also lag behindboys. And in LatinAmerica where primary school enrollment rates of girlshave generally been at parity withthose of boys,most researchon educationdo not focuson gender issues. Despite lear regional differencesin cultres and institutions,chapters 3 through 7 indicate striking commonalitiesin the factors that influencethe educationalattainment of women. Beloware some that standout. They can be groupedmore readilyby contextthan typeof effect--thefamily and the home, the school,and the economyand society at large. Thereader is directedto the appropriatechapters for a fuller discussionof findings.

Famly and te Home Parents'education bears an importantinfluence on the genderdifferences in education. In Ghana,female studentsat the secondarylevel are more likelyto comefrom familieswith more educationas comparedto male studentsat the same level. In Egypt,holding income constant, parental educationhad the most influenceon educationalaspirations for childrenin rural and urban areas. However,in some countries,it Overview 29

is the father's education that appears to make a difference; in others, it is the mother's education. In Cote d'Ivoire, a girl with a university-educatedfather is 35 times more likely to enter an academic secondary school than the daughter of a man with no education. In Peru, the father's education has twice as large an impact on sons' schoolingas does the mother's education, while the effect of each parent's education on a daughter's schooling is equivalently strong and positive. In Thailand, it is father's (not mother's) educational aspirations for daughters that is important in determining daughter's schooling,but not so for sons. In Pakistan, the large majority of illiterate women want only religious education for their daughters, while the bulk of those with some primary education want at least higher secondary education; in lowerincome households, more educated fathers tend to enroll their daughters more than less educated fathers.

But in Sub-SaharanAfrica, there is vast evidencethat women bear a large part of the burden of educating their children, especially in areas where polygamous marriage is common or where male migration is widespread--that is, when women become de facto heads of houseiiolds. And when women are the heads, girls are as likely to be in school and to stay in school as long as boys, as in Botswana. Similarly,in Saudi Arabia, the higher the educational level of mothers, the greater their influence on their daughter's academic plans. What does parental education represent? The literature interprets the effect of this variable in different ways. First, parent's education may represent the value that parents attach to formal education. The expected direction of the relationship is generally that more educated parents value more highly formal education for their daughters as much as for their sons. Second,it measures more generally the degrt. to which parents are open to influences outside tradition. Hence, even in a relatively closed society which restricts the activitiesof girls and women,the more educated parents are less likely to see formal education as a threat to their way of life. Third, parents' education is a limited measure of family income or wealth when more direct measures are not available. When estimates of family income are available, as in Malaysia, income was found to have three times as great an influence on the probability of enrollment of girls aged 12 to 18 as on that of boys. Another home factor that affects girls' education is the demandsfor their time in alternative uses. Although in countries such as Brazil, Chile and Nicaragua, boys perform a larger share of family labor, in many countries, girls do more home and market place work than boys. In Ethiopia, household duties are a primary reason for keeping boys out of school, while the schoolingof girls might be seen as famine-fighting strategy because being literate allowsthem to marry better or to fmd work. However,in Jordan and Nepal, girls drop out to help with domestic tasks, especially the care of siblings. And in India, a rise in female wages has been found to reduce girls' school attendance.

Schoolfand Teadie The school euvironment exerts its own influence on female education. Compulsory education legislation, open admissionspolicies, and "free"education have not guaranteed equal access of rights to education. For many other reasons, schools can be regarded as "closed"or inaccessible to girls and women. In Egypt, Tunisia, and Morocco, parents are reluctant to send their daughters to distant schools because of the fear of moral or physical peril to girls. Even in the relatively more open societies of Malaysia and the Philippines, distance to school is also a greater deterrent of girls' enrollment than of boys'. And school facilities themselves can be hostile to girls. In Bangladesh,parents have withdrawngirls but not boys from schools without latrines. In Kenya and Ghana, girls are over-represented in secondary level institutions of low quality, us'zallyimplying their lack of access to science and math subjects. Kuwait and Saudi Arabia (in contrast to Morocco and Egypt) illustrate how wealth has increased school facilitiesfor girls so that the demands of culture are met. A study in Brazil shows that teachers believe girls to be less capable in math and consequentlyfail to use teaching techniques that might improve girls' achievement in math. In Kenya and Ghana, girls are channeled into domestic science, handicrafts and biology, and boys into chemistry,math and vocational subjects. Various studies have found that single-sexschools may be more effective for girls' learning. In Thailand, these schools make a difference; even after controllingfor such factors as socioeconomichome 30 Overview backgroundand schoolresources, girls achieved more in single-sexschools than in coeducationalschools, whileboys did better in the latter.

Employmen wad Marrge Althoughthe non-marketbenefits from education are manifold,the regionalreviews suggest that the returns to educationin the labor market are also importantin the case of girls. In Nicaragua,higher predicted earningsfor the womenin the householdare significantlyassociated with greater educationalattainment by the children.In Mexico,visible returns to schooling,measured as the prsportionof white-collarworkers withhigh incomes,were more importantfor the decisionto send girls than boysto school. But evenin East Asia and LatinAmerica where more women are enteringthe formallabor market,women workersare stillconcentrated in a fewjobs which are generallycharacterized as of lowskill, low wagesand low mobility.In Malaysia,boys expect their salariesto be higherthan do girls; and girlsbelieve that the range of jobs for them is restricted. These expectations,in turn, affecteducational aspirations. The sharp distinctionbetween male and female socializationstill persists in many countries. In Arab countries,such as Egypt and Morocco,the socializationof girls emphasizesthe acceptanceof the predominantsex-role where marriageand family,not employmentin the labor market,are the ultimate goalsof women. For example,in Ethiopia,20 percentof primaryschool students surveyed in a studywere alreadyeither promised, married, or divorced.In thesecontexts, girls will ' ^ educatedif schoolingis viewed as a positivefactor in marriage. In Zaire, a higherbride price for moreeducated girls induce some parents to send girlsto school.However, in NorthernNigeria, there appearsto be a socialconsensus that education in Westernschools is bad for girlsand for societyas a whole.This attitudeis a barrier to femaleeducation evenwhere the governmentor donor agenciesare able to financethe constructionof schools.

Notes on Estimation Past studies have examinedthe extent of the gender problemthrough comparingaverage schooling or achievementlevels of both sexes,using either two-waytables or a dummyvariable for sex in the case of multipleregression analysis. These comparisons,however, do not elucidatethe more difficultquestion: whataccounts for sex differencesin education?In order to do so, multipleregression models will need to examinewhether family background, school-related variables, or individualcharacteristics besides sex have differenteffects on whethergirls enroll in school,how long they stayin school,and howwell they perform. This requires estimatingseparate regression parameters for girlsand boys. Conclusionsregarding causality from aggregate two-way relationships can alsobe misleading.For example, the marginalreturns to educationin terms of market wageshave been found to be at least as high for womenas for men (chapter2). That the averageeducation level of womenis lowerthan that for men does not necessarilymean that women'sschooling does not respondto market incentives.Rather, the rate of return to femaleeducationi may be highersimply because the averagelevel of educationis lower. And even when the underlying research has been executed with great care, methodologicaland measurementproblems persist. Studiesthat attempt to quantifythe strength of the differentfactors affecting,for example,the decisionto continuein or drop out from a givenlevel of schooling,often fail to correctfor the fact that the membersof the samplestudied (girls of a particularage groupin schoolin this instance)are not a randomsample of the populationat large (all girlsin that age group). Certainfamily characteristicsare highlyassociated with the probabilitythat girlswill ever enroll in school. To assessthe Overview 31 relativestrength of schoolor other factorsthat affectcontinuation and attainment,it is necessaryto control for the likelihoodthat girlsalready in schoolare a non-randomsample from the relevantage cohort.24 The problem is perhaps most obviouswhen dealingwith female school populationsthat are a small percentageof the relevantage groupson a nationalor regionalbasis, but it alsopertains when dealingwith apparentlyhigh enrollment and continuancerates wherethere is a divergencein the typesof schoolsmales and femalesattend. For example,when both coeducationaland single-sexinstitutions exist, it is very commonfor single-sexschools to be more selective,both academicallyand socially.A findingthat girls in single-sexschools experience the highestachievement levels is of limitedvalue unless statistical controls demonstratethat this effectremains after accountingfor the backgroundsof thosestudents who do attend single-sexschools. Therefore, it is necessaryfirst to correctfor the typesof girlswho eitherattend school or attend a certaintype or levelof school,before one can beginto make supportedstatements about those factorswhich can increaseor decreaseenrollment, persistence, attainment or scholasticperformance. The use of correlations,while illustrative and descriptive, does not provideadequate informa%tion on the net effectthat any particularvariable might have on the dependentvariable. Often,studies make no attempt to control for confoundingfactors when tryingto identifythose variableswhich influence various aspects of women'seducational participation. For example,the attributionin Akande (1987)of differencesin achievementand aspirationto rural or urban residenceis problematicwhen consideringthat parental income,parental education, school quality and labormarket opportunities (to name onlya few factors)are also highlycorrelated with livingin urban/rural areas. Cooksey(1980) made a similar point when he arguedthat comparisons of schoolpopulations from high and lowenrollment areas were generally not valid becausechildren in lowenrollment areas were usuallyselected based on highability, high socioeconomic status,or both. Specifyingincomplete models of educationchoice, often due to measurementproblems, also leads to biased estimates of the effects of factors. Geneticallydetermined ability affects learning and educational attainment,and thus income;but, due to limiteddata on cognitiveability, the effect of this factor on educationand incomehas been neglectedin moststudies of the determinantsof schoolingand its rate of retun. Grilichesand Mason(1972), in a studyfor the UnitedStates, estimated that failureto controlfor the effectof abilityoverstates the estimatedrate of return to educationby between7 and 15 percent. In a studyfor Tanzaniaand Kenya,Boissiere and others (1985)found that controllingfor abilitylowered the estimatedrate of return by about 60 percent. Measurementproblems regarding costs and benefitsare pervasive,and withoutappropriate measures of these factors,our understandingof the sourcesof gender differentialsin schoolingis severelyhampered. The gender differencein opportunitycosts to the family,for example,has been measuredby wagesthat boysand girls could commandin the labor market. Yet many of the work activitiesthat childrenengage in are outsidethe wagesector. Also, usingtime spentin alternativeactivities as a measureof the valueof time is flawedsince it is an outcomeitself, being determinedtoo by the value of time. The measurement of benefits as perceivedby parents of the individualis even more elusive. Whereas the market and nonmarketbenefits are plainto see in the aggregate,choosing appropriate measures that reflectindividual or familydifferences in benefitspresents problems. Is the wagethat an individualreceives in the labor market the correct measure of economicreturns expectedat the time schoolingdecisions were made? When explaininga student'slikelihood to continuein school,whose wage is relevant? Finally,sampling methods used in studiesshould be made explicit.Identifying the numberand gender of pupilsis not helpfuleven to the informedreader when informationon the way the pupilswere selectedor their backgroundcharacteristics are not part of the report. Unfortunately,many studies do not describe their samplingmethods sufficiently welL

24 There is also the more generalestimation problem related to analyzingschooling outcomes that are still incomplete. When studentsare still enrolledin school,using educational levels attained thus far as a measureof outcomesis essentiallyflawed. The finallevels are, in reality,unobserved until schoolinghas stopped. 32 Overview These estimationissues imply that readers of the literatureshould interpret findingswith caution. It is reassuring,nevertheless, that the nun'erous studies reviewed in the followingchapters demonstrate remarkableconsistency across countries in the directionof the effectsthat of perceivedcosts and benefits have on educatingwomen. Theysuggest to us the broad directionsthat policyshould take in improving female education. Designingspecific interventions, however, would require estimatingmore precise magnitudesof these effectsand, therefore,greater attentionto methodology. Overview 33 References Acharya, M. and L. Bennett. The Rural Women of Nepal: An AggregateAnalysis and Summary of Eight VillageStudies. Center of Economic DevelopmentAdministration. Khatmandu: Tribhuvan University, 1981

Akande, B.E. "Rural-Urban Comparison of Female Educational Aspirations in Southwestern Nigeria," ComparativeEducation 23:75-83,1987. Barrera, A. "The Role of Maternal Schooling and Its Interaction with Public Health Programs in Child Health Production,"Journal of DevelopmentEconomics 32:69-91,1990 Bellew, R. and E. King. "Promoting Girls' and Women's Education: Lessons from the Past," PRE Working Paper No. 715. Washington, D.C.: World Bank, 1991.

Behrman, J. and A. Deolalikar. "Are There Differential Returns to Schoolingby Gender?" Development Studies Project II Research Paper. Jakarta, Indonesia. 1988 Behrman, J. and B. Wolfe. "The SocioeconomicImpact of Schooling in a Developing Country,"Review of Economics and Statistics, 66:2, May 1984, 296-303. Bellew, R. and E. King. "PromotingGirls' and Women's Education: Lessons from the Past," PRE Working Paper No. 715. Washington, D.C.: World Bank, 1991. Biazen,A. and B. Junge. "Problemsin Primary School Participationand Performance in Bahis Dar Awraja," United Nations Children's Fund. Addis Ababa, Ethiopia, June 1988. Birdsall, N. "PublicInputs and Child Schoolingin Brazil,"Journal of DevelopmentEconomics, 18:67-86,1985 Boissiere, M., J. Knight and R. Sabot. 'Earnings, Schooling, Ability and Cognitive Skills,"American Economic Review, 75(5):1016-30,1085. Butz, W., H.P. Habicht and J. DaVanzo. "EnvironmentalFactors in the Relationship Between Breastfeeding and Infant Mortality,"American Joumal of Epidemioloqg,119:516-25, 1982. Cabanero, TA. "The Shadow Price of Children in Laguna Households."The PhilippineEconomic Joumal 17(1&2):62-87,1978.

Caldwell,J.C. "Educationas a Factor in Mortality Decline,"Population Studies 33:395-413,1979. Chernichovsky,D. and OA. Meesook."School Enrollments in Indonesia,"World Bank Staff Working Paper No. 746. Washington, D.C.: The World Bank, 1985 Cochrane S. and D. Jamison. "Educational Attainment and Achievement in Rural Thailand," in AA. Summers (ed.), ProductivityAssessment in Education: New Directionsfor Testingand Measurement. San Francisco: Jossey-Bass, 1983 Cochrane, M., and T. Osheba. "The Educational Participationof Egyptian Children."EDT DiscussionPaper No. 45. Washington, D.C.: World Bank, 1986.

Cooksey, B. "Social Class and Academic Performance: A Cameroon Case Study."Comparative Education Review, 25:402-17,1981 DeTray, D. "Children'sEconomic Contributionsin Peninsular Malaysia,"A Rand Note, N-1839-AID. Santa Monica, CA: The Rand Corporation, 1983. 34 Overview

Greenhaigh, S. "SexualStratification: The Other Side of Growth with Equity in East Asia,"Population and DevelopmentReview, II(2): 265-314,1985. Griliches, Z. and W. M. Mason. "Education, Income, and Ability,"Joumal of Political Economy, 80(3), 11:574-3103,1972. Gujarati D. Basic Econometrics,Second Edidon. New York: McGraw Hill, 1988. Haveman, R. and B. Wolfe. "Schoolingand Economic Well-Being: The Role of Non-Market Effects," Joumalof Human Resources, 19(3):377-406,1984. Heyneman, S. and WA. Loxley."The Effect of Primary School Quality on Academic AchievementAcross Twenty-NineHigh and Low Income Countries,"AmericanJoumal of Sociology,88(6): 1162-94,1983. Kelly, G. and C. Elliott eds., Wonren'sEducation in the Third Wortd. ComparativePerspectives. Albany. State Universityof New York Press, 1982. King, E. "Investments in Schooling: An Analysis of Demand in Low-Income Households." Ph.D. dissertation, Yale University,New Haven, Connecticut. 1982 King, E. "The Effect of FamilySize on FamilyWelfare: What Do We Know?"in D.G. Johnson and R. Lee (eds.), Population Growth and Economic Development (Issues and Evidence). Madison, WI: The University of Wisconsin Press, 1987 King, E. and R. Beliew. "Gains in the Education of Peruvian Women in 1940-1980,"PRE Working Paper No. 472. Washington, D.C.: World Bank, 1989. King, E. and R. E. Evenson. "Time Allocation and Home Production in Philippine Rural Horiseholds,"in M. Buvine, M. Lycette and W.P. McGreevey (eds.), Women and Poverty in the Third World. Baltimore, MD: John Hopkins University Press, 1983 King, E. and L. LiDard."Determinants of SchoolingAttainment and Enrollment Rates in the Philippines," TheRAND Corporation,N-1962-AID, April 1983. ------. "Education Policy and Schooling Attainment in Malaysia and the Philippines," Ecvnomics of Education Review, 6:167-81,1987. Lau, LJ., D. Jamison, and F. Louat. "Educationand Productivityin DevelopingCountries, An Aggregate ProductionFunction Approach,"World DevelopmentReport, WorkingPaper, WPS 612.Washington, D.C.: World Bank, March 1991. Lockheed, M. and A. Verspoor, et.al. ImprovingPrimary Education in DevelopingCountries: A Review of Policy Options. Bank, 1990. McGinn N. and others. Education and Development in Korea. (East Asian Monographs). Cambridge: Harvard University Press, 1980 Moock, P.R. and J. Leslie. "ChildhoodMalnutrition and Schoolingin the Terai Region of Nepal,"Journal of DevelopmentEconomics 120:33-52,1986. Mueller, E. "The Value and Allocation of Time in Rural Botswana,"Joumal of DevelopmentEconomics 15:329-360,1984. Nag, M., B. White, and R. C. Peet. "An AnthropologicalApproach to the Study of the Economic Value of Children in Java and Nepat" in RuralHousehold Studies in Asia, edited by H. Binswanger, R. Evenson, C. Florencio, and B. White., Singapore University Press, 1980. Overview 35

Newman,J. "Labor Market Activityin Cote D'Ivoire and Peru," LSMS WorkingPaper No. 36. Washington, D.C.: World Bank, 1988.

Psacharopoulos,G., J.P. Tan and E. Jimenez. FinancingEducation in DevelopingCountries: An Exploration of Policy Options. Washington D.C.: World Bank, 1986. Rosenzweig,M. and R. Evenson. "Fertility,Schooling, and the Economic Contribution of Children in Rural India: An Econometric Analysis,"Econometrica, 45:1065-79, 1977.

Rosenzweig, M. and T.P. Schultz. "Market Opportunities, Genetic Endowments, and the Intrafamily Distribution of Resources: Child Survivalin Rural India," Ametican Economic Review, 72:4, 803- 815, 1982.

Schultz, T.W., "The Value of the Abilityto Deal with Disequilibria,"JournalofEconomic Literature,13:872- 876, 1975.

Sen, A. "More Than 100 Million Women are Missing,"New York Review of Books, December 20, 1990. "Gender-...... and Cooperative Conflicts,"in Irene Tinker,ed., PersistentInequalities: Women and World Development. Oxford UniversityPress, 1990. ------.Resources, Values and Development, chapters 15 and 16. Cambridge: Harvard UniversityPress, 1984.

Smock, A. C., Women's Education in Developing Countries: Opporfunitiesand Outcomes. New York: Praeger, 1981.

United Nations Statistical Office, Women's Indicatorsand Statistics (WISTAT) - User' Guide to the United NationsMicrocomputerData Base FinalDraftAnnexes and OrderForme. New York: United Nations, 1988.

UNESCO. "Education of Girls in Asia and the Pacific: Report of the Regional Review Meeting on the Situation of Education of Girls for Universalization of Primary Education." Bangkok: Regional Office for Education in Asia and the Pacific, 1986.

World Bank. WorldDevelopment Report Washington, D.C.: Oxford UniversityPress, 1985through 1989. ------. FinancingHealth Servces in DevelopingCountries: An Agenda for Refonn. Washington, D.C.: World Bank, 1987. ------. World Tables. Washington, D.C.: World Bank, 1989. 36 Overview Appendix 1A: Description of the Data Sources

Data compiled for our analysesare derived primarily from United Nations statistical sources25and World Bank documents26. The final analytical database contains pooled time series data covering 155 countries and quinquennial years during the period 1960-1985.

Countries included in the data set span the entire globe, but we incorporate into our analysisonly those reporting consistent data. Table A.1 lists these countries by broad regionalcategories: Sub-SaharanAfrica (with data on 42 countries), South and Southeast Asia (17), the Middle East and North Africa (20, excluding Israel), Latin America and the Caribbean (31), East Asia (5), Other Asian (2), North America (excluding Mexico) (2), Europe (31), Oceania (4), and the USSR. However, data for each of the variables were not always available in each year for every country. Details on the availability of data for each of the regressions are presented in table A.2, which also displaysvariable definitions,sources, and means for each year.

2 The Research Triangle Institute kindly provided information on 17 variables from 80 countries. We expanded this initial data set with additional information taken from the United Nations 1988database entitled Women's Indicators and Statistics (WISTAT) and from the World Bank database containing UNESCO and WHO data.

26 World DevelopmentRepoit (Washington,DC: Oxford University Press for the World Bank, 1985-1989); FinancingHealth Senivcesin DevelopingCountries: An Agendafor Reforn. (Washington,DC: World Bank, 1987); World Tables. (Washington, DC: World Bank, 1989). Overview 37 Appendix T AbleA.1 Countries Included in the Analytical Data Base, By Region

Sub-Saharan Afilca South and Southeast Asia Ladn America and Caribbean Other Asian

Angola Afghanistan Antigua And Barbuda Japan Benin Bangladesh Argentina Israel Botswana Bhutan Bahamas North America Burkina Faso Burma Barbados Canada Burundi Democratic Kampuchea Belize United States Cameroon India Bolivia Europe Cape Verde Indonesia Brazil Bulgaria Central Afr. Rep. Iran Chile Czechoslovakia Chad Lao People's Dem. Rep. Colombia German Dem. Rep. Comoros Malaysia Costa Rica Hungary Congo Nepal Cuba Poland Cote D'Ivoire Pakistan Dominica Romania Equatorial Guinea Philippines Dominican Republic Denmark Ethiopia Singapore Ecuador Finland Gabon Sri Lanka El Salvador Iceland Gambia Thailand Guatemala Ireland Ghana Viet Nam Guyana Norway Guinea Middle East and North Afrca Haiti Sweden Guinea-Bissau Algeria Honduras United Kingdom Kenya Bahrain Jamaica Albania Lesotho Cyprus Martinique Greece iberia Democratic Yemen Mexico Italy Madagascar Egypt Nicaragua Malta Malawi Iraq Panama Portugal Mali Jordan Paraguay Spain Mauritania Kuwait Peru Yugoslavia Mauritius Lebanon Puerto Rico Austria Mozambique Libya Suriname Belgium Niger Morocco Trinidad & Tobago France Nigeria Oman Uruguay Germany, Fed. Rep. of, Rwanda Qatar Venezuela Luxembourg Senegal Saudi Arabia East Asia Netherlands Sierra Leone Sudai China Switzerland Somalia Syriat Arab Republic Hong Kong Oceania South Africa Tunisia Korea, Dem. People's Rep. Australia Swaziland Turkey Korea, Republic of New Zealand Tanzania United Arab Emirates Mongolia Papua New Guinea Togo Yemen Fiji Uganda USSR Zaire USSR Zambia Zimbabwe 38 Overview

AppendixTable A.2 Definitions,Sources, and Mean Values for Variables Icluded in InitercountryAnalysis

Mean Value and Number of Counaes Variable Deflnion Rtportng Data

1975 1980 1985 Ln GNP 1980 U.S. Dollars 22.7366 23.0449 22.8538 126 125 124 Female Life Expectancy Expectedyears at birth 60.8 62.5 64.2 148 147 147 Mate Life Expectancy Expected years at birth 56.8 58.4 60.0 148 147 144 Infant Mortality Rate Deaths per 1,000live births 825 73.1 66.6 138 138 144 Maternal Mortality Rate Deaths per 100,000live births 382.0 269.9 a/ 68 108 Total Fertility Rate Children per woman 5.0 4.7 4.4 126 125 125 Ln Capital Stock (if reported) Estimated by summing capital 22245 22.6431 225693 investmentsusing quinquennial 104 109 114 data; assumed total depre- ciation over 20 years. Ln Labor Force 153902 15.5268 15.6465 130 129 129 Female Primary Enrolhment Gross Enrollment Ratio, 784 84.7 87.0 lagged 10 years 126 125 116 Female Secondary Enrollment Gross Enrollment Ratio, 35A 43.5 484 lagged 5 years 127 125 111 Female/Male Enrollment Ratio Dummy variables based on 0.30 0.20 0.15 < 0.42 primary or secondary ratio, 142 136 137 whicheveris smaller 0.42-0.75 0.23 0.25 0.24 142 136 137 0.7S-0.95 0.24 0.28 0.30 142 136 137 La Total Population 15.4938 15.6076 15.7147 152 IS1 151 Population (1000's) per physician(if reported) 121784 7764.8 77903 58 91 108 % Population with safe water (if reported) 48.1 70.4 635 92 21 104

E/ Only nine countries reported maternal mortality for 1985,therefore we excludued 1985 from our analysis of this variable. Source: World Bank database (includes data from IMP, UNESCO, and WIHO). Overview 39

AppendixTable A.3 RegressionResults for Gross National Product Models

Indepdent Variable Coeffiden T-Swstbkc

Intercept 25152 3.66

Female/Male Enrollmcat Ratio

<0.42 -0.2621 -1.74

0.42- 0.75 402484 -1.9I

0.75 - 0.95 0.0162 021

Female Secondary Enrollment 0.0288 2A6

Log Capital Stock 0.7143 20.49

Capital'Enrollment 0.0020 -7.09

Log Labor Force 03585 8.43

Labor ForceEnrollment 0.0012 1.46

Adjusted R-square 0.9586

F-Statistic 239.37

Sample Size 289

Regression also includes dummy variablesif missingcapital stock mesure, for year, and for detaited relsons. 40 Overview

IppendixTable AA Effects of Female Education on Indicators of Social Well Being

Life Expectancy Mortality TotalFerilty Independent Variable Female Male Infant Maternal

Beta t Beta t Beta t Beta t Beta t Intercept 40.2941 7.58 35.0791 7.10 186.4631 6.16 -996.6836 -1.42 6.5313 4.84 Female Primaq Enrollment 0.0964 7.71 0.0915 7.87 -0.4130 -5.80 -3.3111 -1.98 -0.0064 -2.01 Female SecondaqyEnrollment 0.1224 5.52 0.1072 5.21 -5649 -4.48 3.0165 1.00 -0.0260 -4.62 Female/Male Enrollment Ratio < 0.42 4.8044 -3.82 -3.8525 -3.30 21.1575 2.95 99.8237 0.65 0.7176 2.25 0.42-0.75 -3.4071 -3.20 -2.7511 -2.80 11.3668 1.89 111.8381 0.88 0.7300 2.72

0.75-0.95 -0.6856 -1.09 -0.5168 -0.88 1.8770 0.52 82.5575 0.98 0.2608 1.63

Ln GNP (Predicted) 1.7773 4.01 1.7165 4.17 -10.7132 -4.24 -17.0170 -0.30 -0.2136 -1.90

Ln Total Populetien -L.i254 -3.77 -1.6475 -3.88 11.7548 4.51 104.1428 1.68 0.2223 1.91

Population (1000's) per Physician -0.0279 -1.25 -0.0341 -1.65 0.2431 1.91 0.0105 3.72 -0.013 -2.30

% Population with safe water 0.0136 1.11 0.0124 1.10 -0.0168 -0.24 -2.2997 -1.41 -0.0117 -0.38

REGIONS: Eastern Africa -2.2332 -0.96 -0.4342 -0.20 -25.2085 -1.90 180.9228 0.50 1.2033 2.03

Middle Africa -6.3130 -2.58 4.6285 -2.04 1.5855 0.11 347.0491 0.91 0.3535 0.57

Northern Africa 0.7706 0.34 3.5143 1.66 4.6560 -0.36 280.5897 0.83 0.8885 154

thern Africa -9.3626 -354 -7.7855 -3.17 12.7618 0.85 358.8022 0.89 1.8227 2.71

Westem Africa -4.3821 -1.92 -2.7853 -1.31 6.2085 0.48 380.1984 1.05 1.0987 1.90

Caribbean 0.7682 0.32 25630 1.13 -16.6017 -1.20 271.6048 0.59 0.3122 051

Central America 0.9S47 0.42 3.0891 1A7 -11.8763 -0.92 105.1453 0.29 1.2251 2.13

South America -2.2592 -1.05 -1.0138 4051 5.7105 OA7 174.9261 0.51 0.7252 1.33

Eastem Asia 2.3547 1.11 3.3019 1.67 -21.4497 -1.77 -136.1847 -0.39 -0.6231 -1.15

Southeastern Asia 05069 0.22 2.5290 1.20 -24.9603 -1.93 8.8818 0.03 -0.3510 -0.61

Southern Asia -0.5231 -0.22 4.8795 2.22 -20.8305 -1.55 445186 0.12 0.1105 0.18

Westem Asia 0.9442 0.44 2.9308 1.45 -0.0608 -0.01 173155 0.05 1.7244 3.18

Northem Europe -0.9263 -0.48 -0.1272 -0.07 -2.3011 -0.21 -85.8959 -0.26 0.2628 0.53

Southem Europe 1.7595 0.85 2.6271 1.37 -12.8484 -1.09 -55.9124 -0.14 -0.6410 -1.22

Western Europe -1.0469 0.53 -1.0219 -0.55 0.1473 0.01 -1.1001 0.00 -0.1616 -0.32

Australia and New Zealand -1.1982 -0.55 -0.5999 -0.30 0.3836 0.03 985990 0.27 0.2403 0.43

Melanesia -1.8003 -0.64 2.2609 0.86 -27.2732 -1.69 733.8979 1.79 0.2571 0.36

Adjustel R2 0.9378 0.9318 0.8948 0.5070 0.8463

F-Statistic 137.900 124.990 77.680 6.160 51.550

Sample Size 289 275 275 147 275 egression also includes dummy variables for year, and if missing data on population per physician and/or safe water measures. The equation for redicting GNP is displayed in Table A.3. Overview 41 Appendix 1.B: An Economic Model of Gender Differences In order to illustratethe potentialmechanisms by whichpolicy can affectgender differences in educational outcomes,one can specifya simpleintergenerational model that depicts heuristicallyhow parents in developingcountries may chooseto investin their children'seducation. For simplicity,we assumethat parentalpreferences embody no genderdifferences. Gender differences in educationthus arise from social, economic,and culturalenvironments within which rewards and restrictionsdiffer by sex.27 Supposethat parentalpreferences may be describedby a simpletwo-period model, in whichthe firstperiod representsthe time duringwhich children depend on their parents and duringthe secondperiod, parents rely upon their children: (1) U = U(C,, C2),

whereC 1 representsparents' consumption in the first periodand C2represents consumption in the second period. We assumethat first-periodparental income (VI) is exogenousand that boys (m) and girls(f) may spend their childhoodyears in two activities,namely, investing in schooling(Ej) and working(L). (2) T = Ei + Li, wherej = f,m. Childlabor maybe suppliedto a well-developedmarket or maybe employedin homeproducdon at wages (or shadowwages), wm and wf. During the secondperiod, we assume that parents rely solelyon their childrenfor economicsupport. Parentalincome (and hence,their consumption)depends both on the incomeof their children(Y 1) and the rate at whichthe childrenmay transfer incometo parents (a.).

(3) m= C2(anm Ym . af Yf). Each child'slevel of income,in turn, dependson his or her investmentin education(Ej).28 (4) Yj = y(Ej), j = f,m, wherethe functiony, whichcan be interpretedgenerally as an earningsfunction, can differ for males and females(see empiricalevidence on this in chapter2). Parentsthen face a cleartradeoff in choosingto educatetheir children.If childrenspend time in schooling today,the parent's future incomewill rise,but onlyat the expenseof current familyconsumption. By substitution,we can simplifythe problemto that of maximizingthe followingobjective function,

(1') U = U (C; C2 [amYr., (Em)' af Yf (Ef)]), subjectto

27 This modelmakes other simplifyingassumptions, among which are the following (a) that the number of yearsattended by a studentis a sufficientindicator of learning;this then disregardsalso grade repetition; and (b) that the only Ists to schoolingare time costs--anassumption that is not appropriatewhen the familyincurs expendituresfor learningmaterials and transportation,and for enrollmentin secondary educationwhich often requires more privatespending. 28 This model can be extendedto allowfor the effectsof assortativemating--education may increasean individual'sdesirability as a marriagepartner and thereforeincrease the child'sfuture income by raisingthe expectedincome of the spouse. 42 Overview

(5) Pi C1 = VI + w. (T-E.) + wf (r-Ef). The firstorder conditionsimply that parentswill choose the levelof their sonsand daughters education such that

aC 2 /(Ef) Wf a C2 /a (Em) WM

This conditionimplies that the benefits of educatingsons relative to daughters (in terms of income transferredto the parents) must equal the relativeopportunity cost of educatingthe'4 (the2is, income foregonewhile in school).Suppose then that the relativeopportunity cost of educatingdaughters (Wf / Wm) rises. As longas there existdiminishing marginal returns to further education,the educationof daughters would fail relativeto that of sons. Alternatively,if the relativereturn to educatingdaughters ([8 q2/8 (Ef)J/a C2/8 (E%)D)rises, relative parental investment in their daughters schooling will also increase. Without appealingto assumptionsabout tastes or preferencesof parents, this model can explainwhy daughters might systematicalyattain less schoolingthan sons. As modeDied,the parental return to educationdepends on both the mannerby whicheducation affects the child'sincome and the rate at which the child'sincome is transferredto the parents. Any factor that increasesthe parental return to educating sons more than that for daughterswill result in the parents providingrelatively less educationto their daughters. Thismay result if the labor marketrewards the educationof malesmore than that of females. For example,parents may anticipate that the earningsprospects for their daughtersare poorerthan for their sons (that is, 8Y / 8E > aY / a;r). Alternatively,custom and social norms may dictate that sons, rather than daugAer;,take responsibilityfor their parents(that is, a. > af). For example,girls may "marry out' of their own familyinto their husband'sfamily so that parentscan recoup little if any of the returns from their daughterseducation. Returns to Education 43 Chapter 2. Returns to Women's Education

T. Paul Schultz

Remarkablyfew detailedstudies of returnsto schoolingfor womenhave sought to clarifysocial or private investmentpriorities. As a consequence,no one has lookedfor an explanationof whythe highreturns to femaleeducation have not attractedmore publicand privateinvestment in women'sschooling, particularly in those countries where women receive much less educationthan men do. Analysis is needed to understandthe originsof the genderdifferentials in returnsto schooling.They may be relatedpartly to the st, uctureof aggregatedemand for labor and partlyto economicconstraints such as per capita incomeand the costs of deliveringschool services (Schultz 1987). The gender differencesin investmentbehavior may be perpetuatedby the structure of regulationsand incentivesin public (and private)education systems. Theymay also reflectfamDly decisionmaking and the preferencesof parentswho value greater productivity in a daughter less highlythan in a son or who are unable to appreciatefully the enhancednonmarket productivityof better educatedwomen. Understandingwhy this pattern occursin somesettings and how policyinterventions can changefamily behavior are challengesfor researchersand programdesigners.

Rates of Return to Schooling Estimatesof the rates of return to educationare calculatedin twoforms. The privaLhreturn is the internal rate of return that equalizesthe present discountedvalue of the privatecost of attendingschool with the present discountedvalue of the private after-taxgains the individualrecoups in subsequentproductive acti"ities. The scial return indudes,in additionto these privatecosts and gains,the cost of publicand privateschool subsidies and the gains in increasedtaxes more educatedworkers pay, as well as any net positivesocial externalities that educationgenerates that the individualdoes not capture. In practice,relatively few studies of privatereturns to educationeven deduct from labor earningsor wage rates what the more educatedworker is likelyto pay in increasedtaxes (income and indirect taxes), althoughdoing so is not conceptuallydifficult. This deductionwould be irrelevantif the numberof hours workedwere independentof educationand taxeswere proportionateto wages,assumptions that may be plausiblein analyzingthe returns to male education. The first assumptionconflicts with what is known about the behaviorof femalelabor supply,however. The distinctionbetween private and socialreturns to schoolingin mostempirical studies merely involves includingthe additionalsocial costs of publicexpenditures per pupilin the schoolsystem in the calculation of the internal socialrate of return. Thus, socialreturns to educationare predictablylower than private returns in proportionto the share of the total cOstsof schoolingthat the publicsector absorbs. Thisgap betweenprivate and socialreturns to schoolingis particularlylarge for highereducation in some relatively poor countries,where attendanceat institutionsof highereducation is rationed and studentswho gain admissiondo not necessarilypay tuitionand mayeven receive cost-of-living stipends (Psacharopoulos and Woodhall1985). In primaryand secondaryschools, the opportunityvalue of the time of those studentswho are removed from productivework in the familyis the primaryprivate cost of schooling,augmented in some casesby privateoutlays for books,school materials, uniforms, and transportation.On the publiccost side,salaries of teachersare the dominantcost in poorer countries;indeed, teachers' salariesin many such countries absorb 80 to 95 percentof publicexpenditures on education(see UNESCOStatstcal Yearbook,various years). The privatereturn to schoolingprovides the incentivefor individualsand familiesto investin education,to the extentthat theyview schooling as an economicinvestment in the futureproductivity of the humanagent. 44 Returnsto Education The consumptionvalue of educationprovides another motivationfor privateexpenditures on schooling. This motive is not commonlyobserved, however, and therefore is not analyzed a differentialforce affectingthe leveland mixof a society'seducational investments. Families and ind'.' - s withequal access to creditand facingthe same costof capitalwould efficiently invest in educationas K tn of humanc..pital until the privatereturn on additionalschooling declined to the levelof the privatecost of capital. Thus, efficiencycriteria alone would predisposea familyto investdifferentially in the human capital of their children,depending on their perceptionof the abilityand opportunitiesof each child--assumingability and opportunityenhance the privatereturns to schooling(see, for example,Becker 1981). Socialreturns to educationprovide one set of guidelinesby whicha societycould efficiently allocate social investmentsand set prioritiesamong alternative educational programs, producing C5tinctive types of skills and workers. Individualscapture muchof the socialreturn to education,however, so those groups that obtain high privatereturns for their childrenwill be strong supportersof the status quo. Rent-seeking behaviorin the upper and middleclasses is commonlya potent politicalforce in low-incomecountries to raise quality,maintain free access,and if necessary,even ration entryinto the better secondaryand higher publiceducational institutions. Though high social returns signal the need to expanda particularsegment of the schoolsystem, if theyare buttressedby highprivate returns, the publicsector should consider using fees (and scholarships)to financesuch an expansionor allowinga privateeducational system to satisfythe excessdemand. If socialreturns are moderateprimarily because of socialexternalities, and consequently privatereturns are relativelylow, the publicsector shouldexpect to rely more on its own financingfor schoolexpansion. An alternativecriterion for allocatingpublic subsidies to educationcould emphasizethose activitiesfor whichthe socialexternalities alone would justify public subsidies for education.Externalities from education are thoughtto be substantialat the basicprimary level and to diminishat the secondary,higher, and more technicallyspecialized levels of universitytraining--levels at whichindividuals capture more of the social benefits (minus taxes) from their education(Weisbrod 1964). Socialexternalities are particularlyweUl documentedin studiesof effectsof women'seducation, but they are rarelycited as a reasonfor expanding publiceducation for women. Research that is often producedin combinationwith universityeducation may itself generate external benefitsthat are an importantsource of economicgrowth. The productsof universityresearch are generally freelyavailable to firmsand households,but partlybecause of their accessibility,their productivevalue is difficultto assess. This potential externalityof universitytraining allied to research has not been quantitativelyevaluated in many low-incomecountries, to my knowledge.A possibleexceptibn is in the agriculturalsciences, in whichresearch and extension are sometimesresponsively connected (Evenson 1988).

Patters in OverallLevels of Social and Private Returns Psacharopoulos(1973 and 1985)has summaizedseveral estimates of rates of return to educationby region and schoolleveL Table 2.1 impliesseveral general patterns in these returns. For example,the more developedthe country,the lowerthe socialreturn to education,both acrosscountries and withincountries over time. With notable exceptions,the higherthe levelof schoolingis withincountries, the lowerthe return. The social returnsappear to be about twice as large in Africa and LatinAmerica as they are in industriallyadvanced high-income countries. Socialreturns in Asia failbetween those extremes. With the slowingof economicgrowth in some countriesin the late 1970sand 1980s,returns to educationmay have declined,such as in Ghanaand inArgentina, altering the patternsshown in table2.1 whichare based mainly on studiesconducted on data fromthe 1960sand early 1970s. In manylow-income countries, the privatereturns to highereducation are twicethe socialreturns, because the publiccosts of educationare a relativelylarge share of total costs. The exceptionis Asia, wherethe socialreturns to secondaryand higher education for the samecountries are onlymoderately lower than the privatereturns. In this region,the publicsubsidies are onlya moderateshare of the total costs. In contrast, Africa and, to a lesser degree, Latin Americahave providedlarge public subsidiesfor secondaryand particularlyfor higher education. Returnsto Education 45

Althoughthe use of the mostrecent studymay make these figuresmore up to date, many countrieshave onlya singleset of estimates,and these come fromthe 1960sor early 1970s.More recentstudies tend to find lower returns in many parts of Sub-SaharanAfrica and Latin America,and to documentdeclining returnsfor primaryschooling as the majorityof new entrantsto the labor force in these countriesextend their schoolingbeyond the primarylevel.

Table 2.1 AverageSocial and Private Rates of Returns to Education by SchoolLevel

Social Private Region Pdmay Secondaiy Higher Pdmary Secondary Higher

Africa 27 19 14 45 28 33 (12) (12) (12) (9) (9) (9) Asia 18 14 12 34 15 18 (9) (11) (11) (5) (8) (8) LatinAmerica 35 19 16 61 28 26 (8) (8) (8) (5) (5) (5) High-Income 12 Countries 13 10 8 19Y 12 11 (6) (15) (15) (7) (14) (15)

Note: Numbersof countriesreported in parenthesesbelow mean. I/ Europe,United States,New Zealand,Israel. k/ Notcalculable in the majorityof high-incomecountries, where the comparisongroup withouta primary educationis smalland highlyunrepresentative at youngerages. Source: Calculatedby the author for mostrecent yearswhen social returns were availableat all levelsor all but primarylevel for high-incomecountries. Originalstudies summarized by Psacharopoulos (1973)table 14,and (1985)table A-1.

Withina country,the patternof diminishingsocial marginal efficiency of humancapital investments supports the viewthat publicsubsidies should focus first on the expansionof primary and then on secondary schooling. These levelsnot onlyoffer the highestreturns, they are also the most widelydistributed and henceare more likelyto be equitablydistributed across economic classes. Private returns tend to be highest in Africa,where large educationalinvestments began only recently. The high privatereturns in Latin Americamay reflect the sluggishexpansion of publicschools at the secondarylevel in recent decades. Portionsof East and SoutheastAsia and high-incomecountries have already achieved substantial levels of human capital investment. In fact, some studies suggestthat the social returns to schoolingin these countriesare roughlyon a par with the privatereturns from physicalcapital after taxes. Thus, in some regions, public investmentsin education could be increased efficientlyand equitably,whereas in others--morerapidly growing countries--the case for publicinvestment in schoolexpansion depends in part on new evidenceof socialexternalities (see, for example,Lucas 1985; Romer 1983). Studiesof the rate of return to schoolingbased on aggregatedata are often replicatedby estimating earnings functionsfrom individualsurvey data. In these logarithmicwage or earnings functions,the coefficienton yearsof schoolingapproximates the privateinternal rate of returnto schooling(Mincer 1974; Rosen1977). The estimatedcoefficient on yearsof schoolingusing this method tends to be slightlysmaller 46 Returnsto Education than the estimatedprivate return to schoolingderived from tabulated data on earningsand educationcosts by agegroups, as ri unallyperformed by Becker(1964). Wage function approximations of the privaterate of return to educa . rangefrom 13-25percent in LatinAmerica and Africato 6-9percent in high-income countries(see, for e-vnple, Psacharopoulos1985, table 3). This growingbody of data and analysisleads to the conclusionthat in most developingcountries, the demandfor moreeducated labor appearsto be increasingat least as rapidlyas the supplyof more educated labor, althoughin somecases macroeconomicswings in economicgrowth causetemporarily high or low returnsto schooling.Relatively little analysis has addressedhow structural adjustment has affectedprivate returns to educationin the short run and over the longerterm. Moreover,al the calculationsof returns to educationneglect the consumptiongains associated with schooling and anyexternalities, or "publicgood," attributesof education,because consensus is lackingon howto measure and value these benefits.

Estimating Schooling Retuxia Manyissues arise in the empiricalmeasurement of returnsto schooling.The firstclass of problemsinvolves the estimationbias introduced by inadequatespecification of studentability, parent background,and school quality. Ideally,this bias could be correctedby agreementon specificationand by better data (Schultz 1988b). Most studies,however, must build on imperfectspecifications and data. The directionof the bias these problemsmay cause in estimatingreturns to educationis unclear,as is their influenceon the level of returns for womenrelative to those for men. The seconddass of problemsarises because only a portionof the populationis askedto report the wage and related productivityinformation needed to estimatea wagefunction. A bias is likelyif selectioninto this sample of the populationis not independentof the schooling-wage-productivityrelationship. The selectionproblem is likelyto be evenmore seriousif the criterionfor inclusionin the sampleis choiceof occupationor migration;plausibly, both are linked to the educationof workers. In these cases, the problemsof sample selectioninvolve jointly explainig labor market behaviorand the determinantsof wages. Many limitationsto the use of intergroupcomparisons for estimatingreturns to schoolingare beyondthe scopeof this paper; they are discussedelsewhere (see, for example,Griliches 1977, Schultz 1988b). One such Uimitationis the inadequacyof the statisticalcontrols to account for individualability or job characteristics(that is, the biasof omittedinputs or selectionby comparativeadvantage) (Willis and Rosen 1979;Heckman and Sedlacek1985). Anotheris the neglectof qualityof educationin the wage function; becausequality tends to be directlyrelated to the quantityof schoolingindividuals receive, the returnsto additionalyears of constant-qualityschooling may be overestimated(Welch 1966; Behrman and Birdsall 1983). No consensusexists on preciselywhich variables should enter the wage function or how to identify statisticallythose individualswho invest in schoolingor work for wages in the labor force. Different strategiesfor constructingthe comparisongroups may yield privaterates of return to schoolingthat are substantiallylower, or sometimeshigher, than the simplestlogarithmic wage functionthat includes(1) yearsof schoolingin linearor splineform and (2) a quadraticin yearsof post-schoolingexperience (Mincer 1974). The controversiesover the methodsof estimatingreturns to schoolingdo not appearto be germane to the purpose of comparingthe rates of return and determininginvestment priorities between male and femaleeducation, because most of theseproblems that mightbias estimates of returnsto educationoperate similarlyfor womenand men,with the possibleexception of laborsupply. Nevertheless, assessing precisely the absolute levelof such returns to privateand public resourcesinvested in the educationof women Returnsto Education 47 remains an important,if subsidiary,objective. 1 ' AppendixB to this chapter discussesmore fully these differentestimation problems.

Sex-specificEstimates of Market Returns to Education: A Selective Review Many studies have reported private rates of return to schoolingof men and women (summarizedin table 2.2). The studiesare not comparable,however, because they use differentconceptual and empirical methods. Severalrelated aspectsof the estimationmethodology need to be revisedand standardizedto improveour confidencein the comparabilityand accuracyof such estimatesfor determininginvestment priorities.

Table 2.2 Returns to Educaton by School Level and Gender

Estmation wah . Second=r Higher Country(Year) Method Male Female Male Female Male Female

Colombia,Bogota (1965) II 18.2 nr 34.4 18.9 4.5 5.3 Kenya(ca 1960) I 21.7 7.1 23.6 19.5 nr nr Malaysia(ca 1960) 9.4 9.3 12.3 11.4 10.7 9.8 Brazil(1960) 17.9 38.6 nr nr nr SouthKorea (1971) V/ nr nr 13.7 16.9 15.7 22.9 Taiwan (1982) 8.4 16.0 nr nr nr nr Puerto Rico (1959) 29.5 18.4 27.3 40.8 21.9 9.0 Andra Pradesh, II 8.9 11.8 8.7 11.9 6.2 8.9 India (1977) I 7.2 0.3 6.8 2.41 5.5 5.5 IvoryCoast (1985) II 18.3 5.5 17.0 28.7 21.1 13.6

nr = Not reported. N.t: I - includes participationrate in labor force to deflate returns, depressingfemale returns disproportionately. 11- estimateswage rate relationshipwithout labor force adjustment. i/ Estimationmethod not known. Sources: Colombia,Bogota--Schultz 1968, table 7, hourlywage; Andra Pradesh, India--Tilak1987, table 6.8), averagereturn for schoollevel; Ivory Coast--van der Gaagand Vijverberg1987, appendix 2; other countries--Psacharopoulos1973, table 4.5, and 1985.

First, how shouldthe potentialproductivity gains attributable to schoolingbe adjustedwhen peoplework differentamounts of time in the labor force? Second,how should the effectof educationon nonmarket

I/ For example,if more able studentsself-select themselves into the more educatedcomparison group, the returns to educationcould be overstatedunless a selection-correctedmeasure of the privatereturns to schoolingis computed(Willis and Rosen 1979). But this fonn of selectionmight not necessarilybias the comparisonof the directlymeasured retuns of men relativeto women,if abilityoperates in an analogous wayto influencewho goes to schoolamong both boys and girls. 48 Returnsto Education productivitybe estimated and combinedwith market productivityeffects to yield average returns to schoolingfor the entire popu!ation?Third, whattechniques can correctfor the potentialbias that enters these studies when direct obse:vationson productivityare inevitablylimited to only a portion of the population--say,to those who report labor incomeand hours worked? As mentionedabove, many studies of the returnsto educationtend to includeonly wage earners and thus excludethe self-employedand unpaid familyworkers for whom a wageis difficultto infer or measure precisely.This excluded group is a relativelysmall segment of the laborforce in high-incomecountries, but it is a major part in low-incomecountries; for example,nonmarket workers, the self-employedand family unpaid workers togetherrepresent about half of the adult male populationand two-thirdsof the adult femalepopulation (3chultz 1989). Thus, any bias from sampleselection could be importantin analyzing the educationalreturns for men, as wellas for women,in low-incomecountries (Nakamura and Nakamura 1988). Table 2.3 summarizesthe privaterates of return to yearsof schoolingfor men and womenin severalLatin Americancountries. All of the related studiesused a commonconceptual and statisticalmethodology (Schultz1980a). Although these estimates deal plausiblywith the laborsupply issue by analyzingthe hourly wagewhich does not erroneouslydeflate female returns by labor force participation,they do not correct for wagesample-selection bias. Because the samples are small, some of the estimatesare imprecise. Overall,however, these studiesprovide no clear evidencethat returnsto schoolingdiffer systematically by gender. Returnsdo differconsiderably across countries,however, presumably because of differencesin macroeconomicconditions and the levelof past investmentsin schooling(Schultz 1988c).

Table 23 Estimates of Private Retuns to Education, in Selected Labor Markets and Yes (pecen per annum)

Intemal Rate of Retu Year of Ages 25-44 Ages 45-65 Site Survey Men Women Men Women

Argentine, BuenosAires 19801/ 9.3 6.6 10.0 110 (estimatessimilar in 1976) Bolivia,La Paz 1980 9.8 11.0 9.6 67 (estimatessimilar in 1976) Brazil,Sao Paulo 1971 5.4 6.3 6.0 61 (estimateshigher in 1980) Colombia (estimateslower in 1980) 1973 18.0 18.0 16.0 140 Paraguay,Asuncion 1979 11.0 8.0 10.0 110 (estimatessimilar in 1977) Peru 1974 14.0 14.0 11.0 190

Note: Private returns refer to the estimatedcoefficient (times 100) on the variableyears of completed educationin a logarithmichourly wage rate regressionwhich also includes post-schooling experience, experiencesquared, and someregional or migrationorigin variables. Samples vary in size from21 to 3,478,but all estimatesare statisticallysignificantly different from zero at the .001confidence level (t > 2.83). The selectiononly of workersin the laborforce for whomwage rate couldbe calculated is not treated as a specificsource of bias in these estimates. A/ The availableage groupsin Argentinaare 25-49and over49. Source: Schultz(1989). Returns to Education 49

An analysisof the returns to education in Andra Pradesh, India illustrates how sensitive the calculationsof returns are to the treatment of the rate of labor force participation by women (Tilak 1989). Both the private and the social returns to schoolingat virtuallyevery level are greater for women than for men when the returns are adjusted only for unemployment. When nonparticipationin the labor force is also factored into the calculation,as Becker proposed, the private rate of return for women is less than it is for men (see results for Andra Pradesh in table 2.2). This study also documents the lower public cost of female versus male education and the lower opportunity cost of time for female than for male students. The adjustment for nonparticipationof women apparently is not introducedto deflate the opportunity cost of female student time but only to deflate the stream of benefits from work in the labor force. A study of Sri Lanka also confirmed that women's return to schooling, when not deflated by labor force participation rates, exceeds that of men. The private rate of return to completing the general certificate of exams at the end of secondary school is three times higher for women than for men in urban areas (36 versus 13 percent) and twice as high in rural areas (14 versus 7 percent). At the university leve-l,however, the rates of return for men and women in Sri Lanka appaar to converge (Sahn and Alderm-. 1988,table 17).

Fixed-effectestimation procedures provide another approach to eliminating the bias that may arise from certain types of unobserved or owmittedvariables in a relationship. A cross-sectionis drawn from a number of distinct localities in which the price and quality of market goods and public services may differ. If these local market variables influence tht productivityof schoolingor its quality, then they should be controlled for in estimating the returns by varying only the years of schooling. In this case, introducing a fixed effect for every school district into the wage function removes any bias attributable to the oiaission of school quality, which might be correlated with the quantity of schoolingreceived. The effect of other local market variables, such as size of schools, cannot then be estimated, because they do not varywithin the community. To the extent that the quality of local schools changes at different rates over time across regions and that individuals in a school district move across regions or attend more than one school and work in regions different from those in which they attended school, the community fixed effect becomes a less adequate control for school quality. Family background probably has its own impact on average ability, through genetic and eavironmental mechanisms that instill motivations and habits and also influence the quality of schooling that siblings receive. If these family background characteristics affect productivitv and are correlated with years of schooling,their omissionfrom the wage function would in all likelihoodalso bias upward the estimated rate of return to schooling. One strategy for dealing with these unobserved characteristics is to introduce fixed effects for each family. The estimates of schooling returns are then based only on within-ramilyvariations in worker productivity. This procedure, however, may increase the relative importance of measurement error by eliminatingall between-familyvariation. Exaggerated measurement errors would bias to zero the within household fixed-effectestimates of schooling returns (Griliches 1977 and 1979). )usehold fLxed effects are, therefore, likely to represent a lower bound on the estimates of the effects < schooling on market productivity. Moreover, the restriction of the estimatioa sample to those residinp -n a family that has another wage earner may itself seriouslydistort the comparisongroup and thus bias t return estimates in other directions.

Private rates of return to schooling have been estimated for men and women from a 1986 survey of Indonesia. This study compares standard estimates of the wage function to those that include both communityfixed-effects (proxy for school quality and the like) and householdfixed-effects (proxy for family background correlates) (Behrman and Deolalikar 1988b). Table 2.4 shows returns for three increments of 50 Returns to Education schooling--primary,general senior high school and university./ Privaterates of returns to schoolingfor womenin all comparisonsexceed those for men, and (as expected)the estimatesthat includecommunity and householdfixed-effects are between9 and 24 percentsmaller than thoseobtained from the standard regressionsthat includeinterfamily and intercommunityvariation. These estimatesdo not attempt to controlfor the potentiallyunrepresentative character of the sampleof wageearners, nor are the two sets of estimatesbased on the same sample.,because 16 percentof the wageearners includedin the first set of estimatesapparently did not reside in a householdwith another wageearner and are thereforeexcluded fromthe fixed-effectestimates (Behrman and Deolalikar1988b, table 3). Differencesin the privaterates of return estimatedfor womenand men in Indonesiachange moderately as fixedeffects are addedfor the communityand household,with the differencedecreasing returns at the primaryschool level and increasing returnsat the universitylevel.

Table 2.4 Implied Private Retu-n to an AdditionalYear of S&hoolingin Indonesia by Gender @en per annm)

General Se; Controls Primwy Senior High University Females: Withoutfixed-effect 9.1 11.8 12.4 controls (21.1) (43.4) (27.8) YWithcommunity and 6.9 9.6 10.9 familyfixed-effects controls (173) (353) (24.7) Males: Withoutfixed-effect 7.6 8.2 9.2 controls (2.64) (10.9) (6.41) With communityand 6.1 6.2 8.4 familyfixed-effects controls (1.43) (10.8) (5.23)

Note. The absolutevalue of t ratios are reported in parenthesesbeneath the coefficientsin the case of femalereturns, and for the differencebetween the maleand femaleregression coefficients beneath the male returns. Thus, a significantt ratio under a male return suggeststhat the rate of return on schoolingfor men and womendiffer by a statisticallysignificant amount in this pooledearnings regression. Source: Behrmanand Deolalikar(1988b), table 2.

Chiswick(1976) developed a techique for includingself-employed workers in the estimationof an annual earningsfunction along with wageearners, thereby avoidingsample-selection bias due to analyzingonly

V A logarithmicmonthly earnings function is estimatedpooling men and women;this functionincludes dummyvariables for nine levelsof schoolingand a quadraticin age. Parametersare estimatedfor the differencebetween male and femalecoefficients for all variables,including the intercept. The community and householdfixed effects are believedto controlfor possibleschool quality variation and the effect of familybackground on eanings. Unfortunately,the ordinaryregression estimates and those includingthe fixedeffects are for differentsamples, raising the possibilitythat the differencesreported may be dueto the differentsamples and not due to the introductionof the fixed-effectcontrols. Returns te Education 51

wage earners. Her approach attributed a share of self-employedearnings to entrepreneurial capital or risk- taking. Based on an analysis of Bangkok from the 1971 SocioeconomicSurvey of Thailand, male wage earners (not self-employed at all) received a 10.4 percent return on their years of schooling, whereas females received a 14.5 percent return. Includingpart- and full-time self-employedin the sample reduced the returns to schooling only slightly, to 9.1 percent for men and 13.0 percent for women. The inclusion of the self-employedincreased the urban estimation sample by 39 percent for males and by 53percent for females. In both cases, women's returns exceeded those for men, but those who are self-employedreported slightlylower returns on their schooling than did wage earners. This is bro&dlyconsistent with the pattern of more women with education beyond primary school working in wage jobs than the less educated. Whether under-reporting of incomes by self-employedbiases such estimated returns to schooling--andif so, by how much--is uncertain.

Few studies of the relationships between wage rates and schooling have assessed how taking the selective sample of wage earners biases fndings. (see, for example,Anderson 1982;Mohan 1986;Griffm 1987;King 1989;Schultz 1988b). Moreover, these studies often deal with men alone or women alone and thus do not help assess whether the bias due to sample selection modifies systematicallycomparisons of male-female estimated returns to education, as reported in tables 3.4 and later in table 3.5. This is an important issue for public policy but one that has receivecdsurprisingly little empirical study.

Griffin (1987) analyzes the earnings of married women in the Philippines in 1980to appraise estimates of schooling returns subject to alternative methods for dealing with sample-selection bias. He estimates a nonmarket (reservation) wage function and a function for market wage offers. The reservation wage function determines the shadow value of nonmarket time of the individual,and hence what the individual requires to induce him or her to enter the market labor force. Heckman's selection-corrected model is identified within the context of the family or bargaining labor supply model. A standard log-linear speification of the earnings function is estimated in which returns to schoolingare constant across schooling levels. The selection-correctedmaximum likelihood estimate of schoolingreturns is 18 percent, compared with the conventional estimate of 14 percent (based on only the one-third of the sample who earn wages). In this case, adopting a sample-selectioncorrection procedure increases, the estimated returns to schooling for women and the selectivityterm is statisticallysignificantly different from zero, showingthat the sample of wage-earning women is not a random sample of the population with regard to their wage rates. King (1989) analyzed the earnings of women in the 1985-86Peruvian LivingStandard Survey. A probit equation for women participatingin paid employment(that is,both wageearners and self-employed)is used to estimate an hourly earnings function with Heckman's (1979)two-stage procedure. The sample-selection correction decreased the rate of return for women from 12.2to 12.0percent for primary school, from 8.0 to 7.8 percent for secondary school, and from 6.8 to a -1.7 percent at the universitylevel (if a diploma is received after four years of study). As in Griffin's study, the family's nonearned income and husband's characteristics are included only in the paid-participationequation (along with the woman's marital status and a variety of more controversialidentifying variables).

Khandker (1989) subsequentlyused the same Peruvian LivingStandard Surveydata to examine the returns to schooling for both men and women. He restricts his analysisto wage earners and identifies the sample- selection probit equation by the family's land holdings and unearned income, as well as the individual's marital status. The return estimates appear to be relativelyrobust to variations in this list of identifying variables included only in the sample-selection equation. For the country as a whole, women's returns increase when controls are introduced for sample selection, and returns are then marginally higher for women than for men at the secondary and higher schooling levels. At the primary school level in the metropu2itanarea of Lima, however,the returns are low for both sexes, but they are, lower for women than for men (2 percent compared to 2.5 percent). The same pattern of low returns to primary schoolingfor

2/ The nonmarket (reservation) wage of the wife is affected by the family's land, ownership of a business, assets, nonearned income, and her husband's education and experience. These variables are added to the wage-status probit equation. The Heckman (1979) two-stage estimate, which is less efficient, yields an estimate of private returns to schooling of .16 (Griffin 1987,table 3). 52 Returns to Education

women has been noted elsewhere in metropolitan Latin America (see Schultz 1968) and has led some researchers to attribute this pattern to domestic servantswhich they then excludefrom samples in estimating wage functions because of the difficulty of valuing the income they receive in kind, i.e., food and bread (Mohan 1986). The 1976, 1981, and 1986 Socioeconomic Surveys of Thailand permit further evaluation of the effect of samnple-selectionbias on estimates of the private rates of return to schooling for both women and men. In this case, the analysis incorporates two selection correction terms representing the probability of being in the labor force and being a wage earner. These selection equations include family nonearned income, hectares of family land that are irrigated or unirrigated, and the standard market wage rate determinants, including years of schooling completed at the primary, secondary, and higher education levels. Land ownership and nonearned income raise the nonmarket reservation wage and thereby reduce the likelihood that a person will take a wage job or work at all in the market labor force. Education does not exert a monotonic effect on the labor force and wage earner status of Thai men or women (Schultz 1989). An individualwith primary schoolingis less likelyto be a landless wage laborer in this primarily agricultural country. The more years of secondaryschooling an individualhas, the higher are the chances that she or he is working for a wage. Each year of universityeducation strongly increases the likelihood of working in a wage or salaryjob. Herein is a clue why the sample-selectionbias can operate in different directions at different levels of schooling. Many landowners are also wage earners. If they have enough land, however, they presumablywithdraw from the wage market to cultivate their own land full time. The critical question is whether land is exogenousor merely a proxy for self-employment;in other words, is land a legitimate variable to use to identify the selection model? An additional problem in specifying a wage function to estimate returns to education in low-income countries is how to model the regional segmentation of labor markets. Without interregional migration, wage functions should be estimated separately for each region. The wage differences related to education within a region would then be the appropriate parameter determining priorities for private investment in schooling in that closed region. Interregional migration does occur, however, and more educated persons tend to migrate more frequentlythan others. Generally,they move from lower to higher wage markets and from rural to urban areas. In Colombia, for example,as much as half of the lifetime returns to schooling for the children of rural residents is realized by the increased likelihood that the children will migrate from the rural to the urban labor market (Schultz 1988c). Analysisof this would require knowingwhere people migrated from, the costs they incurred in moving,and where they received their schooling. Alternatively, holding an individual's current residential regional labor market statisticallyconstant purges from the wage function an estimate of the return to schoolingthat arises from higher wage regions which is more frequent among the better educated (as it clearly is in Latin America and the United States) (Schwartz 1976). Models with and without regional shifters provide a useful range of estimated returns to schooling. The greater the interregional mobility (as, for example,in Taiwan compared to China), the stronger the case for treating the entire country as a single labor market when estimating school returns. Regional labor market nominal wage differences alo may reflect compensatingvariation for price levels and reinforcing variation in the quality of subsidized public services. Nominal wage differences may not, therefore, measure accurately real wage differences. Urban high-wage regions have more and better schooling,and regional shift variables in a wage function may also reflect this difference in the quality of schooling embodied in workers across regions (Behrman and Birdsall 1983). Other than school and health services,other prices particularlyfor housing,are generally higher in urban high-wageregions. On balance, regional nominal-wage differences probably exceed real-wage differences,if public services are a relatively small part of familyconsumption. Estimating the participation and wage functions with and without regions as explanatory variables should at least help to assess the importance of migration in the estimation of school returns. The lack of information on migration in the Thailand survey data does not permit any further analysis of this issue here. The selection-correctedprivate rate of return estimates are contrasted in table 2.5 with those based on the ordinary least squares (OLS) estimates for wage earners that ignore the potential sample bias arising from selecting only wage earners. Two selection probit equations predict the probability that the individual is Returnsto Education 53 in the laborforce and is a wageearner (Catsiapisand Robinson1982). Both the wagefunction and the two selectionequations vary across regions in Thailand: the least developedNortheast region, the rural population,the suburban sanitarydistricts, the urban municipalareas, and finallyBangkok. Because regionalshifters are specifiedin the wagefunction, the privatereturns to educationexdude the gainsthat accrueto educationthrough the more frequentmigration of more educatedpersons to regionswith higher wages.

Table 2.5 Estimats of Private Rates of Return to Schooling in Thailand,by Gender, With and Without Statistical Correctionfor Sample-Selection Bias"

Without Comecdon#' With Selection Cofection-l Year-Unit of Earnings (Sample of Eamners/Population)Primary Secondary Higher Primaiy Secondary Higher I. 1986--MonthlyEarnings Female 8.2 3 09.5 13.0 25.0 18.0 (2,709/8,606) (4.75) (18.7) (4.31) (7.00) (9.84) (5.45) Male 14.0 18.0 12.0 17.0 6.8 7.8 (4,199/7,685) (9-40) (14.4) (6.81) (113) (5.34) (4.61) 1I. 1981--MonthlyEarnings Female 4.6 30.0 2.2 9.0 22.0 12.0 (2,419/8,816) (2.41) (19.7) (.74) (4.56) (6.09) (332) Male 15.0 20.0 4.2 15.0 8.8 2.9 (4,525/7,986) (10.2) (17.4) (1.72) (9.22) (6.24) (122) m. 1981--HourlyEarnings Female 5.2 34.0 1.6 10.0 25.0 11.0 (2,419/8,816) (2.7() (22.1) (.55) (5.00) (6.77) (3X4) Male 16.0 24.0 5.4 14.0 13.0 4.1 (4,525/7,986) (9.98) (20.9) (2.16) (8.95) (9.04) (1.66) IV. 1976--MonthlyEarnings Female 11.0 17.0 7.8 11.0 31.0 17.0 (1,464/9,430) (8.87) (13.8) (3.63) (8.70) (7.04) (6.35) Male 5.7 15.0 8.4 5.6 7.4 11.0 (3,783/8,836) (6.37) (16.5) (5.50) (533) (7.65) (685)

Note: Doubleselection correction terms are includedin the earningsfunctions to capturethe probability of participationin the labor market and of selectingwage employment See text for identifying restrictions. i/ The estidmationsample is restrictedto wageand salaryearners betweenthe ages of 25 and 54. k/ The absojutevalue of the t ratio is reported in parenthesesbeneath regression coefficient on years of educationwithin each levelof schooling. ~/ The absolutLvalue of the t ratio is reported in parentheses,but it has not been adjustedfor the selectioncorrection procedure, and is thereforepotentially biased. Source: Author's calculationsfollowing methodology outlined in text;and Schultz(1989, tables 4,A-7 and A-8) for 1981survey. 54 Returns to Education

For most of this century,Thailand has investedheavily in primaryeducation. Still, it enrollsa smaller proportion of its populationin secondaryschool than do other countries at a similar stage in their development,such as South Korea,Taiwan, the Philippines,and Malaysia,or the two city states of Hong Kong and Singapore(Sussangkarn 1988). On the other hand, the proportion of the Thai population enrolledin higher educationis relativelylarge for a countryat its incomelevel. The relativesupply of workersby educationallevel would lead to the expectationthat in Thailandthe returnsto educationwould be relativelyhigh at the secondarylevel an4 relativelylow at the universitylevels, comparedto other countriesat Thailand'sstage of development. Primaryschooling has been nearlyuniversal in Thailandfor sometime, and hencethe differenceby gender is smalL The differencesbetween male and femaleenrollments at the secondaryschool level are more substantial,but are narrowerin the 1975-85period. Onlyabout half as manywomen as menwere enrolled in Thai institutionsof higher educationin the 1970s(UNESCO 1984). As reported in table 2.5 for 1981,for example,without sample-selection correction, women's hourly rates of return appear to be 5 percent a year for primaryeducation, 34 percent for secondaryschool, and 2 percent for universityeducation. For men, the OLS primaryschool returns are 16 percent; secondary school,24 percent; and universityeducation, 5 percent. The statisticalcorrection for the two sample- selectionprocesses that mightbias these estimatesmodifies the estimatesmarkedly in four out of the six cases. Three out of four selectionterms are statisticallyhighly significant (Schultz 1989). The return to primaryeducation for womendouibles to 10 percent,whereas the returnsto secondaryschool decline for both women(to 25 percent)and men (to 13 percent). Higher educationprivate retuns increaseto 11 percentfor womenand declineto 4 percentfor men. The analysisis repeatedwith 1981monthly earnings, thereby induding labor supplyadjustments related to schoolingas part of the privatemarket returns. In four out of six comparisons,the selection-corrected monthlyretuns are slightlylower than the hourlyretuns. The better educatedwork fewerhours a month, or enjoymore leisure,responding to educationas a gain in their wealth. Exceptionsare womenat the highereducation level and men withonly some primary schooling, who increasetheir labor supplyas their educationincreases. OveralLhowever, the hourlyand monthlyearnings data yieldsimilar results for the directionand magnitudeof the effecton schoolingreturns of the correctionfor sampleselection. Onlythe monthlyearnings functions can be estimatedfrom the earlier and later Thai surveysbecause respondents were not askedabout hoursworked during the previousmonth. By 1986,male and femalereturns to primaryschooling had increasedfrom 1981,and the correctionfor sampleselection continues to raise these returnsfor women.At the secondaryschool level, women's high privatereturns to schoolingare diminishedsomewhat by the correctionprocedure, but theyare more than three timeshigher than male returns at this schoollevel. Fmally,at the higher educationlevel, private returns are higher for both Thai womenand men, thoughthe sample-selectioncorrection has again a greater effect of doublingfemale eturns. Only at the primary schoollevel are the returns to male educationgreater than those to femaleeducation. Goingback to 1976,the pattern of schoolingretuns is roughlycomparable. Higher education earned slightlyhigher returns for men in 1976,11 percent, compared with3 percentin 1981and 8 percentin 1986;the rapidexpansion of public"open" universities in thisperiod may have reducedthe overallquality of a year of highereducation. Muchmore workis neededto assessthe effectof alternativemethods for copingwith sample-selecion bias as it affectsestimates of returnsto educationof womencompared to men. Nevertheless,these data from Thailandspanning eleven years suggest that estimatedlevels of returnsto schoolingmay be sensitiveto this sourceof samplingbias The overalltendency is for the sample-selectioncorrection to raise returns to female schoolingand lowerthose to male schooling.With the exceptionof primaryschool returns after 1981,women's schooling appears to earn a more favorablereturn than does men's schoolingin Thailand. An understandingof whatunderlies an individual'sallocation of time isrequired to correctanalyses of wage earners that seek to infer the effect of investmentsin schoolingon the productivityof all people. This sample-selectioncorrection procedure depends on whatthe relevantfamily unit is that may poolresources Returns to Education 55 and coordinatelabor market behavior,and how these forms of famDlybehavior are modeled. The bargainingmodel implies a few,possibly useful, differences between the empiricalspecification of the family Nash-bargainedlabor supplymodel and that impliedby the unifiedfamily demand model. On the whole, however,both modelsrequire similarvariables for identificationof the sample-selectionrule determining wage earner status. Conclusionsdrawn from this section are likelyto be robustto changesin how the familydecisionmaking process is eventualy modeled.

Non-market Returns to Schooling The effectsof schoolingon marketearnings are relativelywell documented, although sometimes subject to uncertaintybecause of problemsof measurementand estimation. The evidenceon the returnsto schooling in non-marketproduction within the householdis morefragmentary, however, and is inherentlydifficult to aggregateor summarizein a singlemeasure such as an internalrate of return. The greaterthe levelof educationof marriedwomen, the more likelythey are to workin the labor market and in wage employment,given their husbands'education and businesscapital. This pattern has been observedfrequently in surveysin low-incomecountries, particularly in urbanareas and in rural areas where off-farmemployment opportunities for educatedwomen are reasonablydeveloped. Thus, to workin the labor market,more educatedwomen must curtail their nonmarketproduction activities or at least find substitutesfor their time in nonmarketproduction. At the same time, many studies reveal that quantifiableincreases in home output occur as women's educationincreases, despite the fact that they are likelyto spend less time in the home. That is, the productivityof womenin homeproduction appears to increaseas their educationincreases, indicating that nonmarketreturns to schoolingare positive.

Determinantsof Child Health and Survival Studiesin demography,economics, anthropology, and sociologyconclude that a stronginverse relationship existsbetween a mother'sschooling and the incidenceof mortalityamong her children. This relationship is particularlystrong in low-incomecountries. The patternhas been widelyreplicated across comparative surveys,such as the WorldFertility Surveys, and overtime based on repeated censuses. Manyhypotheses have been advancedas to why(see, for example,Farah and Preston1982; Schultz 1984; Barrera 1988b). Data that are nowwidely coDected from women in low-incomecountries on their age, childrenever born, and childrenstill living facilitate analyses of the determinantsof childmortality. Additional information on the timingof each birth and survivalor date of death of the offspring,improves the measurementof the mortalityrisk faced by each of a woman'schildren. These refinements are particularlyuseful in comparing the childmortality experience of youngerwomen. Educationaldifferentials in childmortality are not very sensitiveto whichof these proceduresis used to comparewith simple child survival (Preston and Trussell 1982). The statisticalstrength of the relationshipand its replicabilityacross surveysand societiesis reminiscentof the "discovery"in the 1960sof the logarithmicwage function, which also depended centraly on education-baseddifferences in wagerates. One is a measureof the marketrate of return, the other a form of the nonmarketreturn to women'seducation. An added year of maternaleducation tends to be associatedwith a relativelyconstant percentage change in childmortality rates. Althoughmortality tends to be higherin rural than in urban areas in manylow- incomecountries, the proportionatereduction in child mortalityassociated with an additionalyear of mother's schoolingis about the same,between 5 and 10percent. The mortality-reducingeffect of father's educationis smaler, especialyin rual populations(Mensch, Lentzuer, and Preston1986). Studiesin Latin Americahave noted that the differentialsin child mortalityassociated with maternal educationwere more moderatein Costa Rica and Cuba. The hypothesisfor these deviationsis that these countries' strong public health programs have improved access to health care, even among the 56 Returns to Education

least-educated mothers (Beb;n 1980). Other economic hypotheses for the differences in the relative magnitude of the effects of schoolingon child mortality are analyzedby Rosenzweigand Schultz (1982)and are discussed further elsewhere (Schultz 1984;Thomas, Strauss, and Henriques 1987). Is education simply correlated with the use of mvorehealth inputs, or does education provide a mother with the capacity to cope with health risks and bet;er manage her child's environment? An analysis of the 1973 census in Colombia indicated that controlling for household income, husband's education, or the marital status of the mother did not eliminate or even greatly reduce the independent role of the mother's education as a partial explanation for her children's suruival (Schultz 1980b). Studies elsewhere have shown that although controlling ior many lifetime evcnts and changes in socioeconomic status in relation to child mortality rates is possiole, the mother's edr!cationstill had a substantial effect (Farah and Preston 1982). In Brazil, a third of the mother's education effect on child mortality could bc explained through controls for family income variables (Thomas and others 1987).

In additionto influencingchild mortality,a mother's education undoubtedly influencesmany intercorrelated variables such as migration, labor market behavior, use of health care, and modern attitudes. Controlling statisticallyfor these types of variables is, therefore, likelyto result in an understatement of the net effect of schooling (Mensch, Lentzner, and Preston 1986). The puzzle that remains is why a mother's education explains more of the variation in child mortality than do other variables such as an individual'saccess to health care, the prices of health care, or even total family income that could be spent on health care? Three competing hypotheses are that (1) the better educated mother uses a different mix of observable health inputs; (2' -'ie uses these inputs more effectively,or (3) her education is positivelycorrelated with the use of many ..2inorhealth inputs that are not observed, and her education is credited with the effect of these unobserved inputs on child health (Schultz 1984). The most important health worker for children is their mother. How well she performs this task depends "on her schooling,which equips her with general and specificknowledge, and the means and confidence to seek new ideas" (Barrera 1988a). How does education influence the use of health inputs to reduce the probabilityof childmortality? The answeris sought by studyingthe variations in more continuous indicators of child health status that can be measured through surveys,including the child'sheight and weight at birth. Anthropometric indicators predict lifetimehealth problems and mental and physicaldevelopment handicaps accurately, as well as subsequent age-specificmortality. Further, the health care used by pregnant women has been analyzed to evaluate the effect of this input on the "production"of child healtlh (Schultz 1984). Selection of some health inputs occurs in response to the mother's expectation of a good or poor birth outcome. Consequently,the simple correlation between these forms of self-selected health care and health outcomes can be biased or misleading. For example,prenatal care is sought from a doctor early in difficult pregnandies;such early prenatal care is thus not surprisinglycorrelated with havinga less healthy and lower than average weight of the child at birth. But early prenatal care is nonetheiess beneficial for an average women or for a mother and child whose initial health condition can be statisticallyheld constant. In this United States study, the effect of the mother's education on birthweight is transmitted largelythrough the variation in four measured prenatal health inputs: age, parity, smoking, and timing of prenatal care (Rosenzeig and Schultz 1989). The effect of maternal education on child health in this case is fully explained by education's effect on the use of observed health inputs. The inputs that play a major role in producing good nutrition, good health, sound development,and survival in an older child arc more difficult to measure. A statistical explanation must also be found for which mother uses each of these health inputs, if their effect on the production of child health is to be estimated without a self-selectionbias. Impact on child health of maternal education, health care facilities, aud interactions between mother's education and her constraints in caring for her children's health may be simpler to examine directly. Estimates of the health effects of these interactions document how maternal education exerts its elusive effect on child health. For example,Caldwell (1979)hypothesized that in West Africa, a mother's education enabled her to exploit local public health care more effectively. He suggested that the interaction between mother's education and local public health infrastructure was complementaryor positive: more educated Returns to Education 57 mothers gained most from local public health clinics. According to my reinstatement of his hypothesis, differentialsin child health or mortality,by mother's education,should increase in communitiesserved more intensivelyby a public health system. Rosenzweigand Schultz (1982),however, found the opposite pattern of negative interactions or substitution in Colombia. There differences in maternal education had a smaller impact on child mortality in urban populations that received more public and private hospital and clinic servicesper capita. Their findingscan be viewed as consistent with the aggregate patterns reported in Latin America by Behm (1980) and Palloni (1981) and in Sri Lanka by Meegama (1981). Other studies have examined the relationship between mother's choices of health inputs and environmental constraints on child mortality. In Malaysia,among those householdswith poor water and sanitation facilities, breast-feedingwas associated with reductions in child mortality (Butz, Habicht, and DaVanzo 1984). Estrey and Habicht (1987) found that safe water supplies reduced child mortality by a greater amount for more educated than for less educated mothers, whereas access to toilets in the household was less effective in reducing child mortality in the case of educated mothers. In Costa Rica, Haines and Avery (1982) found that an additional year of a mother's education reduced her children's mortality 6 to 7 percent, holding constant household sanitation, quality of the dwelling,and communitychild mortality levelsand health care facilities. Haines and Avery concluded that the child health gains related to mother's education were smaller in urban areas, a result also found by Schultz (1980b) for Colombia and by Behm (1976) for several Latin America capital cities. Similar studies for Malaysiaand Costa Rica treated the household water and sanitation infrastructure, as well as the mother's breast-feeding, as exogenous, in other words, as not affected by maternal education or unobserved variables that might themselves have otherwise influenced child health outcomes. Barrera's study (1988b)of householdand communitydata from the Bicol Provinceof the Philippines refined these earlier studies. He assumed that the water and sanitary facilities of households were endogenous choice variables that may be correlated with unexplained variations in child health. He first analyzed the relation between maternal education and child health, conditional on the community's average levels of water and sanitation but not on the household's actual variables, which are assumed to be spuriously correlated with the family's other choices. Barrera found that mother's schooling had a larger protective effect on child health in unsanitary communitieswhere signs of excreta were visibleand in communities that were farther (in time) from outpatient health care facilities. In a communitywhere piped water was the predominant source of supply, the impact of mother's education diminished. Where water-sealed toilets were more prevalent in the community, the differences in the effect of maternal education on child health were larger. Because Barrera replaced household-level measures of water and toilet facilities by community-levelones representing the local availabilityof these facilities,he obtained conclusionsthat were diametrically-opposedto those of Estrey and Habicht (1987): the community water supply appeared to substitute for mother's education, while modern toilets complemented it (Barrera 1988b;table 16). At the same time, Barrera ;howed that higher income and mother's education increased the chances that a household had acquired piped water and water-sealed toilets. The duration of breast-feeding, another important input to child health, is inversely related to mother's education in many countries (see, for example,Blau 1984;Wolfe and Behrman 1982). Table 2.6 illustrates how in Africa, Latin America, and Asia, women with seven or more years of schooling tend to breast- feed their children seven or eight months less than do women with no schooling. Breast-feedingis beneficial to child health primarily when it is supplemented by other foods before the end of the baby's first year. In Barrera's rural Philippine population, mother's education shortened only the duration of unsupplementedbreast-feeding (1988b,table 23). Moreover, he estimated that unsupplemented breast-feeding was "beneficial' only up to six months. The more educated mother can also replace her milk with more sanitary substitutes. For the less educated, Barrera hypothesized,supplementation of breast- feeding at less than six months was harmful. Thus, the optimal duration of breast-feeding and the optional time to introduce supplementaryfoods in the child's diet depended on the education of the mother who had to provide sanitary substitutes for her own supply of milk. In sum, the duration of unsupplementedbreast- feeding and education appeared to be substitutes in their effect on child health. This finding may partially account for why more educated mothers breast-feed less,but their children's health is better (Barrera 1988b, table 29). 58 Returnsto Education

Table 2.6 Womens AverageAge at Marriage,Breast-feeding, and Contraceptionby Region and Bducation,for World FertilitySurvey Counties

Region(Number of Countries), Age at Contrace tive Years of SchoolCompleted Marriagel/ Breast-feedingk/ Usage-'

Africa (8 to 12) 0 years 17.8 19.9 7 1-3 19.2 18.5 14 4-6 203 15.7 17 7 or more 23.0 13.4 27 Difference(7+-O) 5.2 -6.9 20 LatinAmerica and Caribbean(13) Oyears 19.5 15.0 24 1-3 19.5 12.1 33 4-6 20.4 9.1 43 7 or more 22.6 5.4 53 Difference(7 -O) 3.1 -8.7 29 Asia and Oceania(7 to 13) 0 years 20.2 20.1 16 1-3 19.5 18.4 26 4-6 20.6 16.0 28 7 or more 23.8 10.6 39 Difference(7+-4) 3.6 -7.1 23

/ Singulatemean age at marriagein years. §/ Meanduration of breast-feedingin monthsusing current status estimates based on survivingbirths only usinglife table methods. s Percent of currentlymarned women aged 15-49 currently using contraception,adjusted for age differencesbetween education groups. Source: United Nations(1987), tables 119,121, and 122. Returns to Education 59 Table 2.7 Median Age of Women at First Birth by Age and Education, Selected African Countries

Am County, years of YoungerThan schooling comsleted 25 25 to 34 3or Older Senegal 0 years 18.3 18.0 17.9 1 - 4 19.3 (18.6) (19.6) 5 - 7 19.7 20.4 (21.0) 8 + 22.4 (23.0) (23.0) Difference (8-0) 4.1 5.0 5.1 Ghana 0 years 18.9 19.3 20.0 1 -4 18.6 19.7 20.3 5 - 7 18.8 20.2 19.3 8 + 20.5 20.7 20.5 Difference (8-0) 1.6 1.4 0.5 Lesotho 0 years 18.7 19.5 22.2 1 - 4 19.1 19.9 (19.8) 5 - 7 19.8 205 (21.1) 8 + 20.8 22.9 22.8 Difference (8-0) 2.1 3.4 0.6 Benin 0 years 19.5 19.1 19.7 1 -4 19.4 19.9 (19.8) 5- 7 19.3 19.7 20.7 8 + 20.2 21.2 21.0 Difference (8-0) 0.7 2.1 1.3 Kenya 0 years 18.2 18.4 19.4 1 - 4 18.6 18.2 19.3 5 - 7 19.1 19.3 19.2 8 + 21.3 20.5 (21.5) Difference (8-0) 3.1 2.1 2.1 Cameroon 0 years 18.2 19.3 20.6 1- 4 18.4 18.8 19.7 5 -7 19.0 19.0 19.1 8+ 21.4 20.6 (24.3) Difference (8-0) 3.2 1.3 3.7 Ivory Coast 0 years 18.3 18.7 19.3 1 - 4 17.6 19.2 20.2 5- 7 18.2 18.7 19.9 8 + 20.2 18.8 20.2 Difference (8-0) 1.9 0.1 0.9 Sudan 0 years 18.0 18.5 20.0 1 - 4 17.6 18.6 20.1 5 - 7 19.1 (19.6) (21.0) 8 + (20.3) 21.6 (23.6) Difference (8-0) 2.3 3.1 3.6 Not: Means reported in parentheses are less reliable because they are based on fewer that 50 observations in the specific age-education category. Source: World Fertility Surveysas tabulated by Eelens and Donne (1985),tables A.9-A.16. 60 Returns to Education Determiants of Child Achievementsin School Surprisinglyfew studieshave addressedthe householddeterminants of schoolenrollment and attendance rates by gender in low-incomecountries. In most studiesof the schoolingof childrenin high-income countries,the mother'seducation has a larger effectthan the father's,even thoughthe father's education impliesa larger marketincome effect because he tends to receivea higherwage and to workmore hours (Leibowitz1974; King and others 1986). Evidencethat the mother's schoolingexerts a greater effecton the schoolingof daughtersthan sonsis less wel established(see, for example,de Tray 19&i:table 5). That hypothesisrequires further study.

Deteminants of Fertility More educatedwomen marry later, as shownin table 2.6 for the countriesincluded in the World Fertility Surveys. In Africa,women with sevenor more yearsof schoolingtend to marry five yearslater than do womenwith no schooling;the differentialis about three years in Latin Americaand Asia. Table 2.7 examineshow median age at first birth varies by a woman'seducation in eight Africancountries, with further disaggregationby age of the woman. The effectof female educationon the age when a woman's child-bearingbegins is not uniformacross Africa, but it is of growingimportance in suchcountries as Kenya and Ghana,where overallfertility levels may soon begin to decline. The countervailingeffects of decreasedbreast-feeding on fertilityas mother'seducation increases (table 2.6) mayremove a monthor so from the interbirthintervals of the most educatedmothers (related to breast- feeding seven to eight months less). The much greater prevalenceof contraceptionamong the more educatedwomen (table 2.6) more than fullycompensates in its impacton fertilityfor this shorterduration of breast-feeding. As shownin table 2.8, total fertilityrates are lowerfor womenwith sevenor more years of schooling comparedto those withno schooling.The differencesby women'seducation are larger in LatinAmerica (-3.6 children),than in Africa (-2.0 children)or Asia (-3.1. children),but they are relativelyuniform regardlessof fertilitymeasure. If one holdsconstant for marital status,or essentiallyage at firstmarriage, (marital) fertilitydifferences by educationare abouta third less than differencesin total fertilityin Africa and Asia,but theyare only20 percentless in LatinAmerica, where overall contraceptive prevalence is high but variesgready by woman'seducation. To comparechildren ever born, the figuresare restrictedto the age group of womenwho havecompleted their childbearing(ages 4049). Thesedata are not informative, therefore,on reproductivepatterns amongyounger women or recent trends. The last aspect of fertilitymeasured through the WorldFertility Surveys is women'sdesired family size. Desiredfertility also falls monotonicallywith a woman'seducation. Subtractingthe desiredfertility from the current total fertilitysuggests that the potentialfor increasedcontraception to reduce fertilitytoward desiredlevels is concentratedamong women in LatinAmerica with less than four years of schoolingand in Asia amongwomen with less than sevenyears of schooling.Among the better educatedwomen, total fertility rates are already approachingdesired fertility levels. This reflects of the fact that women's educationsubstitutes for familyplanning by helpingwomen reach their desiredreproductive goals (Schultz 1989). Returns to Education 61 Table 2.8 Measuresof Recent, Cumulative,and Desired Fertlity: Averagesfor World Ferdlity SurveyCountries Reporting, by Region and Respondenfs Education

Regions (No. of Countries) Total Marital Desired Years of Schooling Fertility Fertility Children Fanib Completed by Women RateC Rate Ever Bomtl Size Afrki (8 to 10) 0 years 7.0 6.3 6.4 6.9 1-3 7.2 6.6 6.5 6.4 4-6 6.2 6.3 6.1 5.9 7 or more 5.0 5.4 4.8 5.0 Difference (0-7+) -2.0 -1.2 -1.6 -1.9 LatinAmerica (13) 0 years 6.8 6.8 7.1 4.8 1-3 6.2 6.2 6.8 4.7 4-6 4.8 5.2 6.0 4.2 7 or more 3.2 3.8 4.2 3.7 Difference (0-7k) -3.6 -3.0 -2.9 -1.1 Asia and Oceania (9 to 13) 0 years 7.0 6.6 6.7 5.4 1-3 6.4 6.4 6.7 4.3 4-6 5.8 6.1 6.4 4.2 7 or more 3.9 4.7 4.9 4.0 Difference (0-7+) -3.1 -1.9 -1.8 -1.4

AAge standardized. / Age 40-49 years. s/ Age adjusted.

Source: United Nations (1987), table 112 and 115.

Other studies also indicate that women's education helps couples to avoid exceeding their reproductive goals. This is partially achievedby delayingmarriage (Cochrane 1979),but more educated wivesalso have fewerunwanted conceptions and births in marriage (Rosenzweigand Schultz 1985 and 1987). Although the husband's education may also enhance the effectivenessof contraception, the wife's education has at least as strong an effect on these forms of reproductive behavior, whether inferred from a respondent's own classificationof conceptionsas unplanned or derived from econometric analysesof the reproduction function and its residual (Rosenzweigand Schultz 1985, 1987,and 1989). Reduced fertility may be another sphere of nonmarket production in which the education of women generates an important beneficial social externality. Although fertility during a woman's lifetime appears to decrease as her education increases, in some cases the fertility of unschooledwomen is slightlylower than that of women with one to three years of schooling (see African region total fertility rates in table 2.8). This occasional reversal in fertility differentials by women's schooling has two interpretations. Easterlin (1975) proposed a framework for describing the demographic transition that accommodatesthe tendency for birth rates to increase at the outset of modern economic growth before birth rates begin their secular decline. This early rise in fertility has been attributed to improved maternal health and decreased breast-feeding,both thought to increasereproductive 62 Returnsto Education potentialbefore contraception was available and widelyused. The rise in fertilityamong women with a few yearsof schoolingcould, therefore,be attributedto the enhancedreproductive potential of womenwho receiveonly a coupleyears of schooling(Cochrane 1979 and 1988). The secondexplanation for this occasionalreversal in the fertilitydifferential associated with a few years of women'sschooling relies on the householdeconomic model of fertility. Economicmodels of family decisionmakingfocus attention on the potentiallydifferent signs of the effects of men's and women's schooling(or valueof time) on their demandsfor childrenand hence lifetimecompleted fertility (Schultz 1981). These differencesin wageeffects on fertility,by source,follow from the customarytendency for womento spendmore time in childcare activitiesthan do men. This economichousehold demand model of fertilityhelps to accountfor the observedtendency in multivariateanalyses for female education(or wages)to be inverselyrelated to fertility,while male education (or wages),land or asset income,and cbild wagesare all directlyrelated to fertility,at least in a traditionalagrarian society (Schultz 1973). At a later stage in the developmentprocess, the fertilityeffect of male education(wages) may also becomenegative becausethe parents investmore heavilyin the schoolingof their children. Thus, the simpleone-way correlation between women's education and their fertilitycan be a misleading indicatorof education'seffect. To estimatethe relevantparial effect,one must hold constantstatistically the other major exogenousdeterminants of fertility.The bias causedby omittingthese other determinants, in the case of men's education(or wages), is undoubtedlyto reduce the estimated negativeeffect of women'searly schooling,because the years of schoolingof husbandand of wifeare stronglypositively correlatedin ail societies,and men'seducation (or wages)is often positivelyrelated to fertility,at least at low levelsof education. All theseroutes by whichwomen's education may improve society-- reduced child mortality, enhanced child nutrition,increased child schooling, and reducedfertility--involve benefits that are partlycaptured by the women'sown children. If parents are entirelyaltruistic toward their offspringin a dynasticform of intergenerationalfamily and theythus view their children'sutility as identicalto their own,these nonmarket activitiesof womenwould be adequatelyencouraged within the famiy. Parents may not viewsons and daughtersas equallyimportant branches of the family,however, or their perceptionsof the payoffto educationmay be basedonly on education'seffects on wagesin the laborforce. Many societieshave strengthened the rights of childrenand weakenedthe economicclaims parents can placeon their children--forexample, through restrictions on the conditionsof childlabor, mandatory school attendance,and penaltiesfor truancy. Also,many societieslink improvementsin women'seducation to increasesin the levelof investmentin the humancapital of children(Thomas 1989). A subsidythat favors women'sschooling would help shift privatehousehold resources toward investments in the qualityof the youngergeneration. Empiricalstudies have not yet widelyappraised how responsivefamilies would be to practicalinterventions that seek to educatewomen in order to increasetheir investmentin the human capitalof their children,but somedevelopment projects are testingthese linkages(Martin, Flanagan, and Klenicki1986).

Social Returns to Schoolingand Externalities of Education The socialcosts of schoolingare traditionallycalculated by addingpublic expenditures on educationto the privateopportunity costs of the timeof students. Hence,the socialreturns to educationare lowerthan the privatereturns, often by 20 to 30 percent. This adjustmentshould not cause the returns to educating womenand men to differ,unless the publicresources devoted to teachingwomen and men differ. In many parts of the world, highereducation for womenis focusedon specificprofessions, such as teachingand nursing,which may be less costlyto providethan other curricula.For example,women in somecountries receivemuch of their highereducation in "normalschools," which cost less per studentper yearthan regular academicinstitutions (see, for example,Birdsall and Fox 1985). Thispattern is alsonoted in India (Tilak 1987). Given higher costs for educating men, the social retums from their schoolingshould be commensuratelylower. Much better data are neededon these differentialcosts, however. Returns to Education 63

If a government taxes labor earnings, it reduces private returns to human capital and recovers some of its expenditureson education, health, and other public services. In most low-incomecountries, however,only wage and s,lary workers are effectively taxed, and no country has designed a tax on nonmarket production. The tendency for more men than women to work for a wage and be taxed might appear to favor social investments in the schoolingof males over females. This supposition,however, depends on how the labor supply of men and women responds to the market wage offered to them (after taxes). Most empirical studies of labor supply in high- and low-incomecountries indicate that women respond positively to an increase in their own market wages and negatively to an increase in their husbands' wages. Consequently, the taxable supply of women's labor rises with an investment in women's capabilities and decreases with a comparable investmentin men's capabilities. Moreover, estimates of the male labor supply often reveal a tendency for men to reduce their labor supplyas their own wages rise.J Thus, to the extent that governments recoup public expenditureson investments in human capital by taxing wages, the social return from investing in women should be greater than that from investing in men.W A more widely accepted rationale for public expenditureson education for both men and women is that an educated population enjoys a higher level of welfare. Moreover,this gain is not entirely captured by more educated individualsor their families. That human capital may be a source of increasing returns helps to explain the puzzle of modern economic growth (T.W. Schultz 1988;Kuznets 1966).

Supporters of public education may also argue that a more educated societyis more capable of managing a political system that protects individuals'rights and facilitates efficient and equitable growth. Although these claims are difficult to substantiate, more concrete examples have been empiricallyconfirmed. As noted earlier, most favor the schooling of females.

4i Taxes on personal income are a relativelysmall portion of governmentrevenues in low-incomecountries. The exception may be Latin American if one includes "socialsecurity' taxes (World Bank 1988a,p. 84) In the case of Thailand in 1981,direct taxes on personal incomes amounted to less than 5 percent of household expenditures, and only 10 percent of expenditurefor wagerecipients in Bangkok (National StatisticalOffice, undated). Most of government revenues in low-income countries are from trade and excise taxes on companies and Gommodities. Thus, adjusting for taxes may be of secondary importance in many low- income countries, but it could be quite important in industrially advanced countries and moderately important in Latin America.

V/ The marginal tax revenu'., dR, generated by the expansion of education for women, Ew, and men, Em, can be expressed as follows:

dR/dE, = tw&i (10w + 12m)

dR/dEm = twm(tlm (61m + 2w

Where ctlm and alw are the private rates of return to schooling of men and women, which are assumed equal; Olw> O is the wife's uncompensated own wage effect; P2w < 0 is the wife's uncoLipensated husband's-wage effect;, p m S 0 is the husband's uncompensated own-wage effect; and P2;ma 0 is the husband's uncompensatedlwife's-wageeffect. The above indicated signs are those commonlyobtained in static instrumented family labor supply models (Cogan 1980;Schultz 1981).

§/ A parallel public finance argument can be made for taxing more heavily inelasticallysupplied factors, if the goal is not to distort the optimal (that is, untaxed) allocation of factors. This tax criterion implies that, given the labor supply parameters outlined in footnote 5, male labor outcomes should be taxed more heavily than female labor outcomes because male labor supply is inel-stically supplied. 64 Returns to Education Conclusionsand Directions for Research and Policy

Investments that increase the primary and secondary schooling of women are economicallywarranted on several grounds, based on the currently available evidence from many countries. The public and private direct and opportunity costs of schoolingappear to be recovered fullyby the increased market productivity or potential wage gains experiencedby better educated women during their adult lifetimes. When internal rates of return are calculated based on private costs of attending school and potential wage gains, the returns tend to be at least as high for women as they are for men, varying from 30 to 10 percent in the less and more developed countries, respectively(Psacharopoulos 1973 and 1985). Social rntes of return calculated to recoup public expenditures on education are fractionally lower than private returns at the primary school level. The social returns to higher education are in some instances markedly lower than private returns, because of the relatively high costs of university training in low- income countries, and the reluctance of governments to cover these costswith higher tuition. In some cases, the public sector even pays for the opportunity costs of students' time by providing scholarships and cost-of-livingstipends regardless of the financial need of the student (World Bank 1986). Although public expenditures on education are rarely disaggregatedby sex, the pablic costs of education are likely to be lower for temales than males. If so, the social returns are correspondingly higher for females than males, at least at the university level, when other things are equal. Evidence is accumulatingthat schooling increases the nonmarket productivityof women. For example, a mother's education can improve her children'shealth, measured by their birthweight,nutritional status, and the absence of mortality. Research surveyedearlier describes how maternal education produces these child health benefits. Studies that estimate the interactionsbetween maternal education and program and policy interventions confirm that in some circumstances public health and family planning programs provide services that appear to substitute for maternal education (Schultz 1984). The largest payoff in terms of nonmarket returns to maternal education are realized where public health and family planning programs are least developed. Differentials in child mortality by mother's education are smallest, therefore, in urban areas served by relatively well-funded public health programs, and they are largest in rural areas that generally lie outside of the reach of hospitals and public health and family-planningclinics. The education of a woman is also associatedwith the educational achievementof her children, her fertility, and her regional and occupational mobility,controlling for her other household resources and productive opportunities. Though less widelystudied than the linkagebetween female education and child health, these parallel relationshipsalso confirm the importance of female education to nonmarket productivity. For *he most part, the woman and her family captut: hese returns, which thus compensate the intergenerational family for the costs (if her education. The nonmarket returns to women's education may also represent a social externality, howrever,and therefore warrant special public subsidyinsofar as society at large benefits from reduced fertility, dirmiished population growth, improved health and education of children, and increased population mobility.

Investments in the schooling of women encourage a shift in the allocationof women's time towa-rdmarket work and away from home-based work. Market-based work is counted in conventional national income accounts-and hence adds to GNP--whereas home-based work is not. But this is no reason in itself to encourage female education. This shift in the allocation of women's time may have other desirable consequences for the productiveuse of social resources. For example,because better educated women are more likelyto work in wege employment than are less educated women, they tend to pay more direct and indirect taxes, and given the government's requirements for resources, lowering the tax rates on others. Increases in men's education that contribute to increasing their wages tend to reduce the number of hours their wives work in the market labor force, and frequently, the men themselves work fewer hours as well. Thus, social rates of return on women's education are, higher than those on men's education, because the women's education returns more in taxes to support public schooling. Returns to Education 65 EAucationand Fertility

Social scientists and statisticians studying child mortality and fertility have used very different methods to estimate these relationships. Unless the common core of major exogenous economic variables is held constant, findingson the partial effects of women's education on child mortality and fertilitywill be unstable and potentially misleading. Any focused literature survey in these fields requires a clearly articulated research methodology.

Although parents universallyprefer a reduction of child mortality,there are times when couples do not want to limit their fertility. The effect of a variable, such as mother's education, on fertility involvesboth its effect on desired fertility (if birth control were costless and perfectly reliable) and its effect on unwanted fertility (by reducing the cost of using birth control with less uncertainty). Therefore, surveys of the relationship between education and fertilitythat do not control for the economic resources available to a woman from her husband and from nonearned family income and for the relative benefits and costs of children, do not alwaysfind a monotonic inverse relationship between female education and fertility. Use of the economic household demand model of fertility should make assessmentsof this important aspect of fertility variation across low-incomecountries more uniform.

Regional and Occupational Stratification

Returns to education are sometimes calculatedwithin subpopulations. If these subpopulations are defined by characteristics that are fixed for individuals,such as sex, race, or caste, interpretation of the returns withineach strata is straigh.forward. if the subpopulationsare not closed, however,as in the case of regions or occupations, empirical estimates of within-group returns to education are ambiguous and potentially misleading.

More educated individuals,male or female, are more likelythan the less educated to migrate from one region to another, holding constant differential wage gains earned by migration (Schwartz 1976). For example,estimation of the returns to schoolingwithin a decliningsector, where wages are generally lower than elsewhere, is likelyto introduce a downwardbias to estimates of the returns to schooling. This occurs because part of the return to schooling in the declining sector involves migrating out of that sector. Consequently,rural-urban stratified estimates of educational returns are potentiallymisleading unless the strata can be defmed by birthplace or residential area where the individualattended school. If migration rates differ for men and women, this bias in estimating the retums to education within a sector may distort the estimated returns to schooling for women relative to those for nen.

For the same reasons, assessing the returns to schoolingamong the self-employedis difficult. In the rural sector, the self-employedtend to be landholdersor tenants, and in the urban areas, small businessmen. The probability of being self-employed increases generally over the life cycle in low-income countries, and therefore wage functions estimated across ages, but within such an occupational group, are not readily interpreted in the human capital life-cycleframework (Mincer 1974). The assessment of labor market returns to schooling may be more accurate if analysis is restricted to wage earners, whose wage rates approximate hourly productivity. Such an analysismust then be corrected, however, for the systematicbias introducedbecause only wage earners remain in the sample. A joint analysisis required of the individual's occupational choice or allocation of time and the determinants of market wage productivity. This is again the now-familiarproblem of correcting for sample-selection bias, which has emerged repeatedly in this paper as a major limitation to existing research on returns to schooling of women.

To evaluate the resource intensity of the educational process, more detailed information is needed on the amount of t-me students and teachers expend on it. For students, the resource input is the private opportunity cost of the student's time allocated to school attendance and school homework. For teachers, it is the teacher's input or time, adjusted perhaps for the number of students he or she teaches in a class and for the amount of time required for preparation of class material. These school inputs may differ for female and male students, though they are ',oth enrolled in the same level of a school system. More accurate measures of the private and social costs of a year of school enrollment for males and females in 66 Returns to Education distinctivebranches of the educationalsystem might help explainsubsequent differences in their adult productivityand mightmodify resulting estimates for the privateand socialreturns to schoolingfor women and men.

IncreasingFemale School Enrollment The situationof womenvaries widely within and across countries. The challengeis to set prioritiesthat have the highestprobability of increasingfemale enrollmentrates in each specificsociety. Becausethe directand indirecteffects of manyof theseprograms are uncertainin suchvaried circumstances, collecting baselineand longitudinaldata on ,iopulationsserved by oilot and test programs and new policiesis particularlyimportant. Such data would permit future aluation of program consequenceson the educationalattainment of women;on the productivityand behavior of womenand men; and on the welfare of their families,communities, and countries. Parentsappear to be respondingto the growingevidence in the developingworld that more educatedsons and daughtersget distinctlybetter jobs. Generaly,they will earn more than enoughto justifythe sacrifices parents make to send them throughschool. Microeconomistshave amassedenough supporting evidence on wagedifferentials to providea humancapital explanation for the growthin schoolenrollments in most low-incomecountries. Quantitativeassessment of how better educated mothers create healthier childrenwho have a greater chanceto surviveto adulthoodis more difficult. Appraisalof the effect of maternal educationon child nutrition,cognitive development, and schoolachievement is stillmore subtle. The majorityof parentsmay not yet fullyappreciate these complexand extendedlifetime relationships, indeed, social scientists do not yet fuDlyunderstand them either. Evenif the current cluesin the literatureare confirmedby new, more thoroughanalyses and surveys,parents may stillrequire some time to digest and act on this new body of evidence. Howwil this information,as it diffusesthrough society, tend to affectbehavior? Youngmen, encouraged by their parents,should become more willingto "pay'more, or sacrificetheir intereststo a greaterdegree, to marry a more educatedwoman. This willoccur even when a wifeis not expectedto work in the labor force. As this increaseddemand for the nonmarketproductivity of educatedwives becomes apparent, parentsmay also perceivethe monetaryand psychologicalgains they willreceive by investingmore heavily in the schoolingof their daughters. The historicalrecord has not yet been adequatelyanalyzed, but the nonmarketproductivity gains associated withwomen's education may be a relativelyrecent phenomenonin some parts of the world. In the past, when effectivemodern health inputs,such as antibiotics,vaccines, and oral rehydrationsalts, were not readilyavailable to lo,.Y-incomepopulations, the abilityof a more-educatedmother to shelterher childfrom healthrisk mayhave been muchmore limitedthan it is today. Moderninputs to homeproduction activities may be especiallyuseful to better educatedwomen. The better educatedwoman today can control her reproductionwith greater ease and certaintythan was possiblein the past, becauseof the wideningrange of birth controltechniques. Nonetheless, until more women participate in the paid laborforce, parents may persistin their beliefthat the educationof their daughtersis of secondaryimportance to that of their sons. The res.-ing allocationof resourcesis viewedtoday as uneconomic,but existingpolicy research has not addressedthe question of how governmentsmight compensate for this traditionalbias. What policies shouldbe encouragedon the basis of the accumulatingevidence that the market returns to schoolingof womenare at least as high as those of men, and that nonmarketproductive returns to women'seducation are associatedwith major social externalities--in child health, nutrition, schooling, and fertility--whichmany societieswant to subsidize?Higher quality research in a widervariety of settingscould bolster the case for policyinterventions. Development agencies should initiate pilot programsand experimentalpolicy tests now,to begin identifyingpromising, cost-effective policy options. Evenbefore the pilotprograms can be fullyevaluated, however, governments can be involvedin the initiationof researchthat documentsthe local returnsto women'seducation, both in labor marketwage differences and in nonmarketproductivity gains. Returns to Education 67 References

Anderson, KH. "The Sensitivityof Wage Elasticities to Selection Bias and the Assumption of Normality," Joumal of Human Resources, 17 (Fall): 594-605,1982. Arrow, KJ. "Higher Education as a Filter,"Joumal of Public Economics, 2:193-316,1973. Barrera, A. "The Role of Matemal Schooling and Its Interaction with Public Health Programs in Child Health Production."Economic Growth Center DiscussionPaper, 551. Yale University(forthcoming in Journal of DevelopmentEconomics), 1988a. ------. "Maternal Schooling and Child Health." Ph.D. dissertation, Yale University,1988b. Becker, G.S. Human Capital, New York: Columbia University Press, 1964. ------. "A Theory of the Allocation of Ti ne," Fconomic Journalu74 (December): 493-517,1965. ------.A Treatiseon the Family. Cambridge, MA: Harvard University Press, 1981. Behm, H. "La Moitalidad en los PrimerosAr,os de V/idaen los Paises de America Latina, San Jose, Costa Rica: CELADE, 1976. "Socioeconomic------. Determinants of Mortality in Latin America."In SocioeconomicDeterminants and Consequencesof Mortality,Geneva: WHO, 1980. Behrman, J.R., and N. Birdsall. "The Quality of Schooling,"American Economic Review, 73 (December): 928-46, 1983. Behrman, J.R., and A. Deolalikar. "Health and Nutrition."In Handbook on Economic Development,vol. 1, chapter 15, H. Chenery and T.N. Srinivasan(eds.). Amsterdam: North Holland PublishingCompany, 1988a. ------. "Are There Differential Returns to Schooling by Gender?" Development Studies Project H Research Paper. Jakarta, Indonesia, 1988b. Benavot, A. "Education, Gender and Economic Development: A Cross-National Study," Sociology of Education, 62 (January): 14-32,1989. Ben-Porath, Y. "SelfEmployment and Wage Earners in Israel." In Studies in the Population of Israel vol. 30, U.O. Schmeiz (ed.), Jerusalem: Hebrew University,Magres Press, 1986. Berry, A. "Open Unemployment as a Social Froblem in Urban Colombia,"Economic Developmentand Cultural Change,23:276-91, 1975 Berry, A., and R.H. Sabot. "Unemployment and Economic Development,"Economic Developmentand CulturalChange, 33:199-116,1984. Birdsall, N., and M.L. Fox. "Why Males Earn More," Economic Development and Cultural Change, 33 (April): 533-56, 1985. Blau, D. "A Model of Child Nutrition, Fertility and Women's Time Allocation." In Researchin Population Economics, vol. 5, T.P. Schultz and KI. Wolpin (eds.). Greenwich, CT: JAI Press, 1984.

Blaug, M. Manpoweras a Forecasting'Technique,' Not an 'Approach'to Planning 1973.????? 68 Returns to Education

Blaug, M. "The Empirical Status of Human Capital Theory A Slightly Jaundiced Survey,"Joumal of Economic Literature,14: 827-55,September 1976. Blaug, M. "V,here Are We Now in the Economics of Education,"Economics of Education Review, 4:117- 28, `85.

Bongaarts, J., and R.G. Potter. Fertility,Biology and Behavior.New York: Academic Press, 1983. Bustillo, I. Chapter 5, this volume. Butz, W., H.P. Habicht, and J. DaVanzo."Environmental Factors in the Relationship between Breastfeeding and Infant Mortality,"American Joumal of Epidemiology, 119: 516-25, 1984. Caldwell,J.C. "Education as a Factor in Mortality Declihw,"Population Studies, 33: 395-413,1979. Catsiapis, G. and -. Robinson. "Sample Solution Bias with Multiple Selection Rules," Joumal of Econometrica, 18: 351-68, 1982. Chiswick,C.U. "OnEstimating Earnings Functions for LDCs,"Joumal of DevelopmentEconomics, 4: 67-78, 1976. ------.The Determinants of Earnings in Thailand." Research Project 671-36. Washington,D.C.: World Bank, 1979. Choe, M.K. "Sex Differentialsin Infant and Child Mortality in Korea,"Social Biology,34 (Spring-Summer): 12-25, 1987.

Cleland, J., and G. Rodriguez. "The Effect of Parental Education on Marital Fertility in Developing Countries."Fertility Determinants Research Notes, 19. New York: Population Council, November 1987. "The------.. Effect of Parental Education on Marital Fertility in Developing Countries," Population Studies,42:419-22, 1988. Cleland, ., and C. Scott. The WorldFertility Survey:An Assessment. Oxford: Oxford UniversityPress, 1987. Cochrane, S.H. Ferility and Education. Baltimore:Johns Hopkins UniversityPress, 1979.

------. "The Effects of Education, Health and Social Security on Fertility in Developing Countries." Working Paper WPS 93. The World Bank, Population and Human Resources Department, 1988. Cogan, J. "Married Women's Labor Supply: A Comparison of Alternative Estimates." In Female Labor Supply, J.P. Smith (ed.), Princeton, NJ.: Princeton UniversityPress, 1980a. deTray, D. "Government Policy, Household Behavior and the Distribution of Schooling."In Research in PopulationEconomics, vol. 5, T.P. Schultz, (ed.). Greenwich, Conn: JAI Press, 1988. Easterlin, RA. "An Economic Framework for Fertility Analysis,"Studies in Family Planning 6 (March): 54-63, 1975. Eelens, F., and L Donne. "The Proximate Determinants of Fertility in SubSaharan Africa." IPD Working Paper, 1985-3,Brussels: Vrije Universiteik,1985. Estrey, SA., and J.P. Habicht. "MaternalLiteracy Modifies the Fffect of Toilets and Piped Water on Infant Survivalin Malaysia."Processed. Ithaca, NY: Cornell Umversity, 1987. Returns to Education 69

Evenson, R.E. "The Allocation of Women's Time: An International Comparison,"Behavioral Science Research, 17 (Spring-Summer): 196-218,1983.

----. "EconomicIssues in Agricultural ExtensionPolicy." Processed. Yale University,Economic Growth Center, 1988.

Farah, AA., and S.H. Preston. "ChildMortality Differentials in Sudan,"Population and DevelopmentReview, June 8:365-83,1982.

Fields, G., and T.P. Schultz. "Income Generating Functions in a Low-IncomeCountry. Colombia,"Review of Income and Wealth, 28:71-87,1982

Fuchs, V.R. "SelfEmployment and Labor Force Participation of Older Males."NBER Working Paper, 584, 1980.

Gertler, P., and P. Glewwe. "Demand for Education in Peru." Processed. World Bank, 1988. Gregory P. "An Asessment of Changes in Employment Conditionsin Less Developed Countries,"Economic Developmnentand CulturalChange, 28: 673-700,1980. Griffin, C.G. "Methods for Estimating the Value of Time with an Applicationto the Philippines."Processed. University of Oregon, November 1987.

Griliches, Z. "Estimatingthe Returns to Schooling,"Econometrica, 45:13, 1-22, January 1977. ------. "SiblingModels and Data in Economics,"Joumal of PoliticalEconomy, 87 (October): S37-S64, 1979.

Gronau, R. "The Effect of Children on the Housewife's Value of Time." In Economics of the Family, T.W. Schultz (ed.). Chicago: University of Chicago Press, 1974.

Haines, M.R., R.C. and Avery. "Differentials in Infant and Child Mortality in Costa Rica, 1968-73," Population Studies, 36:31-44,1982. Hanushek, EA., and J.M. Quigley. "Life-cycleLeanings Capacity and the OJT Investment Model," Intemational Economic Review, 26:365-85,1985.

Haveman, R., and B. Wolfe. "Education and Economic Well Being: The Role of Nonmarket Effects," Joumal of Human Resources, 19:377-407,1984.

Heckman, JJ. "Shadow Prices, Market Wages and Labor Supply,"Econometrica, 42 (July): 679-94, 1974. "Sample------. Bias as a SpecificationError," Econometrica,47 (January): 153-62, 1979.

------. "Selection Bias and Self Selection." In The New Palgrave:A Dictionaryof Economics. London: Macmillan Press, 1987. Heckman, JJ.. and G. Sedlacek. "HeterogeneityAggregation and Market Wage Functions,"Joumal of PoliticalEconomy, 93 (December): 1077-125,1985. Hines, F., L. Tweeten and M. Redfern. "Social and Private Rates of Return to Investment in Schoolingby Race-Sex Groups and Regions,"Joumal of Human Resources, 5 (Summer): 318-40, 1970.

Hyde, K.A.L. "Improving Women's Education In SubSaharan Africa: A Review of the Literature." Processed. World Bank, Population and Human Resources Department, February 1989. 70 Returns to Education

Khandker, S.R. "Returs to Schoolingand Male-Female Wage Differences in Peru." Processed. PHRWD, The World Bank. February 1989.

King, E. M., and R. Bellew. "Education Policy and SchoolingLevels in Peru: Patterns and Explanations." Processed. The World Bank, May 1988.

King, E.M., J.R. Peterson, S.M. Adioetomo, LJ. Domingo, and S.H. Syed. Changein the Status of Women Across Geneations in Asia, Santa Monica, Cal.: The RAND Corporation, December 1986.

Kuznets, S. Modem Economic Growth:Rate Structure and Spread,New Haven, Conn.: Yale UniversityPress, 1966. Leibowitz,A. "Home Investments in Children." In Economics of the Family, T.W. Schultz (ed.). Chicago: University of Chicago Press, 1974. Lindsay, C.M. "Measuring Human Capital Returns," Joumal of Political Economy, 79 (November/December): 1195-2151971.

Lucas, R.E. "On the Mechanics of Economic Development," Marshall Lecture, Cambridge University. London: Basil Blackwell,1985. Martin, L.G., D.R. Flanagan, and A.R. Klenicki. "Evaluation of the Bangladesh Female Secondary Education Scholarship Program and Related Female Education and Employment Initiatives." Prepared for Office of Population, USAID, (85-63-032),Washington, D.C., March 1986. Meegama, SA. "The Decline in Mortality in Sri Lanka in Historical Perspectives."In Proceedingsof the IUSSPConference, Manila, 1981.

Mensch, B, H. Lentzner, and S. Preston. Socioeconomic Differentialsin Child Mortalitiyin Developing Co*unies,New York: United Nations, 1986. Michael, R.T. "MeasuringNonmonetary Benefits of Education."In FinancingEducation, W.W. McMahon and T.G. Geske (eds.). Urbana, IH: Universityof Illinois Press, 1982.

Mincer, J. "On-the-jobTraining, Costs, Returns, and Some Implications."Journal of PoliticalEconomy, 70, 5 (part 2 supplement): 50-79, October 1962.

Schooling Expenienceand Earnings.New York: Columbia UniversityPress, 1974

Mincer, J., and S.W. Polachek. "FamilyInvestments in Human Capital: Earnings of Women."In Economics of the Family: AMamage,Children and Human Capital T.W. Schultz (ed.), New York: Columbia University Press, pp. 397-429,1974. Mohan, R. Work, Wagesand Welfarein a DevelopingMetropolis. New York: Oxford University Press, 1986 Nakamura, A., and M. Nakamura. "Selection Bias: More Than a Female Phenomena." Processed. Edmonton, Canada, University of Alberta, 1988. Oliveira, M.C., and ES. Berquo. "The Familyin Brazil: DemographicAnalysis and Recent Trends." Seminar on Theories of Family Change, IUSSP, Tokyo, Japan, November 29-December 2, 1988.

Palloni, A. 'Mortality in Latin America: EmergingPatterns," Population and DevelopmentReview, 7:623-50, 1981.

Pitt, M., and M.R. Rosenzweig.'Estimating the Intra-familyIncidence of Health: Child Illness and Gender Inequality in Indonesian Households."Processed. Minneapolis,MN: Universityof Minnesota, 1988. Returns to Education 71

Pitt, M.M., M.R. Rosenzweig, and N. Hassan. "Productivity,Health and Inequality in the Intrahousehold Distnbution of Food in Low-IncomeCountries". Processed. Universityof Minnesota, Frborary 1989. "Estimating------. the Intra-familyIncidence of Health: Child Illness and Gender Inequalityin Indonesian Households."Processed. Universityof Minnesota, 1988.

Polachek, S.W. "OccupationalSegregation Among Women."In Women in the LaborMarke4 C.B. Lloyd,E.S. Andrews, and C.L. Gilroy, New York: Columbia University Press, 1979. Preston, S., and J. Trussell. "Estimatingthe Covariatesof Childhood Mortality from Retrospective Reports of Mothers," Health Policy and Education, 2, 1982. Psacharopoulos, G. Returns to Educadon. San Francisco:Jossey Bass, Inc., 1973. ------. Returns to Education: A Further International Update and Implications,"Joumal of Human Resources, 20, (Fall): 583-604,1985.

Psacharopoulos, G., and M. Woodhall. Educadon for Development. New York: Oxford University Press, 1985.

Ram, R., and T.W. Schultz, "Life Span, Health, Savings and Productivity,"Economic Developmentand Cltural Change, 27 (April): 399-421,1979.

Reyes, A. Detenminantesde la Ofertade Trabajoen America Latna, Programa de Estudios Conjuntos sobre Integraci6n Economica Latinoamericana, Rio de Janeiro, Brasil, 1982. Rodriguez, G., and J. Cleland. "SocioeconomicDifferentials in Marital Fertility in Less Developed Countries." Processed. Department of Statistics, Universidad Catolica de Chile, Santiago, January 1987. Romer, P. "DynamicCompetitive Equilibria with Externalities,Increasing Retuns and Unbounded Growth." Ph.D. dissertation, Universityof Chicago, 1983.

Rosen, S. "Hedonic Functions and Implicit Markets,"Journal of PoliticalEconomy, 82:34-55,1974.

Rosen, S. "Human Capital A Surveyof Empirical Research." In Researchin LaborEconomics, vol. 1, R.G. Ehrenberg (ed.), Greenwich, Conn: JAI Press, 1977. Rosenzweig,M.R., and R.E. Evenson."Fertility, Schooling and Economic Contribution of Children in Rural India," Econometico 45 (July): 1065-79,1977. Rosenzweig, M.R., and T.P. Schultz. "Market Opportunities, Genetic Endowments and the Intrafamily Distribution of Resources,"Amencan Economic Review, September 72:803-15,1982. "Child------.Mortality and Fertility in Colombia: Individualand Community Effects,"Health Policyand Education, 2:305-48,Amsterdam: Elsevier ScientificPublishing Co., 1982b. ------. "The Demand and Supplyof Births,"American Economic ReWew,75 (December): 992-1015,1985. ------. Fertility and Investment in Human Capital: Estimates of the Consequences of Imperfect Fertility Control in Malaysia,"Journal of Econometwir, 36:163-84,North Holland, 1987. ------"The Stabilityof Household Production Technology A Replication,"JournalofHuman Resources, 23:4 (Fall) 535 49, 1989. 72 Returns to Education

Sahn, D.E., and H. Alderman. "The Effect of Human Capital on Wages and the Determinants of Labor Supply in a Developing Country,"Joumal of DevelopmentEconomics, 1988. Schultz, T.P. Returns to Education in Bogota, Colombia, Santa Monica, Cal: The RAND Corporation, September 1968. "A------. Preliminary Survey of Economic Analysis of Fertility,American Economic Review, 63 (May): 71-78, 1973.

------."Interpretation of the Relations among Mortality, Economics of the Household, and the Health Environment." In `ocioeconomic Determinants and Consequencesof MortalityDifferences. Geneva: WHO, 1980a.

------."Estimating Labor SupplyFunctions for Married Women."In FemcleLabor Supply,J.P. Smith (ed.). Princeton, NJ: Princeton University Press, 1980b. Economics---.----. of Population. Reading, Mass: Addison-Wesley,1981. "Studying------.the Impact of Household Economics and Community Variables on Child Mortality."In Child SurvivalW.H. Mosley and L.C. Chen (eds.). Populationand DevelopmentReview, Supplement to vol. 10, 1984. School------Expenditures and Enrollments, 1960-1980."In National Academy of Sciencesbackground papers, Population Growth and Economic Development,D.G. Johnson and R. Lee (eds.). Madison, WI: University of Wisconsin Press, 1987.

------. Women's Changing Participation in the Labor Force: A World Perspective."Paper presented at the PAA meeting, New Orleans, April, 1988 (revised June). Yale University, 1988a. .An-"------. Individualistic Approach to Family Labor Supply and Fertility." Processed. Yale University, (revised April 1989), 1988b.

------."Educational Investment and Returns." In Handbook of DevelopmentEconomics, vol. 1, H. Chenery and T.N. Srinivasan (eds.). Amsterdam: North Holland, 1988c. ------. "-Womenand Development: Objectives, Framework and Policy Intervention." Processed. Yale University, February 1989. Schultz,T.P., and D. Thomas."Expenditure Patterns and BargainingModels of FamilyBehavior." Processed. Yale University (work in progress), 1989. Schultz, T. W. "On Investing in SpecializedHuman Capital to Attain Increasing Returns." In 7he State of DevelopmentEconomics, G. Ranis and T.P. Schultz (eds.). London: Basil Blackwell,1988.

Schwartz,A. "Migration,Age and Education,"Joumal of PoliticalEconomy, 84, pt. 1 (August):701-20, 1976. Strauss, J. "The Impact of Improved Nutrition on Labor Productivityand Human Resource Development." In PoliticalEconomyand Food Consumptionand NutritionImprovements, P. Pinstrap-Andersen (ed.). Baltimore: Johns Hopkins UniversityPress, 1989. Sussangkarn,C. "ProductionStructures, Labour Markets and Human Capital Investments:Issues of Balance for Thailand."Bangkok, Thailand Development Research Institute, January 1988. Teilhet-Waldorf,S., and W.H. Waldorf. "Earnings of Self-Employedin the Informal Sector: A Case Study of Bangkok,"Economk Developmentand CulturalChange 31:587-607,1983. Returns to Education 73

Thias, H.H., and M. Carnoy. "Cost Benefit Analysis of Education: A Case Study of Kenya." Economics Department Report, EC-173, The World Bank, 1969. Thomas, D. "Intra Household Resource Allocation:An InformationalApproach," Economic Growth Center conference paper, Yale University,April 1989. Thomas, D., J. Strauss, and M.H. Henriques. "Child Survival, Nutritional Status and Household Characteristics,"Economic Growth Center discussion paper, 542, Yale University,1987. Tilak, J.B.G. "Female Schoolingin East Asia: A Review of Growth, Problems and Possible Determinants." Processed. World Bank, Population and Human Resources Department, February 1989. The...... --. Economics of Inequalityin Education. New Delhi, India: Sage Publications, 1987. Tinbergen J. "On the Theory of Income Distribution,"Welwirtscha-fliches Archives, 77 155-73,1956. Turnham D. ThieEmployment Problem in Less Developed Countries.Paris: OECD, Development Centre, 1971. United Nations, Department of International Economics and Social Affairs. "Fertility Behavior in the Context of Development,"Population Studies, 100,New York: United Nations, 1987. van der Gaag, J., and W. Vijverberg. Wage Determninantsin Cote d'Ivoire, LSMS Discussion Paper, 33. Washington, D.C.: World Bank, April 1987. Weisbrod, BA. Extemal Benefits of Public Education, Princeton University,Industrial Rclations Section, 1964.

Welch, F. "Measurement of the Quality of Schooling,"American Economic Review,56 (May): 379-92,1966. Willis,RJ., and S. Rosen. "Educationand Self-Selection,"Joumalof PoliticalEconomy,87, pt. 2 (October), 1979. Wolfe, B., and J. Behrman. "Determinants of Child Mortality, Health and Nutrition in a Developing Country,"Joumal of DevelopmentEconomics, vol. 11, pp. 163-93,1982. Woodhall, M. "Investment in Women: A Reappraisal of the Concept of Human Capital," Intemational Review of Education, (Spring) pp.9-28, 1973. World Bank. FinancingEducation in DevelopingCountries: An Explorationof Options 2. Processed. The World Bank, Education and Training Division, 1986. World Bank. World DevelopmentRepot, 1988&Washington, D.C.: World Bank, 1988. 74 Returnsto Education Appendix to Chapter 2. Issues in Estimating SchoolingReturns

In the first studiesof the associationbetween education and labor earnings,private rates of return were calculatedunder the assumptionthat better educatedpeople benefit from schoolingonly to the extentthat they hold a job in the labor force (and report earnings). Thus, if a year of additionalschooling raised the wagerates of men and womenby x percentper year and theyboth workedfull-time in the labor force,the life-cycleinternal rate of return to the opportunitycost of attendingschool for that year was x percentfor men and women. If, however,women worked only half-timein the labor force after completingtheir schooling,while men workedfull time, the conclusionwas that womenearned onlyhalf the rate of return of men on their year of schooling. Becker (1964),in his seminalcontribution on this subject,observed that the lower return to women's education attributableto their lower partiripationin the labor force was consistentwith the smaller proportionof womenthan men attendingcollege in the UnitedStates in the 1950s(p. 178). Laterempirical studies and surveysthat specificallyaddressed gender differencesin rates of return to educationused Becker'smethodology (Hines, Tweeten, and Redfern197C; Thais and Carnoy1969; Psacharopoulos 1973; Woodhali1973). The implicitassumption underlying this researchwas that schoolinghad no effecton the productivityof people workingoutside of the marketlabor force. A large number of subsequentempirical studies have challengedthis assumption. They indicatedthat educationincreases the productivityof time in nonmarketproduction, particularly in the case of women (Michael1982; Haveman and Wolfe1984). Moreover, the opportunitycost of the time of femalesin school was not symmetricallydiscounted; girls were implicitly assumed to be givingup a full-timejob in the labor force to attend school,a patternthat mightnot havebeen true. Theseworking assumptions for estimating the rates of return exert a downwardbias on the rates of return to groups,such as women,that participate in the labor forceless than the averageperson does or that work more often as an unpaidfamily worker. Becker (1964) and Mincer (1974) also analyzedthe differencesacross individualsin annual earnings, combiing the effectof schoolingon the worker'spotential productivity per hour withits potentialeffect on marketlabor supplyand unemployment.The empiricalconsequence of this decisionis not theoretically obvious,but any empiricaltendency for unemploymentrates to be lowerand labor supplyto be larger among the better educatedwould bias estimatesof the rate of return to schoolingupward, when the logarithmof annual or monthlyearnings is used as the dependentvariable rather than the (preferable) logarithmof the hourlywage rate. The usualassumption in labor economicsis that peopleenter the marketlabor forcewhen the marketwage they are offered(after taxes and fixedcosts of enteringthe labor force) exceedsthe reservationwage that is determinedby their marginalproductivity in nonmarketor home activities.When an individualshifts time fromnonmarket to marketwork, the nationalincome may increase, but the broaderwelfare indicator of "fullincome" (Becker 1965), or the potentialproductivity of "'P humanagent, remainsunchanged. As people varythe hours they work in the marketlabor force,the > rvable marketwage rate continuesto approximatetheir nonmarketproductivity. Only when they leave the labor force entirelycan we infer their nonmarketproductivity exceeds their market wage offer. The problem is how to estimate the productivityof nonmarkettime for nonparticipantsin the labor force,by education,so as to incorporatethis informationinto the calculationof the returns to education. Estimatingthe nonmarketproductivity of nonparticipantsrequires that we specifythe mathematicalfunction for nonmarketproductivity function (for example,linear or noninear) and showhow its parametersare identified. There is no consensuson how this shouldbe done, and it is not attemptedhere. The originalmethodology of Beckerand Mincerfor estimatingthe returns to schoolingcan be improved, for example,by assumingthat educationaffects hourly labor productivityin market and nonmarketwork by the same amount,or that it is neutralbetween these sectors. The coefficienton years of schoolingin a logarithmichourly wage functionis then an estimate of the private of rate of return to schooling. Research must now appraise the severity of the bias remaining in this revised simple estimation Returns to Education 75 methodologybecause of the underestimationof the nonmarketproductivity of thosenot participatingin the labor force. Fimngthe effect of schoolingon nonmarketproductivity to zero appears,with hindsight,to have been unjustifiedand to have led to a large downwardbias in the estimated rates of return to educationfor women. The bias in the estimatedrates of return attributableto analyzingannual earnings, rather than the hourlywage rate, cannotbe generallyprescribed, for it dependson the jointdetermination of labor supply behaviorand wagefunctions in a specificsociety. Similarly,research is onlybeginning to evaluatethe bias on estimatesof the return to schoolingof womenattributable to the selectedsample of wageearners. This bias could overstateor understatethe true rate of return to educationfor the entire population (Heckman1979 and 1987). Theoretically,the variableneeded to identifythe nonmarketproductivity (or reservationwage) function is a householdfixed productive factor that affects the individual'snonmarket productivity but that does not alter his or her labor productivityto a firm in the market. In the short run, childrenhave been viewedas such a variablein that they raise a woman'sproductivity only in the home (Gronau 1974). Over the life cycle,however, this variableis alsojointly determined and responsiveto labor marketwage rates. Thus, it should be viewedas endogenousor as determinedwithin the same framework(Schultz 1981). Land ownershipand familybusiness assets, and the market productivityof a person's spouse may be more satisfactorymeasures of fixed iousehold endowmentsthat enhancethe value of an individual'stime in nonmarketactivities. The choiceof thisidentifying restriction determines how the estimatesof the market wagefunction are interpretedand hencewhether the impliedrate of return to schoolingis a satisfactory estimatefor the entire populationor onlyfor the nonrandomsample of wageearners. If more than one selectionprocess is used to definethe samplefor estimation,and the selectionprocesses havedifferent determinants, multiple sample-selection equations and correctionsare used (seefor example, Catsiapisand Robinson1982). For example,participation in the marketlabor force and acceptanceof wage employmentmay be responsesto differenthome and marketconstraints. If the marginalproduct of labor is measuredwith less error for wage and salaryworkers than for self-employedworkers, this schemeof double sample selectionmay be appropriate,to reduce measurementerror, despite the loss in the final samplesize. labor SupplyWad Unemploment Individualswith differentlevels of educationmay chooseto work differentnumbers of hours. The rates of return to educationwiD then differdepending on whetherthey are basedon comparisonsof hourlywage rates or an annual rate of earnings(Schultz 1968). Adjusting for howpeople allocate their time to market work in constructingthe benefitstream fromfemale education can be importantin low-incomecountries. Accordingto Mincer's(1974) equilibrium investment framework, the presentvalue of the sum of human and physicalcapital is not affectedby investmentsin schnoling. In this case, the total wealtheffect of schoolingshould be unimportant(see Lindsay1971). The voluntarylabor supplyresponse to the increasedwage rate offeredto more educatedworkers can be decomposedinto (1) an incomeeffect and (2) an income-compensatedprice (wage)effect. Mincer's frameworkimplies that educationis associatedwith a relativelysmall income effect. Mincerassumes all individualscan borrow at the same interest rate, but this assumptionmay be less tenable in countries withoutwell-developed loan markets. Whereliquidity constraints explain why low-income groups invest less in their children,the more educatedshould tend to workfewer hours becauseeducation will be associated witha wealthadvantage and thusbe reflectedin a demandfor more leisuretime. Thistendency would lead to underestimatingthe privatereturn to schoolingif the comparisonswere framed in terms of monthlyor annualearnings (see the analysisof Thailandaccompanying table 3.5). The more educatedwould receive part of their return from schoolingin the formof increasednonmarket activities, including leisure (see, for example,Mohan 1986,for Colombia).Conversely, the more educatedmight tend to worklonger hours in societiesin whichfamily wealth is more equallydistributed and studentloans and feliowshipshelp the poor 76 Returnsto Education investin humancapital. The income-compensatedprice (wage)effect associated with a worker'seducation wouldencourage the more educatedto worklonger hours, perhaps because of the debt they haveassumed to have an education. If this were the onlyeffect of educationon labor supply,comparisons of annual earningswould overstate the privatereturns to schooling. The partial associationof educationand hoursof marketwork tends to be positivefor youthand married womenwhen other sourcesof income,such as family'sor husband's,are held constant. When returnsto schoolingare based on variationsin annual earnings,the privatereturns may be overstatedbecause the offsettingloss of nonmarketproduction and leisure of the more educatedis not deductedfrom the gains in marketearnings. The changein wage rates (measured,for example,by annualearnings divided by hours) attributableto educationis thusa better approximationof the privatewelfare benefits from schooling than are changesin weekly,monthly, or annual earnings,particularly for women. The preferred dependent variablein the earningsfunction us%4 to estimateprivate returns to schoolingis the logarithmof the hguzly wagerate. deflatedby localprices.= Unemploymentmay be a productiveperiod during which workers search for job opportunitiesthat match their skifs. If unemploymentis greater amongmore educatedyouth during a relativelyshort periodafter they completetheir schooling,the opportunitycost of their job searchshould be includedalong withthe other costs of schoolingthat eventuallywill be recoupedby enhancedearnings in employment(Blaug 1973; Turnham 1971;Berry 1975;Gregory 1980;Berry and Sabot, 1984). Unemploymentis generallylower amongthe more educatedthan amongthe less educateda decadeafter theyenter the labor market. If this patterndoes not reflecta current choiceof the workerbetween nonmarket activities and marketwork, then unemploymentmay be called involuntary.Thus, one of the private gains from increasedschooling is enhancedaccess to regular work opportunitiesin the market labor force and hence a lowerincidence of involuntaryunemployment. Typically,analysts only address patterns of unemploymentby levelof educationamong men; no studieswere found on the differentialeffects of unemploymenton returns to schoolingfor women(see Tilak 1987,for some evidence).

Oupational Choice Returns to educationare ofteL calculatedfor subpopulations,such as occupations. Possiblythe most importantoccupational distinction is betweenwage and salaryearners and self-employedworkers. Most researchon returnsto schoolingfocuses on employeesbecause their labor earningscan be observedmore directly,self-employed respondents must be askedto deductpurchased and imputedvalues of inputsfrom gross income. Yet, relativelyfew studiesanalyze how selectioninto the employeesample could bias the estimatedreturns to education(see, for example,Anderson 1982, Griffin 1987).

1/ To introducemeasures of labor supplyamong the explanatoryvariables in the wagefunction is clearly inappropriateunless they are treated as endogenousvariables that could also respond to education. Another seriousproblem in estimatingschooling returns is to explainthe variationin annual or monthly earningsby the variationin the numberof hours or weeksindividuals work. / It is stilltempting to decomposethe effectsof such exogenoustraits as education,race, or gender on earningsand to appraisewhat portion of the effectoccurs because of occupationalsorting and whatportion occurswithin occupations (Polachek 1979). Becausethe stochasticprocesses determining occupation and earnings are undoubtedlyaffected jointly by unobservedfactors, this form of decompositionof a simultaneousequation system is feasibleonly when identifyingrestrictions are knowna prioin;that is, a factor known to influenceoccupational choice or placement,but not earnings,can be used to explain occupationalsorting but justifiablycan be omittedfrom the structuralwage equation. Studiesnonetheless assume,without justification, that occupationalchoice and earningsare stochasticallyindependent and can thereforebe modeledas block-recursive. Returnsto Edtscation 77 The probablecovariance between an individual'schoice of whetherto be a wageearner or a self-employed worker and that person's potential productivitycan be ignored to simplifythe problem. Separate estimationsare then repoited withoutcorrecting for sample selectionin each wage function for the self-employedand the wageearners. The coefficientson a worker'syears of schoolingare then compared withinthe two strata. The proportionateupward shifts in wagerates or earningsassociated with schooling tend to be of a similarmagnitude in Thailand(Chiswick 1979), Colombia (Fields and Schultz1982), and Israel (Ben-Porath1986). The increasein transitoryincome variations in the earningsof the self-em'ployed is oftenemphasized in the economicsliterature, but its relevanceto the returnsrealized from education has not been explored. These comparisonsof self-employedand wageearners have two weaknesses.First, they assumethat the self-employedare willingand able to report their labor earningsnet of Lhevalue of purchasedand owned inputssuch as rental valueof ownedland and businesscapital. In fact, in developedcountries, such as the UnitedStates, farmers and unincorporated entrepreneurs report incomesto surveysand taxauthorities that are much lower than the incomethe nationalaccounts impute to them. How this understatementof self-employedincome would bias comparisonsamong education groups i' m3nclear.To reduce thispotential source of reporting bias, Teilhet and Waldorf(1983) followeda smali sample of self-employedin the informalsector of Bangkokto derivetheir own estimatesof the net hourlyreturn to the labor of the self-employed.The returns to schoolingappeared to be no less for them than for wageearners, although obtainingmore educationpredisposed men in the Bangkoksample to obtain a job for wageemployment. The secondweakness is that the fractionof the labor forcethat is self-employedoften increaseswith age groups in a cross-section.The life-cycleprocess of accumulatingskills, experience, contacts, and physical capitalappears to increasethe likelihoodthat an individualwill become self-employed (Fields and Schultz 1982,Ben-Porath 1986). In approachingretirement, self-employment may also afford a worker more flexiblework opportunities than does wageemployment (Fuchs 1980). A seculartendency exists, however, for the share of self-employedworkers in the labor force to declinewith economicdevelopment (Kuznets 1966).Little empirical evidence is availableto help disentanglethe effectsof life-cyclechanges from those attributableto secularchanges during development.

Nomnarket Returns to Schoolingand Sample-SelectionBias Nonmarketreturns to educationcannot be valuedin comparablemonetary terms and readilyaggregated. Therefore,evidence of the returns to women'sschooling tends to rely on analysesof differencesin productivityamong wage earners who have differentamounts of education. The central problem in constructingthe statisticalcomparisons from whichto estimate the returns to schoolingof womenis to correctfor any potentialbias that mightbe introducedinto the analysisbecause the womenwho workfor wagesmay be more (or less) productivethan the averageperson. Statisticalprocedures designed for dealing with such a problem of sample-selectionbias have been developedin economicsin the last decade (Heckman1979 and 1987). They req iire informationon a variablefor the entire populationthat is statisticallycorrelated with the probabilityof beingin the wage earner sample,but the variablemust not be related to variationin marketproductivity (wages) that is left unexplainedafter controllingfor education,post-schooling experience, and so on. Thissample-selection methodology permits estimation of the effectof educationon marketproductivity for the averageperson so that an unbiasedestimate can be obtainedof the privateand social returns to schooling.If a particularperson does not work in a wagejob, it is becausethat individualcan work in a more productivejob at home or as a self-employedworker, or because the extra costs of findingand holdinga wagejob exceedthe financialgain. To correctan analysisof wage functionsfor this potential sample-selection bias requiresa variablethat influencesonly a person'snet nonmarketreservation wage but not her or his market wage offer. Sach a variable,it was argued earlier, could be tne family's nonearnedincome, land, or other assetsthat raise the persons'sproductivity in self-employmentand also increasethe individual'sdemand for leisure,assuming leisure is a "normal"economic good. 78 Retuns to Education

Sample-selection corrections should be performed routinely in the estimation of wage functions that are designed to estimate the average private returns to s hooling for women and for men. Although this practice is spreading, the methodologygenerally has not been identified by a common set of economically justified variables. The correction procedure may be particularlyimportant in low-incomecountries, which tend to have a higher proportion of familyworkers and self-employedworkers, and for women, because the proportion who earn wages can be relatively small.

Correcting for this sample-se;ectionbias by satisfactoryidentification restr stions sometimes substantially change the estimated private rates of retuns to schooling. At this time, however, we do not know why these differences occur in certain countries and not others, or why they vary by level of schooling. Improving the empirical research methods, however, should raise our confidence in the basic facts and improve our forecasts of educational investmentpriorities, by level and gender, over the course of economic developmentin a particular country. These improvedempirical estimates of the returns to schooling should also help explain long-run, simultaneousshifts in the aggregate supply of and demand for educated labor that occur during the process of modern economic growth. Sub-Saharan Africa 79 Chapter 3. Sub-Saharan Africa

KarinA.L Hyde*

Some of the world's poorest countries lie within Sub-SaharanAfrica and low literacy levels beset much of the region. This vast, diverse group of nations has been shaped by a mix of influences, among them indigenouscultures, Christianity aiid Islam, and a network of Western-type schools set up by missionaries and metropolitan governments during the colonial era. The region embraces Mauritius where more than 80 percent of the adult population can read and Somalia with less than 12 percent literacy. Wealth also varies greatly among Sub-Saharancountries, with per capita GNPs ranging from $10 to $2,700. The region encompasses Nigeria, where enrohment of girls in school in the Muslim north is lower than for the rest of the country, and Sudan where the Muslim north has higher enrollment rates than the Christian/traditional south. Wide differences exist in higher education as weDl,with women in Chad and the Central African Republic making up less than 10 percent of tertiary school enrolments, and women in Lesotho comprising more than 60 percent of third-level students. Amid the cultural and economic diversity,and despite widespread poverty, Sub-SaharanAfrica has made spectacular progress in expanding education since the countries achieved independence. Gross primary school enrolment was only 36 percent in 1960, half the rates of Asia and Latin America at that time. Between 1960 and 1983, to meet the needs of independence and economic growth, the region quintupled student enrollment in schools at all levels to 63 milion, a higher growth rate than in any other developing region. However, fast-growingpopulations and adver ; economic conditions have caused enroDmentsto stagnate and education quality to decline in many countries of the region in the early 1980s. The chalenge amid fiscal constraints and continuing population growth is to diversifythe sources of education finance, maximize the efficiency and quality of the existing education system, and to expand the education infrastructure selectively. Central to these concerns is narrowing the gap between male and female education by removing the constraints to sending girls to school and keeping them there longer. This chapter examines four aspects of women's education in Sub-SaharanAfrica: enrollment, achievement, attainment, and wastage or dropout. Enrollment refers to the decision to enter girls in school; achievement to the level of academic performance of girls once they are in school; attainment to the length of time girls remain in schoo', and how far they progress; and wastage to the girls' dropping out of school before they have completed a cycle. The chapter reviewswhat has been written about the factors that aff^ct these four aspects of edutcation,assesses policies that have been imnlemented in various countries, and summarizes earlier reviewsof women's education in the Third World to develop a frameworkfor further research, policy formulation and action.

The Economic and Educational Setting Economic Conditions Most Sub-Saharan African countries can be characterized as low income (Table 3.1). The median GNP per capita among all Sub-Saharan countries is US$300,ranging from a low of $130 in Ethiopia to a high of $2,700in Gabon. Total fertility rates are high, with a median of 6.5 children and a range from 2.1 in Mauritius to 8.0 in Rwanda. Overall, the median annual rate of growth in the population of primary- and secondary-schoolage children increased only slightly, from 2.7 percent during the 1960-70periud to 2.8 percent in the 1970-80period, but these growth rates actually declined in nearly half of the countries. Literacy rates have improved during the last twenty-fiveyears, rising from a mean of 9 percent to 41.8 percent, but a great deal of variation across countries remains. In Somalia and Burkina Faso, orty 11.6 80 Sub-Saharan Africa

and 13.2 percent, respectively, of the population can be characterized as literate, while Mauritius' and Zambia's respective literacy rates are 82.8 and 75.7 percent. Table 3.1 Economic Growth, Population, and literacy

School Age Literatein Population Aduk Population GNP Per Total Feyity GrowthRaaes (roente Capita (1987) Rate (oercent) (Dercent) 1960 1985or Countiy (S) 1987 1960-70 1970.80 latestyear

Low-incomeeconomies Angola na 6.4S/ 2.3 2.7 na 41.0 Ethiopia 130 6.5 2.8 2.9 1 62A Chad IS0 5.9 2.3 1.9 6 25.3 Zaire 150 6.1 2.0 2.1 31 61.2 Malawi 160 7.6 2.8 2.9 na 41.2 Mozambique 4'0 6.3 25 4.6 8 38.0 Guinea-Bissau i70a/ 5.4 / 2.8 2.8 5 31A Tanzania 180 7.0 3.4 3.5 10 na Burkina Fato 190 6.5 2.2 2.4 2 13.2 Madagascaz 210 6A 2.8 2.7 na 67.5 Mali 210 7.0 2.7 3.0 2 16.8 Gambia 2302/ 6.4 25 2.9 6 25.1 Burundi 250 6.8 1.7 2.5 14 33.8 Zambia 250 6.8 3.0 3.6 29 75.7 Niger 260 7.0 3.0 2.7 1 13.9 Uganda 260 6.9 4.3 3.5 25 57.3 Somalia 290 6.8 4.7 1.4 2 11.6 Togo 290 6.5 3.2 2.7 10 40.7 Rwanda 300 8.0 35 3.9 16 46.6 Sierra Leone 300 6.5 1.8 1.3 7 29.3 Benin 310 6.5 2.9 2.9 5 25.9 Central African Rep. 330 5.8 1.7 3.2 7 40.2 KLnya 330 7.7 4.1 3.8 20 59.2 Sudan 330 6.4 2.2 32 13 na Lesotho 370 5.8 2.6 2.4 na 73.6 Nigefia 370 6.5 2.6 2.7 5 42.4 Ghana 390 6.4 2.6 25 27 53.2 Mauritania 440 6.5 2.7 2.2 5 na Liberia 450 6.5 2.9 2.3 9 35.0 Guinea na na 1A 1.7 7 28.3 Middle-incomeeconomies Senegal 520 6.5 2.6 2.5 6 28.1 Zibrbabwe 580 5.9 4.0 3.8 39 74.0 Swaziland 600; / 6.5/ 3.6 35 na 67.9 Cote d'Ivoire 740 7.4 3.3 5.6 5 42.7 Congo, People's Rep. 870 6.5 2.6 4.0 16 62.9 Cameroon 970 6.5 2.5 3.0 9 56.2 Botswana 1,050 5.0 3.2 4.2 na 70.8 Mauritius 1,490 2.1 2.6 0.0 61 82.8 Upper middle-incomeeconomies Gahon 2,700 5.5 1.0 2.2 na 61.6 Sub-Saharan Afdea weighted mean 6.6 2.8 2.9 9 41.8 median 6.5 2.7 2.8 na - not available. 3 Averap annual growth rate of population of primary- and secondary-school-agechildren. 1 1985 data. Source: World Bank (498b, 1989). Sub-Saharan AfIr.a 81

Education Levels

One of the most end.ring kinds of educational inequality in all countries is between males and females. Although many countries have made tremendous progress in widening education's reach, in no country have males and females benefited equally. In the poorest and least developed countries, this inequalityis seen in lower enrollment rates, higher wastage rates, and lower levels of attainment for girls. Although many of the richer countries show no major gender difference in enrollments at the primary or secondary levels,females tend to have lowerrates of transition to tertiary education (Harding 1985). Inequality is also apparent in the differences in curriculum choices offered to or made by men and women* the secondary and tertiary levels, most notably in the low science and technology enrollment figures for )men. Tables 3.2 through 3.4 show the levels of primary, secondary and tertiary school enrollment in Sub-Saharan Africa. Of the thirty-ninecountries, twelveapparently had universalprimary school eniolUmentin 1985,with gross enrollment rates above 100 percent1, and females make up at least 46 percent of enrollments.2 Tanzania and Rwanda--countrieswith high but not universalprimary schnol enrollment--alsoappear to have been relatively successfulin extending educational access to girls.

1 Gross school enrolment rates can exceed 100 percent.

2 Regional variation exists within countries. In Kenya, for example, in some districts fema!es compose only a fourth of enrollments (Eshiwani 1982). 82 Sub-Saharan Africa Table 32 Gross Enrollment Rates in Primaiy Education

Females as a GrossP&imry EnrofimenAl Percentageof Total Enrolment Cow*nty 1960 1970 1980 1986 1960 1970 1986

Low-incomeeconomies Angola 17 75 124 134W/ 33 36 4&bJ Ethioa 7 16 35 34 24 31 40 ChadJ 17 35 36d/ 43 11 25 28 Zaire 54 88 94 76 27 37 45 Malawi 38 36 60 62 36 37 43 Mozambique S1 47 99 86 38 34 44 Guinea-Bissau 24 39 67 56 30 30 35 Tanzania 24 34 93 72 34 40 49 Burkina Faso 9 13 18 29 29 37 37 Madagascar S6 90 101 104d/ 44 46 ;8dj Mali 9 22 25 23 28 36 37 Gambia 14 24 51 68 31 31 38 Burundi 21 30 29 53 24 42 40 Zambia 51 90 90 99 40 45 47 Niger 6 14 27 28 30 35 36 Uganda 47 38 S0 70 32 40 45 Somalia 7 11 27 1S 25 24 33 Togo 44 71 122 95 28 31 38 Rwanda 49 68 63 61 31 44 49 Sierra Leone 20 34 52 Si/ 34 40 41 Benin 26 36 64 65 28 31 33 Central African Rep. 30 64 72 77 19 33 39 Kenya 47 58 115 98 32 41 49 Sudan 20 38 S0 50 27 38 41 Lesotho 92 87 103 113 62 60 55 Nigeria 42 37 97 77 37 37 na Ghana 46 64 80 72 35 43 43 Mauritania 6 14 37 49 19 28 39 Liberia 36 56 76 35 29 33 40 Guinea 20 33 31 30 26 32 31 Middle-incomeeconomies Senegal 27 41 46 57 32 39 40 Zimbabwe 74 74 85 135 4S 45 48 Swaziland S7 87 103 107 50 49 49 Cote d'lvoire 43 58 75 70 26 36 41 Congo, People's Rep. 81 130 160 16j3 34 44 49/b Cameroon S7 89 104 108 33 43 46 Botswana 39 65 91 108 59 53 52 Mauritius 94 94 108 105 47 49 50 Upoer-middle-incomeeconomies Gabon 54 85 115 18sw 38 48 49t Sub-Saharan Africa weighted mean 36 48 76 75 34 39 44 median 38 56 73 77 32 37 42

na = not available.

1/ Number enrolled as J percentage of the appropriate age group. b/ Figures are for 1983 £i Figurs are for 1984. / Estimates. Source: World Bank k19S8b). Sub-Saharan Africa 83

Table 3.3 Gross Enrollment Rates in Secondary Education

FemaIes as a Gross Secondaty Enrona Percentage of Toal Enrolment

Counay 1960 1970 1980 1985 1960 1970 1985

Low-incomeeconomies Angola 2 8 19 12J 40 42 33bW Ethiopia 1 4 9 13 14 25 41 Chad 0.4 2 4 6d/ 7 8 17 Zaire 2 9 34 23 24 22 30 Malawi 1 2 3 4 22 27 25 Mozambique 2 5 5 7 36 38 29 Guinea-Bissau 3 8 6 7 40 36 21 Tanzania 2 3 3 3 32 29 33 Burkina Faso 1 1 3 5 27 28 33 Madagascar 4 12 14§/ 15 33 40/dJ 4j Mali 1 5 9 7 17 22 31 Gambia 4 7 11 17 26 24 42 Burundi 1 2 3 4 37 20 38 Zambia 2 13 16 17 ?3 33 36 Niger 0.3 1 5 6 17 27 25 Uganda 3 6 5 13 21 25 36 Somalia 1 3 10 9 9 16 33 Togo 2 7 34 21 23 22 25 Rwanda 2 2 3 6 35 33 42 SierraLeone 2 8 14 15ai 27 28 28 Benin 2 5 16 17 27 30 29 CentralAfrican Rep. 1 4 14 16 15 19 25 Kenya 2 9 20 21 32 30 38 Sudan 3 7 16 20 14 28 43 Lesotho 3 7 17 19o/ 53 54 60 Nigerian/ 3 4 19 23 21 32 na Ghana 19 42 37 39 2i 38 40 Mauritania 1 2 11 15 5 11 26 liberia 2 10 23 21 16 23 29 Guinea 2 13 14 10 10 21 25 Middle-incomeeconomies Senegal 3 10 11 14 27 29 33 Zimbabwe 6 7 8 41 36 39 40 Swaziland 18 38 43 45 44 49 Cote d'lvoire 2 9 18 19 12 22 29 Congo, People's Rep. 4 20 83 87W 28 30 4lJ Cameroon 2 7 19 24 17 29 38 Botswana 1 7 21 30 48 46 52 Mauritius 22 30 48 51 32 40 48 tlooer middle-incomeeconomies Gabon 3 8 21 23/ 16 29 40b Sub-Saharan Africa weighted mean 3 7 16 20 25 31 34 median 2 7 14 16 26 29 33

Number enrolled as a percentage of the appropriate age group. W Figuresare for 1982. FIFigures are for 1984. Estimates. Source: World Bank (1988b). 84 Sub-Saharan Africa Table 3.4 Gross Enrollment Rates in Tertiary Education

Females as a Gross TeriaPy Enroumental Percentage of Total EnroUment

Counhy 1960 1970 1980 1983 1960 1970 1983

Low-income economies Angola na 05 0.3 0 .4 J na 40 16c/ Ethiop4a na 0.2 04 05 5 8 11 Chad?t na na 0.5b/ 1.2 na na na Zaire 0.1 0.7 1.2 1.2 na 6 9 Malawi na 0.3 0.4 O.4 na 23 28J/ Mozambique na 0.3 0.1 0.1 na 44 36 Guinea-Bissau na na na na na na na Tanzania na 0.2 0.3 0.4 na 17 17 Burkina Faso na na 0.3 0.6 na 15 22 Madagascar 0.2 l. 3.1 3 5 23 23 na Mali na 0.2 0.3 4 .9W na 11 12 Gambia na na na na na na na Burundi na 0.2 0.6 0.6 na 6 20 Zambia na 0.4 1.6 1.6 na 15 22J Niger na na 0.3 0.5 na na 22 Uganda 0.2 05 0.5 0.6 12 18 27£/ Somalia na OA 0.6 0.6 13 13 11 Togo na 05 2.2 1.7 na 12 15 Rwanda na 0.2 0.3 0.3 na 9 14 Sierra Leone 0.1 05 0.6 0O6./ 11 16 25t/ Benin na 0.1 1.3 2:0e' na 7 16cJ Central African Rep. na 0.1 0.9 1.2 na 13 10 Kenya 0.1 0.8 0.9 0.9 16 15 19 Sudan 0.4 1.2 1.8 2.1S/ 5 13 27W/ Lesotho 0.2 OA 1.8 2 .i/2 22 34 594/ NigeriaJ 0.2 0.5 2.2 2.1 7 15 na Ghana 0.2 0.8 1.5 1.8 11 14 22 Mauritania na na 0.4 0O4.W na na 1° Liberia 0.5 0.9 2.9 2.1 21 22 26 iuinea§/ na 0.6 4A 3:0c/ na 8 22Pn

Middle-income Economies Senegal 0.5 1.5 2.8 2.2 17 17 21 Zimbabvw_/ 0.1 1.2 1.3 2.6 25 42 42 Swaziland na 0.6 3.9 3.0 na 39 41 C6te a"voire 0.1 0.9 2.9 2.W. 11 14 18W Congo, People's Rep. OA 1.7 5.6 6.0W 7 5 i4J Cameroon na 0.5 1.5 1.6 na 8 14 Botswana na na 1.1 1.6 na na 44 Mauritius 0.1 2.6 1.1 0.6 na 5 30

UMier-middle-income economies Gabon na 0.2 2.2 3.3Y/ na is 2N/

Sub-Saharan Africa weighted mean 0.2 0.6 1.2 1.4 10 16 21 median 0.2 0.5 1.1 1.2 12 15 22 na - not available.

Number enrolled as a percentage of the appropriate age group. I Figurms are for 1982. £ Figures are for 1984. Estimates.

Source: World Bank ()QP-b). Sub-Saharan Africa 85

Levels of enrollment in secondary schools were far lower than in primary schools (table 3.3). Only in Zaire, Congo, and Mauritius were enrollment rates above 50 percent. The rates in most countries were below 20 percent, and Malawi, Tanzania, Burkina Faso and Burundi were notable for having maintained their enrollments below five percent since 1960. Also, quite clearly, the higher the level of school the lower the representation of girls. Only in Mauritius, Botswana, Lesotho, and Swaziland were female enrollments at or above male levels in secondary schools, and at the tertiary level (table 3.4), only Lesotho had parity between mrles and females. These satellites of South Africa showed a number of anomalies when compared with other countries in Sub-SaharanAfrica. The effects of South Africa on the frontline states are myriad, but two that affect school enrollment are: first, the economy of South Africa has high labor demands that are met partly by migrants, especiallyyoung males from neighboringstates; and second, the quality of education available in frontline states for black Africans is better than that accessibleto most black South Africans in South Africa so that a major proportion of the demand for black skilled labor is met through the recruitment of non-citizens. Consequently,all the neighboring states experience major flows of young male workers to South Africa and out of secondary and tertiary education. In addition, the use of boys as herders outside settled areas in their own countries increases the drain of boys out of the school system. Wastage is an umbrella term that has generally included grade repetition (that is, holding children back for one or more years, frequently for poor performance on the end-of-year or promotion examination) and dropout (that is, children leaving school and not re-enrolling in that er any other school before they have completed a cycle). The true extent of these two problems in Sub-SaharanAfrica has generally been difficult to quantify because recordkeeping at all levels of schooling is poor. However, repetition at the primary level shown in table 3.5, can be a problem of very large proportions. Dropout and repetition reflect ineffectiveteaching and weak demand for education, but they are also accepted strategies within the educational system for maintaining standards and for lowering the demand for school places at both the primary and secondarylevels (see Bali and others 1984;and Chishimba 1984). Expulsions for academic failure and failure to pay school fees are widely accepted as legitimate policies. Such expulsions are not distinguished in the data from voluntary exits. In some countries where intense competition for secondary school places exists (for example, Kenya and Nigeria), affirmative action takes the form of lower pass marks for pupils from less advantaged regions. As a result, large-scale, though largelyundocumented, transfers into the last year of primary school have taken place in these disadvantaged regions (Nkinyangi1980). Pupils have transferred to raise their prospects for secondary school places and have changed their names and exercised a number of other subterfuges to conceal their past school histories (Nkinyangi1982). In Kenya, for example, dropout rates in primary schools in the least developed areas were consistentlynegative in the period 1970-76(Nkinyangi 1980).

3 Although it can be argued that the figures at the higher levels reflected the histories of earlier cohorts, there was no evidence that these relative positions were changing. High repetition and dropout imply both ineffectivenessof schools to teach or retain students and weak motivation of students to attend school and learn. Parental attitudes (against exposing girls to "foreign"influences and expending resources on girls), levels of parental education, and demand for child labor at home are some factors that may affect motivation. 86 Sub-Saharan Africa Table 3.5 Percentage kepeaters by Grade at the Primary Level

Grades at PrnmarvLevel

County Year Total I II IF IV V VI Vn7 vIII T F T F T F T r T FT F T F TF T F

Low-incomeeconomies Ethiopia 1981 12 14 17 19 11 13 9 10 8 10 8 10 9 12 - - - Chad 1976 38 36 40 37 32 34 34 35 26 27 31 31 58 50 - - - - Zaire 1983 19 19 21 21 22 19 21 22 18 18 15 15 12 11 - - - - Malawi 1984 16 15 17 17 16 16 13 13 9 10 4 4 11 12 13 15 38 37 Mozambique 1985 24 24 24 24 24 25 22 23 23 25 Guinea-Bissau 1984 50 50 56 54 50 51 50 51 50 51 34 39 25 29 - - - - Tanzania 1982 1 1 3 3 3 3 3 2 0 0 0 0 0 0 0 0 Burkina Faso 1984 14 15 10 10 10 11 12 12 12 13 13 15 35 35 - - Mali 1983 30 30 28 39 28 28 32 33 31 31 31 33 32 31 - - Gambia 1985 17 16 18 17 11 11 11 11 11 10 13 12 37 34 - - - - Burundi 1984 15 16 12 12 13 13 10 10 16 17 22 24 27 30 - - Zambia 1981 1 1 0 0 0 0 0 0 2 2 1 1 1 1 6 6 Niger 1986 15 15 2 2 15 15 15 15 15 15 20 19 26 27 - - Uganda 1982 10 11 11 11 10 10 11 11 10 11 11 11 14 14 4 4 Togo 1984 37 38 38 38 34 35 37 39 30 34 35 38 46 48 - - - - Rwanda 1985 12 12 18 18 13 12 12 11 11 10 9 8 8 8 8 7 7 7 Benin 1985 27 28 21 22 23 23 27 28 26 27 30 32 42 44 - - - Central African Rep. 1982 35 37 35 37 32 33 35 39 31 36 33 34 49 47 - - - - Kenya 1981 13 13 15 14 13 12 11 11 12 13 13 13 15 16 13 12 - - Lsotho 1985 23 21 31 30 25 23 21 19 17 16 15 15 13 14 20 21 - - Ghana 1980 2 2 4 4 2 2 2 2 1 1 1 1 1 1 - - - - Mauritania 1980 14 17 12 15 1 14 12 16 13 16 10 13 27 28 - - - - Liberia 1978 11 12 12 13 a/aj 13 14 12 13 11 13 10 12 9 10 8 10 Guinea 1985 27 31 34 35 16 19 27 31 25 32 27 32 44 46 - - - - Middle-incomeeconomies Senegal 1984 16 16 13 13 12 12 11 12 12 12 15 16 35 36 - - Zimbabwe 1984 1 1 1 1 1 1 1 1 1 0 0 0 0 0 1 1 - Swaziland 1984 13 11 13 11 12 9 13 11 11 9 12 11 11 11 16 15 - - Cbte d'lvoire 1984 29 28 21 22 20 20 23 24 21 23 27 29 54 - - - Congo 1985 30 29 33 32 20 19 37 36 32 31 28 28 24 23 - - - Cameroon 1984 29 28 35 34 25 24 31 30 23 23 26 26 32 31 8 8 - Botswana 1985 6 6 1 1 1 1 1 1 12 11 1 1 1 1 23 24 - - Mauritius Ucoer-middle-incomeeconomies Gabon 1983 33 33 45 44 31 30 30 30 20 20 19 20 33 33 - - - -

Note: T - total; F - female. Prnmay levels vary in length. A/ Data combied with data for Year 1.

Source. UNESCO (1987).

The mean educational attainment of girls is low in Sub-Saharan Africa because enrollmeut is low and wastage high. In most countries, more girls than boys fail to enter school at all, and among those girls that do enroll, the repetition rates are lower than for boys, in some cases being almost 20 percentage points lower as in Cote d'lvoire and Malawi (table 3.6).4 A number of predictors and factors influence this gender difference and the length n. time girls wil remain in school. Subsequent sections will summarize these factors and discuss the ways in which they affect girls' enrolment and attainment

4 See Nkinyangi1980, for Kenya. Sub-Saharan Africa 87

gender difference and the length of time girls will remain in school. Subsequent sections will summarize these factors and discuss the ways in which they affect girls' enrollment and attainment

Table 3.6 Retention Rates at the Primary Level (circa 1970-80)

Country Girls Boys

Chad 22.2 36.9 Zaire 28.1 42A Malawi 26.9 45.9 Tanzania 49.3 65.6 Burkina Faso 55.4 58.0 Mali 42.7 48.8 Buntndi 12.8 13.0 Zambia 643 84.8 Niger 49.7 553 Togo 52.8 66.9 Rwanda 33.0 37.1 Benin 533 57.0 Central African Republic 28.6 44.9 Kenya 68.5 76.8 Lesotho 45.5 32.2 Mauritania 52.5 53.6 Senegal 63.1 723 Swaziland 64.4 593 C6te d'Ivoire 45.9 63.9 Congo 573 61.8 Cameroon 46.5 52.0 Botswana 793 76.0 Gabon 33.9 42.4 Malagasy Republic 20.1 20.7 Source: Robertson and Berger (1986,chapter 6).

Factors AffectingEducation Reach and Attainment

A Framework for Review of the Research The aim of this reviewand discussionis to developa basis for actionand intervention.The emphasis, therefore,is on those studiesthat have tried to identifythe factorsresponsible for the largelysecond-class status of women within the education systems of Sub-Saharan Africa. Many, if not most, of the issues touched on here are not uniquely"African" issues. Ultimately, the reasons why females are seldom as well represented in the education system as males lie completely outside the systems of education. The inequalitiesdiscussed in this chapter simplyshow how gender equality is expressedin the education systems of Africa, but discussionsof the education systemselsewhere would have similar themes. Africa may stand out because it is the poorest of the continents and has the lowest levels of education, but it is also the continent that has made the most progress with regard to female participation in education in the last 25- 88 Sub-SaharanAfrica

30 years (UNESCO 1983b)5. Ideas about the appropriate roles for women in the labor market or society, about the biological unsuitabilityof women for science, about the gender-based division of work in the household (and, in the African context, on the farm), will surface frequently in subsequent chapters of this book. Such ideas are familiar to any reader who has thought criticallyabout the situation of women in any society. Certainly, income, class, religion, and region are familiar explanatory variables for a variety of sociologicallyand economicallyinteresting behaviors in most areas of the world. The emphasis of the review of research is on formal, Western-derived educational systems, although use of organizations outside the formal education system is essential and expedient to increasing female educational participation. And, the natural comparison group for measures of performance, enrollment, attainment, and wastage is boys from the same environment as the girls being discussed. This comparison can often highlight specific schooling levels, regions, and socioeconomic groups in which females are particularly disadvantagedwith respect to education. But the comparison can be misleadingor inadequate in some situations. For example, in Lesotho, more females than males are enrolled at every level of education and in every age group; the situation is similar, though not as extreme, in Swaziland and Botswana. The four aspects of educational participation (enrollment, attainment, achievement,and wastage) all show the lower education status of women. Although these aspects are examined separately because they can be measured separately, they are linked statisticallyin at least two ways. First, the levels of attainment, performance and wastage are all conditional on the initial decisionof whether to enroli. Second,the factors that affect any one aspect of participation also affect the others, so levels of performance, attainment, and wastage are interdependent. For example, school performance can lead to a re-evaluation of plans for continuing or dropping out, although the relationship is not simple (Mann 1986, for the United States; Gambetta 1987, for Italy).

The literature identifies factors that have affected women's - 'ucation in Africa--their enrollment, attainment, and achievement. The studies do not provide data for ,large number of countries, however. In general, educational statistics from Africa are poor and often outdated. Seldom are they collected in a waythat is useful for this type of gender-specificanalysis. The studies suffer from a number of deficiencies: girls were seldom analyzed separately, and authors occasionally neglected to use satisfactory tests of statistical significanceand an appropriate sampling methodologyor adequate controls, which would have made conclusionsabout the strength and robustness of particular effects defensible. The studies reviewed here were selected on the basis of three criteria: 1) that they be empirical, that is, use collected data or try to analyze studies that had collected data; 2) that they be published after 1979, because a number of extensive reviews undertaken around this date covered earlier work; and 3) that they treat gender explicitly. All such studies identified and obtainable are included here; no attempt was made to sample from them. The chapter is divided into two major sections. The first dealing with enrollment, attainment and wastage, and the second with achievement and performance. Each section opens with an analysisof the nature of the phenomena, followedby a summary of the evidence in the literature and ends with a discussion.

Enrollment,Attaiment, and Wastage Decisions to continue in school or drop out are contingent first on the decision to enroll and then on a continuing reassessment of one's position in school, aided by feedback on achievement. This segmental nature of school participation has theoretical and methodologicalimplications. A clear understanding of the underlying structure of the process within which these decisions are made is crucial to correct assessment of the data on continuation rates, on the social background of students, and on labor market

S Obviously, floor effects are operating here, but this progress may suggest the absence of "African" intransigence to the idea of full educational participation for females. Sub-SaharanAfrica 89

activities. Actions that imply entry into, continuation through, or dropping out of school may be taken at several points or decision nodes. Surrounding these decision nodes are characteristics of the individual, home, community, school, or school system that can either support or hinder educational decisions.

At the point of entrance into school, the influencingfactors are competing activities-- household labor demands on children; competing ideologies--communitydisagreement over the mores and values taught or perceived as being taught within school; and an estimate of the expected benefits and costs of entering school. The relevant decisionmakers may differ for each of these factors. The family,individual parents, the community at large, and even the child will have opinions and face influencesthat will have differing weights. In addition, institutional factors, such as the supplyof places, may pose constraints on school entry.

Once a child is in school, characteristics of the school become more salient, influencingthe way parents or other responsible adults approach the decision to keep the child in school. At this stage, the child also becomes more instrumental in deciding whether to continue, and schoolsmake extensive use of suspension and expulsion. Which of these factors is more important in the rate of dropout? What role do such factors as gender bias play in retention or teacher expectations? Which are the public and private paths of achievementaccessible to girls? As a child progresses through school, the costs frequently become greater, and the balance of costs and perceived benefits changes. Is this factor important in the early dropout of girls? Do the curricular choices available to girls differ from those offered to boys? What are these differences,and what are their possible short- and long-term consequences? As the girl approaches puberty, social demands such as becoming engaged to be married or undergoing initiation rites become more important.

Decisions about school exit involvelabor market opportunities, which can be both push and pull factors. If the eventual gains from staying in school exceed the immediate gains from leaving,a girl is likely to stay in. The capacity of the school system is critical. Virtually all school systems in Sub-SaharanAfrica expect only a minority of students to enter the secondary level,and achievement is a major factor but not the only one. Pregnancyand marriage can precipitate exit; girls who become pregnant are usually asked to leave and may experience difficultyre-enrolling even in different institutions. Marriage can have similar consequences.

Apart from these specificfactors, which have an impact at different ages and levels, more general factors can affect decisionmakingabout a child's school life. They include religion, level of economic development, language policy, rural/urban residence, the general availabilityof schooLs,and the economic opportunities open to females.

Figure 3.1 provides an abstraction of the interwovenstructure of decisions as to whether or not a girl enters school and how far she goes once she has entered. The diagram makes plain the interconnectednessof the outcomes of these decisions, as well as their cumulative nature. It emphasizes the option of exit not only at the points of entrance or transition but also at the end of every year, every term, and every week, as family and school continuouslyadjust to changingconditions. A further refinement can be made by adding additional branches at the transition points to secondary or tertiary levels to indicate options in the types of institutions girls enter. 90 Sub-Saharan Africa

Figure 3.1 Decision Structure of Educational Participation

Figure V ': Decision Structure of Educational Participation

Not Enrolled Leave Leave

A Achevement Achievement Ahievement

Enrolled Continue Continue Time

Source: Modified from Gambetta (1987)

Most studies on girls' school enrotlment focus on the initial decision to enter primary school, but a few deal with factors that encourage the enrollment of girls in secondary or tertiary schools, or different types of secondary or higher education institutions (Weis 1981;ELuiwani 1982). The variables that appear to affect female participation have been arranged under three headings: * Fami influence-social class (Weis 1981;Biraimah 1987;A-sie 1983);parental attitudes toward sending girls to school (Csapo 1981); level of maternal education (Kossoudji and Mueller 1983; Chernichovsky 1985); child labor (Biazen and Junge 1988; Chernichovsky1985); and selective education of children (Chernichovsky1985; Orubuloye 1987)

* Sodctal fbctos-marriage and other competing activities(Amon-Nikoi 1978); urban residence (Akande 1987;Robertson 1986;Bedri and Burchinal 1985;Assie 1983);overall levels of enrolhment(Adams and Kruppenbad 1987); agricultural labor force (Wood and others 1986); and government expenditures on education (Wood and others 1986)

* ScbhoolaWn-school quality (Eshiwani 1982; Weis 1981); vocational training opportunities within schools (Robertson 1986;Eshiwani 1982;Csapo 1981); and the lack of role models (Tembo 1984) Sub-SaharanAfrica 91

Fwniy Infleum soci m Girls who come from socioeconomicallyadvantaged homes are much more likely to enter and remain in secondary school than are girls from disadvantagedfamilies. Assie (1983)showed that in the C6te d'lvoire a girl with a university-educated father was more than thirty-five times as likely to enter an academic secondary school as was the daughter of a man with no education. For a son, the comparable advantage was only a tenth as large. The large advantage experiencedby girls from homes with high socioeconomic status is reflected in their relative over-representation in elite lvee in Assie's sample: although only 25 6 percent of all CyO&students were female, 41.7 percent of those in elite Iyce were female. Weis (1981)found that female students at the secondary levelin Ghana were also disproportionatelydrawn from families with more education as compared to male students at the same level. The social class advantage was even more striking when contrasted with historical levels of selectivityamong boys. In other words, even when the Ghanaian educational systemwas much smaller, the sons of uneducated fathers were not unduly disadvantaged. Despite their relativelyhigher socioeconomicstatus, however, female students were over-represented in low-qualityschools; that is, the expanding number of female enrollees was finding its way primarily into the newer and less established secondary schools. This pattern is different from that seen in the C6te d'lvoire. This difference may be attributable to the smaller size of the Ivorian school system and the iower proportion of female students within it. Another possible explanation is that the growing enrollments in Ghana were accor2modatedmainly through an increase in the number of schools rather than an expansion of existingones. rhus,the opportunities for girls and other disadvantagedgroups to gain entrance were greater in the new institutions. The educational advantages conferred by high socioeconomic status were even more pronounced at the university level. Biraimah (1987) showed that women students at the tertiary level in Nigeria were disproportionately drawn from more priv owgedfamilies.

Parental Attitodes

The desire to protect daughters from foreign influences appears stronger in areas that are very traditional. For example, this factor explakaswhy northern Nigeria stands in sharp contrast to other areas in Nigeria in the percentage of girls enrolled in school. The reasons for the relative underdevelopment of education in Muslim northern Nigeria are often held to arise from the British government's laissez-faire attitude toward the region and the resistance of the emirs to the introduction of Christian missionaries and the spread of Western education (Denzer 1988;but see also Tibenderana 1988 for a contrasting view that the lack of schooling in the North was a result of British thrift rather than emir resistance). According to Callaway (1984), the education of girls was hampered because northern Nigeria was the only area in Sub-Saharan Africa that observed p_rdah and the only area in the world that observed it so strictly and among both rural and urban women. Western education was regarded as a threat to both Muslim and Hausa values, a threat seen as particularly dangerousfor women,whose duty it was to protect the traditions. Csapo (1981)reviewed a number of unpublished theses from Ahmadu Bello University and editorials and artides from newspapers published in northern Nigeria on the subject of female education. The consensus appeared to be that it was bad for soceqty,and for girls in particular, to be educated in Western schools. The review did not contain, however, a.y measure of the prevalence and strength of these views or of the extent to which they were shared by women and girls.

6 Elite Iyce are the academic secondary schools that prepare for entrance to university.

7 Purdah is a system of screeningwomen (especiallyMuslim and Hindu) from strangers. It often involves the wearing ot veils in public or the seclusion of women within their homes during the day. Girls enter Offrdakat marniage,generally around puberty. 92 Sub-SaharanAfrica As Robertson (1986) pointed out, however,Islam itself should not be held responsiblefor the low enrollmentof girlsin Afiica. Sudanstands as a counterexamplein whichthe MuslimNorth has significantly higher school enrolhnent fates than the Christian/traditionalSouth. Robertson claimed that, since independence,the predominantlyMuslim countries in Africa have had the highestenrollment growth rates.

Mother'sEducation Evidencefrom a number of countriesindicates that Africanwomen bear a large part of the burden for educatingtheir children(fripp 1988,for Tanzania;Robertson 1977, for Ghana). Tb. mother'sability to pay schoolfees and to encouragecontinued attendance at schoolis an importantfactor in explainingthe levelsof enrollmentand attendanceof children. In areas wherepolygamous marriage is common,many womenare the prime moverswith respect to their children'seducation (Bledsoe 1988), and their own levelsof educationand commandof resourcesare importantfactors in their abilityto keep their children in school. This is especiallytrue where male migrationis widespread,and women become de fact heads of households(Kossoudji and Mueller1983). Householdsheaded by educatedfemales are more likelyto send girls as well as boys to schooland to keep them there longer (Kossoudjia d Mueller 1983,and Chernichovsky1985, each for Botswana). Their abilityto supportthemselves and their childrenin part dependson their own levelsof educationi-i that educationis lisuallywhat allowswomen to enter formal- sectoremployment with its higherand more dependableincome stream. Also,because work in the formal sectoris relativelyincompatible with child care, women employed in that sectorwill generally be both more willingto have the schoolhelp care for childrenand more able to incur schoolexpenses.

Chld Labor Mothers' time appears to be particularlyconstrained, especially in rural areas where householdsmust commitextensive time to such activitiesas collectingwater or firewood(McSweeney and Freedman1980). The high fertilityrates in Sub-SaharanAfrica (seen in table 3.1) may be an adaptiveresponse to the demands of the highly time-intensivehousehold and farm tasks that make women's time a scarce commodity.One of the advantagesof polygamousmarriage is the sharingof householdtasks, and, perhaps as a consequence,women in polygamousmarriages have fewer children (Smith and Kunz 1976;Garenne and vander Walle 1988).8The majorityof Africanwomen are not in polygamousmarriages, however, and the natural sourceof additionallabor for the more time-intensive,low-skill jobs in the householdand on the farm is daughtersor other youngrelatives. The extensiveuse of childlabor has obviousimplications for the enrollmentof childrenin schools.Using 1974data from Botswana,Chernichovsky (1985) showed that the enrollmentof childrenseemed to be influencedby the presenceof elderlypeople in the home (who can be seen as substitutesfor childlabor) and the household's possessionof capital (for example, livestock)that may involvechildren in income-producingactivities. Childrenwere more likelyto be enrolledif their grandparentslived in the householdand to remain in schoolfor a shorter period if, at intermediatelevels of familywealth, 9 the householdhad more liv.stock. But boys' schoolingwas more affectedthan girls' by householdownership of livestock,because boys' herding activities were more likelyto be awayfrom home and schools.Similarly, in rural Ethiopia,labor demandsare more importantfor boysthan for girls. Resultsfrom a sampleof 576

8 The apparentcontradiction arises from the low time-intensityof childcare. Mothers'work is oftenfully compatiblewith childcare, and the use of "foster"care can relievemothers of the need to care for their babies.

9 Each additionalolder person increasesby 26 percentthe likelihoodthat a childwill be in school,while being in the middlecategory of cattle ownershipreduces the levelof educationof enrolledchildren by 0.5 years. Sub-Saharan Africa 91 Famil Iniewce

Girls who come from socioeconomicallyadvantaged homes are much more likely to enter and remain in secordary school than are girls from disadvantagedfamilies. Assie (1983) showed that in the Cote d'Ivoire a girl with a university-educatedfather was more than thirty-five times as likely to enter an academic secondary school as was the daughter of a man with no education. For a son, the comparable advantage was only a tenth as large. The large advantage experienced by girls from homes with high socioeconomic status is :eflected in their relative over-representation in elite Ivcee in Assie's sample: although only 25 percent of all Ive students were female, 41.7 percent of those in elite Ivce were female.6 Weis (1981)found that female s!udents at the secondarylevel in Ghana were also disproportionatelydrawn from families with more education as compared to male students at the same level. The social class advantage was even more striking when contrastedwith historical levelsof selectivityamong boys. In other words, even when the Ghanaian educational system was much smaller, tha sons of uneducated fathers were not unduly disadvantaged. Despite their relatively higher socioeconomicstatus, however, female students were over-represented in low-qualityschools; that is, the expandingnumber of female enrollees was fnding its way primarily into the newer and less established secondary schools. This pattern is different from that seen in the Cote d'Ivoire. This difference may be attributable to the smaler size of the Ivorian school system and the lower proportion of female students within it. Another possible explanation is that the growing enrollments in Ghana were accommodated mainly through an increase in the number of schools rather than an expansion of existingones. Thus, the opportunities for girls and other disadvantagedgroups to gain entrance were greater in the new institudions.

The educational advantages conferred by high socioeconomic status were even more pronounced at the university level. Biraimah (1987) showed that women students at the tertiary level in Nigeria were disproportionatelydrawn from more privileged families.

Parental Attitudes The desire to protoct daughters from foreign influencesappears stronger in areas that are very traditional. For example, this factor explains why northern Nigeria stands in sharp contrast to other areas in Nigeria in the percentage of girls enrolled in school. The reasons for the relative underdevelopment of education in Muslim northern Nigeria are often held to arise from the British government's laissez-faire attitude toward the region and the resistance of the emirs to the introduction of Christian missionaries and the spread of Western education (Denzer 1988;but see also Tibenderana 1988 for a contrasting view that the lack of schooling in the North was a result of British thrift rather than emir resistance). According to Callaway (1984), the education of girls was hampered because northern Nigeria was the only area in Sub-Saharan Africa that observed purdahand the only area in the world that observed it so strictly and among both rural and urban women. Western education was regarded as a threat to both Muslim and Hausa values, a threat seen as particularlydangerous for women,whose duty it was to protect the traditions. Csapo (1981) revieweda number of unpublished theses from Ahmadu Bello Universityand editorials and articles from newspapers published in northern Nigeria on the subject of female education. The consensus appeared to be that it was bad for society, and for girls in particular, to be educated in Western schools. The review did not contain, however, any measure of the prevalence and strength of these views or of the extent to which they were shared by women and girls.

6 Elite Ivcee are the academic secondary schools that prepare for entrance to university.

7 Purdah is a system of screeningwomen (especiallyMuslim and Hindu) from strangers. It often involves the wearing of veils in public or the sedusion of women within their homes during the day. Girls enter Mfrfg& at marriage, generally around puberty. Sub-Saharan Africa 93 households interviewedindicate that household duties represent the primary cause of school absenteeism among 57 percent of boys but only 32 percent of all girls (Biazen and Junge 1988). Girls are also needed for household labor. Women in sedusion in northern Nigeria depend on children to help in their market activities. Ironicallv,being in sedusion does not exempt these women from family obligations that require cash expenditures. Their husbands are unlikelyto assume responsibilityfor these outlays because these obligations are used to support the woman's position in her own lineage (Remy, as cited in Fapohunda 1978). Consequently,women in pur_d4ahengage in several kinds of production for the market, especially food preparation and handicrafts. Confined as they are to their households during daylight, they depend on children or young relatives to sell their wares in the street or marketplace. They also rely on their children to collect raw materials and help with production and sales. Sending children to school, thus, may curtail the income-generatingactivities of such women.

SelectiveEducation of Children Children may be kept out of school for reasons that include a need for their labor and a need for them to learn traditional skills. Chernichovsky(1985) found that schooling levels were higher in families with more school-age children. An examination of the education within each family. however, showed a large inequality among children. Those children who went to school attended for longer periods, but those not enrolled were more likely never to have been enrolled. This pattern suggests that parents select certain children for Western education. The reason Chernichovskygives is the diminishingreturns to labor in a household with any given amount of assets. Thus, while some children will be reserved for productive activityin the household or farm, other children are allowedto study. In Botswana,Swaziland, and Lesotho, this calculusappears to operate to the advantage of girls because they are chosen to fill their families' need to have educated children. In most other countries, however, boys are more likely to receive this educational advantage. The selection of specificchildren for education need not be gender-biased. Faced with limited resources, the education of children can represent a risk dispersion strategy, formal educatioaxcan reduce the likelihood that particular child has to earn a living from agriculture and increases the possible sources of income for the intergenerational family. However, the structure of the formal labor market generally gives boys an advantagein this calculus. Similarly,faced with high dropout rates, parents may decide to send only those children whom they expect wid both do well in school and gain remunerative modern sector employment after school. Sending childrento school means they lose valuabletime learning traditional skills (Bowmanand Anderson 1980;Robertson 1984). Even in an urban setting, the foregone learning can be so costly as to deter the enrollment of girls in school (Robertson 1984). Girls and women are overwhelminglyemployed in the traditional or informal sectors, and school attendance means that they have less time to learn with their mothers or apprentice with other older women. Schooling also entails monetary costs--fees, uniforms, books--whichmay be low relative to modern sector incomes but are high if a family's principal income-generatingactivities do not take place in the modern sector. At the same time, the earnings from modern-sector employment are much lower for girls than for boys. Hence, given the costs associated with schooling and the relatively lower monetary return to girls' education, the choice of whom to enroll in school is often made against girls and for boys. Studies of spending on girls and boys in Nigeria show that parents both expect to and actually do spend more to feed and educate their sons than their daughters (Orubuloye 1987). 94 Sub-Saharan Africa Societal Factors Nuptiality Marketwork, marriage, and initiationrites at pubertycan competewith school for oldergirls' time. Labor forceparticipation rates are highestfor womenin the 15-25age group (Amon-Nikoi1978). The relatively youngage of marriagein manycountries (table 3.7) makesmarriage an importantreason why girls do not enter secondaryor tertiaryinstitutions or, havingenrolled, leave before the cycleis completed.Although the enrollmentof marriedstudents is not unheardof, pregnancyand childbirth Table 3.7 Age by Which 50 percent of Women and Men Have Ever Been Married, by Rural/Urban Residence

Total Rural Uban Country Year Women Men Women Men Women Men

Angola 1970 18 23 na na na na Ethiopia 1970 na na 16 22 na na Chad 1964 17 22 17 22 17 22 Zaire 1955-58 18 23 na na na na Malawi 1977 17 22 17 22 18 24 Mozambique 1970 19 24 na na na na Tanzania 1967 17 23 17 23 17 25 Burkina Faso 1975 17 27 ;1 26 18 27 Mali 1976 17 27 17 27 19 29 Bumndi 1970-71 21 23 na na na na Zambia 1969 18 24 na na na na Uganda 1973 20 30 na na na na Togo 1970 18 25 18 25 20 27 Rwanda 1970 20 22 20 22 20 24 Benin 1961 2OI/ 24 20 24 18 25 Kenya 1969 19 25 na na na na Sudan 1973 18 25 18 25 19 27 Lesotho 1966 19 25 na na na na Ghana 1971 19 na na na na na Mauritania 1965 na na 21 28 na na Liberia 1974 18 26 18 26 18 26 Guinea 1954-55 20' 26 20 26 20PJ 27 Senegal 1972 18 28 17 28 19 29 Cameroon 1976 18 26 17 25 19 27 Botswana 1971 24 30 na na na na Mauritius 1972 22 27 na na na na Seyche;i-s 1977 30 32 na na na na

0/ Preciseage cannot be determined;fgure representsa maximumestimate. na = not available.

Source: Newman (1984, table 6.2). Sub-Saharan Africa 95

usuallyend a school career.10 For older female students,therefore, the competingactivity of family formationis an additionaldeterrent to schooling.Marriage can affectprimary school children in societies where the schoolshave significantnumbers of cver-agechildren or where betrothal takes place at very young ages. In Ethiopia,for example,20 percent of the primary schoolstudents surveyedin a study reported by Biazenand Junge (1988)were either promised,married, or divorced.Both bos andgirls were affected,but this was the most popularreason givenfor the nonenrollmentof girls. On the other hand, Wrzesinska(190) reported that the higher bride price paid for educateddaughters was an important considerationfor parents who sent their girls to secondaryschool in Zaire. The more recent data from Nigeriamay illustratethis effectalso at the tertiarylevel. Althoughinitiation rites at pubertyare becomingmore flexible,they can require extensiveperiods away from school,especially if parentswish their daughtersto marry or becomeengaged at this time.

Urban Residence A numberof studies(Akande 1987, Robertson 1986, Bedri and Burchinal1985, Chernichovsky 1985, Assie 1983)have suggestedthat area of residenceis predictiveof enrolment and attainmentat all levelsof education. The mechanismsthrough which urban residenceoperates are not alwaysclear, however.The availabilityof schools,the relativelyhigher opportunity cost of a rural girl'stime, the greaterwealth of urban families,different attitudes toward the educationof girlsin Westernschools, and the greateropportunities for girlsto obtainmodern-sector employment in urbanareas may all be relevantfactors. Evidence on which to judge the relativestrength of these competingand complementaryexplanations is generallylacking. A recent studyon Ethiopia(Abraha and others 1991)observed that urban girlsenrolled in schoolpersist at higherrates, both in absoluteterms and relaive to boys (table 3.8). The measureof grade 4 female persistence(that is the ratio of grade 4 to grade 1 enrollmentper school)is higherin urban schools(0.61) than in rural schools(0.42). Moreover,grade 4 enrollmentequity, measured as the ratio of femaleto male enrollmentsin grades 4-6, is significantlybetter in urban schools(0.84 versus0.49). Interestingly,the proportionof girls passingthe nationalexam appears lowerin urban areas, a result perhapsof the worse selectionproblems in rural areas. The impactof urbanresidence is misleading,however. Once schooland teacher variablesare taken into accountin a multivariableanalysis, the force of the communityvariables weakens (Abraha and others 1991). Chernichovsky(1985) found that when the effectof familyand other factors affectingschooling have been carefullyconsidered, the correlationbetween schooling attainment and size of the community vanishes.Residence in larger villagesaffects primarily enrollment probabilities, while other factorsexplain levelof educationand time spent in schooL

10 In 1982,80 percentof femalestudents at BayeroUniversity in Nigeriawere married(Callaway 1984). 96 Sub-SaharanAfrica

Table 3.8 Urban-rural diference in girls' enrollment and performances,Ethiopia 1986

Urban Ruran F-ratio

Grade 4 female persistence 0.61 0.42 5.88* (39) (135) Grade 4-6 enrollment equity 0.84 0.49 31.29*** (.41) (141) Percent of girls passing exam 0.81 0.88 5.37* (36) (119) Pass-rate equity ratio 0.89 1.04 1.11 (36) (119)

p < 0.5, ***p <.001,

Notes: (1) Numbers in parentheses are sample sizes. (2) Grade 4 female persistence is grade 4- grade 1 enrollment. Enrollment equity is the ratio of female to male enrollments in grades 4-6. Pass ratio of the proportion of girls to boys passing the national exam. Source:Abraha and others 1991

The followingthree factors--overalllevels of enrollment, percent of the labor force engaged in agriculture, and percent of GNP spent on education--relate to variables that appear to have an effect at the national level and are indicators of the level of development of a country and of its educational system.

OverallLezels of Enrollment Adams and Kruppenbad (1987) suggested in their comparisonof four countries (Botswana, Liberia, Niger, and Somalia) that the overall level of enrollment was one of the factors that explained female access to primary school. In other words, countries enroll boys first, and girls gain access at a delayed rate. An additional factor was the overall level of economic development,with richer countries having higher levels of female enrollment. With a larger sample of countries, however, Robertson (1986) found very little correlation between growth rates in girls' enrollment and levels of GNP or income per capita.

Agriculral LaborForce

Analyzingdata for all developingcountries in Africa and Asia, Wood, Swan,and Wood (1986)reported that t"e percentage of the labor force engaged in agriculture was the economic factor that most strongly predicted a nation's school dropout rate. They hypothesizedthat the size of the agricultural labor force was a measure of the demand for child labor. Particularly with traditional farming methods, the payoff to education was sufficientlylow that parents did not feel the need to send their children to school nor were opportunities for off-farm employment likely to be numerous. Although this argument was not specific to guls, it suggests that in areas where most farming is done by women, the labor demands on girls are greater than those on boys. The situation in Botswana illustrated the effects when boys' labor is in greaterdemand than girls'. Sub-Saharan Africa 97

Government Expendiureson Education Wood, Swan, and Wood (1986) also identified the level of government expenditures on education as an important predictor of the national dropout rate. The authors related this supply variable to a high valuation of education. Expenditures on education were probably also related to the number of school places and the probability that students were not being asked to leave school because of a shortage of places.

School Fators

Schoo Quali

Poor school quality inhibits the educational attainment of girls in several ways. Curriculum is a major factor: a continuingdisproportion existsin enrollment of girls in such subjectsas domestic science (Harding 1985). (See table 3.9 for the distribution of curriculum choices in tertiary institutions.) Apart from the curricular choices that girls may be encouraged to make, the overall quality of the schools they attend has an important effect on their attainment. Eshiwani (1982) and Weis (1981) showed that in Kenya and Ghana, respectively,girls were over-represented in secondary level institutions of low quality, that is, those institutions that had relatively poor success in preparing their students for "O"-level and "A"-level examinations."1 One of the most important disadvantagesthis conferred was a lack of accessto science and mathematics. Associated w"+>this lack of scientifictraining was the idea that physics and chemistry are unfeminine.'2 Eshiwani showed that the majorityof girls in secondary school in Kenya were in harambee schools;13 few of these offered any science,while those that did lacked the infrastructure to do so effectively. For Swaziland,Wheldon and Smith (1986)reported that girls in single-sexschools were less likelyto choose physicalsciences than were girls in mixed schools,despite the fact that the single-sexschools in their sample had selective admissions and were well equipped. The lack of access to science in secondary school was usually permanent; the European type of educational system still in use in African countries ensured that options not taken were usually options lost. For example, courses were more circumscribed with few mechanisms for such options as changing majors.

The poor quality of the institutions girls attended also put them at a disadvantage after leaving school, because they were less likely to have acquired permanent functioaal literacy and the skills that employers value in educated workers. This disadvantagewas distinctfrom any distaste or discriminationthat may have existed among employers. Some evidenceindicated that those women who did manage to complete their education successfullymay have suffered pronounced discrimination on the job. For example, female engineers were sometimes not aDlowedon construction sites or had great difficulty obtaining internships during their training (Adjebeng-Asem 1988, for Nigeria).

1 These exams are often prerequisites for entrance into further education or formal-sector employment.

12 Eshiwani (1982) reported that female students at the Universityof Nairobi commented several times, 'Mathematics and science make you ugly."

13 Schools created and rmaintainedthrough communityself-help. Table 39 Fmal Enrollment at Teriary E du Level, Overal and by Fied of Stdy (Latt Year)

Feid of SAP/

Cousav Total I H Hi IV v VI VII Vill IX X Xl7 xni2 xmI )UV xv XvI XVI XVI

Ethiopia 17.9 19.1 18.9 93 85 143 34A 17.7 375 0.0 75 12.6 143 33 9.2 53 0.0 83 0.0 Chad 8.6 1.7 9.8 0.0 10.0 IOA 0.0 0.0 0.0 0.0 5.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Malawi 275 33.6 V 0.0 6.0 135 13.6 0.0 0.0 0.0 12.3 0.0 71.1 0.0 0.0 0.0 0.0 15.8 9.3 Mozambique 23.0 33.0 36A 0.0 0.0 22.7 0.0 0.0 0.0 0.0 30.4 0.0 50.6 7.8 0.0 0.0 0.0 302 0.0 Tanzania 13.6 20.0 0.0 0.0 16.8 20.1 20.8 0.0 0.0 0.0 22A 0.0 11.9 3.9 0.0 0.0 0.0 0.0 0.0 Burkina Faso 22.2 0.0 265 22.4 32.2 13.9 50.0 0.0 0.0 0.0 16.9 45 19.7 0.0 0.0 0.0 0.0 7.8 0.0 Madascr 38.4 31.6 56.8 0.0 42.8 34.8 34.2 0.0 0.0 0.0 36.9 18.1 43.6 7.4 0.0 0.0 1.8 33.5 0.0 Mali 11.3 435 0.0 0.0 0.0 0.0 17.4 0.0 0.0 0.0 0.0 0.0 12.8 2.1 0.0 0.0 1SA 10.3 0.0 Burundi 26.9 25.8 29.8 0.0 35.6 343 37.6 425 0.0 0.0 19.6 7.0 21.0 1.6 11.8 0.0 0.0 6.1 29.1 Zambia 17.3 19.2 k 0.0 26.8 28.8 21.1 0.0 0.0 0.0 9.1 0.0 31.1 0.0 0.0 0.0 0.0 4.8 11.6 NierF 18A 20.5 28.3 0.0 20.7 17.2 0.0 0.0 0.0 0.0 24.7 5.7 18.4 0.0 0.0 0.0 0.0 63 0.0 Uganda 21.9 24.2 24.5 25.8 32.6 28.1 24.7 0.0 0.0 0.0 13.7 7.5 21.2 3.4 0.0 0.0 0.0 21.0 28.1 Togo 15.0 22.0 k k 19.9 11.40 V l 0.0 0.0 6.0 k 21.3 0.0 1.1 0.0 0.0 2.8 0.0 Rwanda 13.4 20.5 7.0 0.0 17.1 28.6 0.0 31.6 0.0 11.6 k 11.9 0.0 0.0 100.0 0.0 3.3 3.0 Benin 16.4 16.2 26.6 0.0 20.0 17.2 18.7 0.0 0.0 0.0 18.6 6.3 16.0 3.7 0.0 0.0 0.0 E.3 0.0 Cent. Afr. Rep. 9.9 13.4 V 0.0 8.7 7.6 21.8 0.0 0.0 0.0 3.6 0.0 11.5 0.0 0.0 0.0 0.0 0.0 0.0 Keya 26.2 37.0 V 52.6 39.5 285 383 47.5 89.2 0.0 13.9 27.5 24.7 1.7 17.6 27.5 0.0 19.8 0.0 Sudan 305 33.2 31.9 20.3 33.8 363 27.0 0.0 0.0 0.0 38.0 19.9 32.0 7.4 15.8 0.0 0.0 iJ. 97.8 Lesotho 63.2 72.1 S7A 0.0 38.7 56.9 52.0 0.0 0.0 0.0 29.8 5S.6 69.2 0.0 0.0 0.0 0.0 0.0 0.0 Ghana 17.1 25.7 l 30.7 k 21.6 14.4 29 90.4 0.0 1i1 V 20.9 1i 15.2 0.0 0.0 9.5 0.0 liberia 27.7 23.4 41.7 0.0 25A 18.8 33.4 0.0 00 OAu 1S.8 5.6 55.9 0.0 0.0 S.0 0.0 1'.7 34.3 Guinea 14A 11.7 l 0.0 Y V 31.0 0.0 A". 0.0 16.2 k 14.6 3.8 4.0 19.2 12.5 15.7 0.0 Sengal 20.8 I.C 22.1 0.0 17.5 k k 25.5 0.9 28.1 113 V 34.6 14.0 0.0 0.0 0.0 6.2 60.5 Swaziland 38.2 49.1 65.0 0.0 27.1 48.0 40.6 0.0 100.0 0.0 26.6 0.0 0.0 0.0 0.0 0.0 15.1 0.0 Cbte d'Ivoire na na 30.0 16.9 11.7 32.8 46.0 0.0 0.0 0.0 10.6 3.0 32.2 3.6 30.4 2.0 5.2 7.2 0.0 Congo 13.4 13.1 12.5 0.0 133 11A 38.7 0.0 0.0 0.0 6.1 V 20.9 0.0 0.0 0.0 0.0 8A 2.0 Cameroon na 18.6 0.0 0.0 23.4 163 17.8 5.5 0.0 na 14.7 3.8 21.0 na 0.0 0.0 0.0 na 4.8 Botswana 40.1 47.6 382 0.0 0.0 39.3 433 69.0 0.0 0.0 26.1 39.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Mauntius 27.2 333 0.0 0.0 0.0 16.7 22.1 66.7 0.0 0.0 0.0 20.0 25.0 4.C 0.0 0.0 0.0 29.3 0.0 Gabon 27.0 14.8 29.2 0.0 28.0 283 48.7 24.2 0.0 0.0 15.9 20.6 42.1 4.8 4.7 0.0 0.0 9.7 0.0

I Education scienceand teacher training Vl Mass communicationand documentation. XIII Engineering nI Humanites, religion and theoloV'. VEII Home economics(domestic science). XIV Architecture and town planning III Fine and appied arts. IX Service trades. XV Trade, craftand industrial programs IV Law. X Natural science. XVI Transport and communications. V Socialand behavioral science. Xl Mathematicsand computer science. XVII Agriculture, forestry and fishery. VI Commercialand businessadministration. XIl Medicalscience and health-related areas. XVII Other and not specified.

V Date induded in another category. Source: UNESCO (1987). Sub-Saharan Africa 99

Vocational Timig Oppotunitiw

A number of authors have attributed reluctance to send girls to school in part tc a societs response to a perceived lack of fit between the vocations for which schooling is supposed to prepare students and the vocations that are regarded as suitable for girls. Csapo (1981) and Wrzesinska (1980) explicitlyreferred to the wife-mother role, the former suggestingthat in northern Nigeria this was the only approved role and the latter indicating that it was the major role in which the subjects of her study--secondary school girls--appeared to have an interest. Csapo reports that a girl's expected vocation was to be a wife and mother, and therefore school was deemed unnecessary.

These views must be examined in the light of two different issues: the availabilityof modern-sector employment opportunities for women and the curricular segregation by gender, which often starts in the higher grades of elementary school and continues on to the secondary and tertiary levels. Girls are channeled into domestic science, handicrafts, and biology,while boys go into chemistry, mathematics, and, more directly,vocational subjects (Eshiwani 1982, for Kenya; Harding 1985).14 One of the results of this lack of preparation for the labor market may have been the fail in female labor force participation between 1960 and 1981 from 45 percent to 12 percent 4espite a massive expansion in education in the same period (Kelly 1988). The evidence for Africa muczak s that the causes for this disjuncture may have arisen from both the curricular choices women have been givon and the quality of the schools in which they were enrolled (Eshiwani 1982, Weis 1981).15 Consequently, neither the belief that women only need to be housewives(refuted by the overwhelming evidence that most women engage directly in economically productive activities in the informal and traditional sectors)16 nor that secondary school students are predominantly bnterestedin becoming wives and mothers (given the kind of training many girls are exposed to in school) is a useful or realistic generalization about the social or personal expectations of African girls. Unfortunately,curricular materials withinthe school too often exaggerate and perpetuate this unrealisticidea of a woman's role being confined to that of wife and mother. Both of these factors, however, determine the supply of places as weli as demand for schooling.

Lackof Role Models

Tembo (1984) surveyeda number of primary and lower secondary texts from Zambia as part of a UNESCO project on gender stereotypes in textbooks. The methodologywas simple: the number of occurrences of male and female characters were summed and their activities and characteristics noted. The results were instructive. The books contained many more male than female characters, and those female characters who appeared did so primarily in domestic roles and were characterized as passive,stupid, and ignorant. Men's activities were admired, women's ignored. No study that attempted to measure the impact of such stereotypes on girls was identified. One can conjecture, however, that the probable educational and occupationalaspirations of girls will be those of wife and mother when those roles are portrayed as the only appropriate ones for women. The presence of female teachers in schools and classrooms is often held to be a strategy for countering these images.

14 The efficacyof these school-basedvocational programs is dubious for both sexes. For girls, the training they receive is often irrelevant to the livesthey lead; for boys, quality is the more salient issue, although the relevance of the training can also be doubtful.

15 Of course, this is not the only explanation for this fall in female upply. Another is that the rise in women's edication has meant an increase in their nonmarket pru.,ctivity relative to productivity in available (and probably quite limited) labor market opportunities for more educated women.

16 See Dixon (1982) and UNECA (1984) for evidence on women in the agricultural sector. 100 Sub-Saharan Africa

SumMwy The factors shown to have affected girls' levels of enrollment, attainment, ara wastage in African countries include negative parental and communityattitudes toward the Western education of girls; the opportunity cost of a girPs school time and indirectly of her mother's; general levels of wealth and economic development; unfav. rable labor market opportunities;poor-quality schools, particularly with respect to the type of curriculum effered to girls; and regional disparitiesbetween urban and rural areas. Some of these factors are more amenable to interventionthan ott-ws and, except for curriculum offerings and the number of femaie teache-s, all require strategies that are cA. munitywideand not confined to the school. Conspicuous by its absence is any study that uses the lack of supply of places in primary school to explain the enrollment disparitybetween boys' and girls' enrollment rates. The tacit, and perhaps valid, assumption appears to be that even if enough places were available for all children of primary school age, girls would still be under-represented in many countries and school places would go unused on a regional basis (Jones 1980, for Tunisia; Bowman and Anderson 1980). The stud; that comes closest to linking supply factors to school enrollment found that, on average, primary school students were doser to the school than was most of the community (Biazen and Junge 1988, for rural Ethiopia). The conclusion is either that school enroltees are disproportionately drawn from families livingclose to the school or that those students who are enrolted move to live closer to it. One of the more fruitfut strategies for increashngfemale participation appears to be one that is aimed not at school-age girls but at their mothers. Suc'na strategy would have at least two prongs: one would be to reduce the mother's dependence on the labor of her children, and the other would be to raise the educational levels of mothers themselves. The first prong would involve the introduction of time-saving amenities, such as convenient water supply and fuel for the household. At least two projects have used such technologicalinnovations to reduce the time constraints on women and to free them for education: a project in Burkina Faso (McSweeney,Freedman, and Freedman 1980) and a USAID-funded soil and water project on agricultural innovation in Kenya (Knowles 1988). Literacy programs are an ongoing feature in many African countries, such as Tanzania, that have had much success in educating females. These programs must be continued. Closer integration with existing formal educational structures and grassroots organizations,such as cooperatives,could lower their costs and make them more salient to the livesof newlyliterate persons. Organizationalneeds, such as recordkeeping, can spur demands for literacy by providing an immediate use for it and can encourage the maintenance of literacy by providing regular opportunities for its use. Such factors as general levels of enrollment, leveti.,of economic development, and poor labor market opportunities for women require major long-term economic and planninginitiatives. Evidence from other continents suggests some pessimism as to the impact such changes would have on women's participation in education. These changes might be necessary, but they are far from being sufficient. While economic development might increaseenrollment, the effecton the other aspects of women's educational participation and indirectly on their social and economic welfare is dubious.

To allay fears that schools would be corrupting, indigenous individualsand groups should sponsor schools. For example,in parts of West Africa, includingNigeria and Sierra Leone, Muslim organizations (including women's groups) have opened schools that presumably do not engender fears that formal education will lessen girls' commitment to Islam or encourage non-Muslim values.

Achievement and Perforrance

The evidenceon th 4', jnal pf ..tor,--anceand achievementof girls usuaLlycomes from comparisons of exam results, *.. -- :.onwtexaiinations. The evidenceon the -lationship between gender and achievementis mixcrd.. 4atnple,Heyrieman (1975) found that genaer was the singlemost important variablein explainitgvariations i1 t achievement(that is,performance on the PrimaryLeaving Examination) in Ugandanschools before 1975,wita boys performingat a markedlyhigher level than girls. Lookingat mathematicsachievement in Nigeriaand Swaziland,Lockheed and Komenan(1988) found that for both Sub-SaharanAfrica lOi

countries,girls in single-sexschools achievedthe highest levels of performance. \ 'hile the reasons for the better performanceof all-femaleschools is not directlyexplored, the authors s,ate: "The picture that emerges from this comparisonbetween high, averageand low performingschools/classes is one of substantialdifferences between students, teachers and teachingpractices" (p. 25). The implicationis that all-femaleschools attract higherquality students and teachers,and havemore effectiveteaching than other schools.Amuge (1987), however, using a nationalsample of secondaryschool leavers in Tanzania,found that boysoutperformed girls in secondaryschool in almostevery subject. An intensivestudy of performance in Mauritiusprimary schools found that girlsoutperformed boys in both urban and rural areas (Chinapah 1983). Girlsin governmentand government-aidedNairobi secondary schools performed as weUlas boys, althoughpupils in single-sexschools performed better than those in mixedschools (Boit 1986). Unfortunately,none of the studies attempted to investigatedirectly the factors important for the achievementof girls. Several factors probablyaffect achievement,including family influences--family obligations(Fapohunda 1978; Wrzensinska 1980) and gender-baseddivision of roles (Wheldonand Smith 1986)--andschool factors--q1''-ty (Akande 1987; Eshiwani 1982; Weis 1981), teacher attitudes (Wondimagegnehuand Tiku. .,; Boit 1986;Amara 1987),and single-stoxschools (Wheldon and Smith 1986;Lockheed and Komenan1988; Boit 1986;Amuge 1987). The findingsrelated to family factors have been discussed earlier. For example, girls are more likelythan boysto have a numberof householdtasks to complete,and these are more likelyto take precedenceover schoolwork for them than for their brothers(Fapohunda 1978- Wrzesinska 1980). Less time for studying is likelyto lowerachievement. Of schoolfactors, those mentionedearlier includelow quality of schools. Eshiwani(1982) indicated that girls primarilyentered harambeeschools that had poorer equipmert,less qualified teachers, more limited curriculathan did the governmentor government-aidedschools that boyswere more likelyto attend. Weis (1981)indicated that 86.1percent of secondaryschool girls in Ghanawere enrolledin low-statusschools, as against43.0 percent of secondaryschool boys. Akande (1987)found that urban schoolshad higher achievingfemale studentsthan did rural schoolsbut did not discussthe mechanismsof this advantage. Giventhe financialconstraints most African governments face, the wholesaleupgrading of physicalfacilities, provisionof textbooks,and increasesin the numberof qualifiedteachers are not feasiblein the near future. Nevertheless, schools could make some changes that would improve school quality without necessarily incurringmajor new costsby addressingwhat Dreeben and Barr (1988)call technologicalchoices in the classroom.For example,Lockheed and Komenan (1988) showed that teachingpractices rather than teacher qualitywere predictive of higherattainment in schools.Barr and Dreeben(1983) made a similarpoint in a differentcontext: that the organizationof classroomtime and use of classroommaterials could effect major gainsin achievementlevels, independently of the abilitylevels of students. Outsidethe classroom, Eshiwani(1982) strongly urged organizational changes in the educationalsystem to removeor ameliorate the disadvantagesto girls. Increasingthe number of femaleteachers is also often recommendedas a strategyfor raisingachievement and attainmentamong females. Unlessfemale (and male)teachers are trainedto be sensitiveto gender equity, however, increasing the number of female teachers is not likely to be a source of positive images. A few studiesreported on the effectof teacherattitudes on achievementbut did not report the gender of the teacher. Nevertheless,the pointwas made that a negativeattitude toward the abilityof girls,and even in some cases toward their right to an education,was surprisinglyprevalent among teachers. From Ethiopia,Wondimagegnehu and Tiku (1988)reported that eighteenof the thirty-oneteachers interviewed felt that boyswere better than girlsin all academicsubjects. Boit (1986)reported that only40 percentof the Nairobiteachers and schoolheads interviewed felt that girlswould do as wellas boysif giventhe same opportunities.Based on an informalsurvey, Amara (1987)reported that teachersin Sierra Leone had higherexpectations of boys than of girls. The evidenceon the effectivenessof single-sexschools in raisinggirls' !..rformanceis also mixed(Table 3.10). In a sample of third-yearstudents from Nairobi secondary schoo,.., Boit (1986)found that girlsin single-sexschools performed as wellas boysfrom single-sex schools and significantlybetter than both boys and girlsfrom mixed-sex schools on a test of mathematicsachievement. In Swaziland,Wheldon and Smith 102 Sub.SaharanAfrica (1986)found that the relativepass rate in sciencewas the same for girls in both typesof schools. Girlsin single-sexschools were more likelyto choosemathematics and agriculturethan the physicalsciences, a pattern that may have accountedfor Lockheedand Komenan's(1988) fmding of superiormathematics achievementby girlsin single-sexschools. Although single-nex schools in Britainhave less gender bias than do mixedschools, gender biaswas greater in single-sexschools in Swaziland(Wheldon and 3mith 1986). Swazigirls in mixedschools may be takingadvantage of curriculumofferings intended for boys,or single- sex schoolsmay not be able to hire femaleteachers to teach subjectssuch as physicsand chemistry. Table 3.10 Effects of Single-SexSchools on Girls' Achievement

Source County Level Sample Comment

Amuge(1987) Tanzania 2 National,4,181 Form IV students No effect Boit (1986) Kenya 2 Nairobi government schools,3rd Positive year math class Lockheed and Nigeriaand 2 Eighthgrade, 8 southernstates; Positive Komenan(1988) Swaziland eighthgrade, volunteer Wheldon and Swaziland 2 Cambridge Overseas School Positive Smith(1986) Certificateentrants (1985)

Amuge (1987) found that among a sample of fourth-yearsecondary school studentsin Tanzania,no significantdifference was apparentbetween the performanceof girls in single-sexand mixed-sexschools when a wide range of subjects vas considered(commerce, agriculture, English, verbal ability,home economics,mathematics, math aptitude,Svahili, and technicalsubjects). Boysoutperformed girls in all subjectareas exceptcommerce, however.

Concluding Remarks Only a few direct programsby governmentagencies have been aimed specificallyat increasingthe participationof womenin schools(Damiba 1982;de Souza1982). Nongovernmentalorganizations and internationaldonors have made someefforts to raise levelsof femaleschooling, but their resultsappear to hav- been negligible,largely because these effortshave been small-scaleand limited(Stromquist 1986). In addition,sponsorship by ,.,reignorganizations may reinforce traditional beliefs about protectingwomen fromWestern corruptio, hich in turn can pose obstaclesto raisingtheir levelsof schooling. Expansionof educationalsystems has led to increasesin accessnot onlyfor girlsbut also for other groups withhistorically low levelsof schoolparticipation. This sa ategyhas been successfulin raisingthe levelsof femaleenrolment and attainmentbut has had less successin reducingor eliminatinggender differentials. For example,the provisionof 100 primary s ;ols in Kano State in Nigeria between 1976and 1981 quadrupledthe levelof girls' enrollment,but their relativeshare of schoolplaces only increased from 25 percentto 30 percent(Callaway 1984). Expansionis also a veryexpensive option. Extendingschooling to clents who do not now haveaccess by expandingthe systemwould require an administrativeand financial effort that wouldstrain most Africaneducation budgets. Chamie(1983) suggested, as an alternative,multiple use of teachersand buildingsas a way of expanding placeswithout incurring as large capital costs. A number of countrieshave tried this option with some Sub-SaharanAfrica 103

success.Such strategy may widen regional gaps, however, because those regions already well provided with schoolswould be able to generatethe largestnumber of additionalschool places. Moreover,unless teachers are to be paid extra, they may resist takingon increasedworkioads. Countrieswith fiscalconstraints and lessdeveloped educational systems will fnd it impossibleto provide more publicschooling to all children. Suchcountries as Nigeriaand Kenya,which have used publicand privateexpansion, respectively, to increaseaccess to education,still have not achieveduniversal primary enrolHment.Supplementary suppliers of educakionalservices nerd to be developed,but privatizationis only part of the answer,and in somecountries only a smaUlpart. Stromquist(1986) recommended the use ofwomen's organiz ;.ns andother nongovemrnentalorganizations to promoteand organize nonformal educational systems that couldsupplement formal systems. She reported that the record of these organizationswith internationalaid agencies has generally been positive. Nongovermnentalentities already have a frameworkand organization in placeaid are orientedtoward cost- effective,community-level work. Women'sorganizatior "otlhi. -'itional and international,already exist in Afr-an countries,and they servewomen in a numbero. - ormaleducational settings. Women in these countriescan overcomethe de-ficienciesof publicand privateiormal - 4ucationsystems by seekingtraining in such settings. The existenceof nonfortnaleducation settings am) tends to relievethe pressuresfor reform of formaleducational systems. The dangerhere is that such organizationsmay be overburdened by such demands. Religiouseducation systems can also be developed.Morgan and Armer (1988),with referenceto boys' education,point to the vigor and growth of Koraniceducation in Northern Nigeriaalongside Western education. Policiesfollowed by a number of internationalagencies have also affectewwomen's participation in education.The packageof structuraladjustment policies being advocated by the WorldBank/International MonetaryFund for manyAfrican countries could have some negative consequences for women'seducation, however. For example,the cutbackin the publicsector has a number of implicationsfor women and women'seducation. This sector is a disproportionatelylarge employerof womenin the modernsector, includingin education.The cutbackcan be regardedas an attackon twofronts: schoolplaces couldbe imperiled;17 and employmentopportunities for thosewho do attendschool will be reduced,a situationthat could further reduce the alreadylow labor market returns to educationfor women. In addition,the privatizationof the publicsector, induding education, could widen the chasmbetween those girls who have educationand those who do not. As more of the burden of educationalcost is shifted to the pupils' families,the strongconsumption aspect of girls'Western education at the secondaryand tertiarylevels will discouragepoorer parents. In thoseareas where enrollment is at or approachinguniversality, privatization may introducean economicconsideration into a decisionthat had becomeautomatic. Recommendationsmade in the WorldBank's Policy Stidy on Educationin Sub-SaharanAfrica appear more sensitiveto the problemswomen face as they try to enter and move through the educationsystem. Reconcilingthese recommendationswith WorldBank fiscal policy will not be simple,however. The AfricanDevelopment Bank'sEducation SectorPolicyPaper, published in 1986,names access and equity, indudinggender equity, as the firstof fivemajor problemareas that willguide its fundingpolicies. Support for proje-tsthat - to remedyinequalities and to ease enrollmentin institutionsin whichwomen have been under-represen.. ' willbe a special priority (p. 15). Such an emphasis presumaoly will be affected by the willingnessof memberstates to propose and supportsuch projects. A brief reviewof the most recent developmentplans from African countries suggests that projectsthat specificallyaddress gender inequality do not appear to havereceived much attention at the nationallevel, where general economic development and fiscalstability appear to be the major concerns(Metra Consulting1983). Any attempt to increaseaccess to schoolsor improvetheir qualityhas iaanrent politicalcontradictions. Suchefforts are likelyto increasethe demandfor educationfrom both genders,a situationthat in turn will

17 For example,the Governmentof Gambiareleased 2,000 teachers in 1988. 104 Sub-SaharanAfrica lead to additionaWpressure in Suchareas as unemploymentamong school leavers and furtherexpansion and differentiationof the educationalsystem. These pressureswill be equallydifficult whether governments chooseto withstandthem or givein to them. Futurepolicies and actionsneed to be more firmlygrounded in a dearer understandingof the barriersto womea'seducation. Research needs to focuson specificcontexts on a nationalor even regionalbasis. An important initial step would involvemarshalling research already completed,but not made generally available,by research institutes,university departments of education,teacher traininginstitutions, and ministriesof educationwithin each country.Enough data are availableon most countriesto describefully how gender inequaiitymanifests itself withinthe educationalsystem. More analysisof the process is needed,however--that is, how females are disadvantaged--beforestrategies to remedythe disadvantagescan be developed.The analysisneeds to be donein a multivariatecontext, with appropriate controls, including parentalsocioeconomic status or the socioeconomicstatus of the relevanthousebold. Sub-Saharan Africa 105 References

Abraha, S., A. Beyene, T. Dubale, B. Fuller, S. Halloway and E. King. "What Factors Shape Girls' School Performance? Evidence from Ethiopia?" InternationalJoumat of EducationalDevelopment, 11: 107- 118, 1991. Adams, M.N., and S.E. Kruppenbad, "Gender and Access in the African School." IntemationalReview of kuucation, 33:457-63,1987. Adjebeng-Ascm, S. "Womenand Technology."Paper presented at the Aftican Studies Association Meeting, Chicapo, Illinois, November, 1988. Akande, B.1. "Rural-Urban Comparison of Female Educational Aspirations in South-Western Nigeria." ComparativeEducation, 2.:75-83,1987. Amara, J.M. "IndigenousTechnology if Sierra Leone and the Science Education of Girls." Intemational Joumal of Science Education, 9.317-24,1987. Amon-Nikoi, G. "Women and Work in Africa." In Human Resourcesand African Developmen4 edited by U.G. DamXchiand V.P. Diejomaoh, pp. 188-219.New York: Praeger, 1978. Amuge, I.M. "Gender Differences in Academic and Post-school Experience among Tanzania Secondary School Students."Unpublished Ed.D. dissertation. State Universityof New York, Albany,New York, 1987. Assie,N. "EducationalSelection and Social Inequalityin Africa."Unpublished Ph.D. dissertation. University of Chicago, Chicago, 1983. Avalos,B., and W. Haddad, Review of TeacherEffectiveness Research in Africa,India, Latin America,Middle Eas4 Malaysia, Philippines and Thailand.t Synthesis of Results. Ottawa, Canada: International Development Research Centre, 1979. Barr, R. and R. Dreeben, How Schools Work. Chicago: Universityof Chicago Press, 1983. Bali, S.K., PJ. D. Drenth, H. -an der Flier, and W.C.E. Young, Contribution of Aptitude Tests to the Predictionof School Performancein Kenya.A LongitudinalStudy. Lisse: Swets and Zeitlinger, 1984. Bedri, N.S., and L.G. Burchinal, "EducationalAttainment as an Indicator of the Status of Women in the Sudan." The Ahfad Jouma4 2:30-38,1985. Biazen,A., and B. Junge, "Problemsin Primary School Participationand Performance in Bahir Dar Awraja." Report prepared for UNICEF and Ministry of Education, Addis Ababa, Ethiopia, 1988. Biraimah, K.L."Class, Gender and Life Chances:A NigerianUniversity Case Study."Comporative Education Review, 31:570-582,1987. Bledsoe, C. "The Politics of Polygynyin Mende Education and Child Fosterage Transactions." In Gender Hierarchies,edited by BD. Miller. Submitted to The University of Chicago Press, Chicago, 1988. Boit, M.K. "The Relationship of Teacher Behaviorto Student Achievement in High and Low Achievement High Schools in Nairobi, Kenya." Unpublished Ph.D. dissertation. University of Oregon, Eugene, 1986. Bowman, MJ., and CA. Anderson, 'The Participation of Women in Education in the Third World." Comparive Education Review, 24:S132, 1980. 106 Sub-SaharanAfrca

Callaway,BJ. "AmbiguousConsequences of the Socializationand Seclusion of Hausa Women."Journal of Modern African Studies, 22:429-50,1984.

Chamie, M. National,Institutionaland Household FactorsAffectingYoung Girls'Attendance in Developing Societies. Washington, D.C.: International Center for Research on Women, 1983. Chernichovsky,D. 'Socioeconomic and Demographic Aspects of School Enrollment and Attendance in Rural Botswana."Economic Developmentand CulturalChange, 33:319-32,1985.

Chinapah, V. Panlcipationand Perfom,ancein PrimarySchooling. Studies in Conparative and International Education, 8. Stockholm: Institute of Interoational Education, University of Stockholm, 1983. Chishimba, M.M. "Language Policy and Education in Zambia." IntemationalEducation Joumal, 1:151-80, 1984.

Cooksey,B. "Social Class and Academic Performance:A Cameroon Case Study." comparativeEducation Review, 25:402-17,1981.

Csapo, M. "Religious,Social and Economic Factors Hindering the Education of Girls in Northern Nigeria." Comparatve Education, 17:311-19,1981. Damiba, A. "EducationalInequalities in Upper Volta." In Inequalitiesin EducationalDevelopmren4 edited by M. Blaug and others, pp. 31-38.Paris:IIEP, 1982.

de Souza, A.R. 'The New School Polik, to Reduce Inequality in the People's Republic of Bel en." In Inequalitiesin EducationalDevelopment, M. Blaug (eds.) and others, pp. 39-43. Paris: IIEP, 1982. Denzer, L. "Colonial Education for Women in Nigeria, 1861-1945."Paper delivered at Women and Education in Africa Symposium,University of Wisconsin,Madison, November, 1988. Dixon, R. "Women in Agriculture: Counting the Labor Force in Developing Countries."Population and DevelopmentReview, 8:539-66,1982.

Dreeben, R., and R. Barr, 'The Formation and Instruction of Ability Groups." American Joumal of Education,97:34-64. Chicago: Universityof Chicago Press, 1988. Eshiwari, G.S.A Study of Women'sAccess to HigherEducation with a Special Referenceto Science and Math Educaton. Working Paper No. 5003. Nairobi: The Bureau of Educational Research, Kenyatta University College, 1982.

Fapohunda, E.R. "Women at Work in Nigeria: Factors AffectingModern Sector Employment."In Hruman ResourcesandAfrican Developmen4 U.G. Damachi and V.P. Diejomaoh (eds.), New York: Praeger, 1978.

Gambetta, D. Wer They Pushed or Did TheyJump? Cambridge:CambridgeUniversity Press, 1987. Garenne, M., and E. van der Walle, 'Polygynyand Fertility among the Sereer of Senegal."Paper presented at the University of Chicago, Chicago, Spring, 1988.

Gordon, LX,and others. "Women in the World of Science and Technology." Working Paper. Paris: UNESCO, 1985.

Harding, J. 'Girls and Women in Secondary and Higher Education: Science for Only a Few." Prospects, 15:553-64,1985.

Heyneman, S. "Influences on Academic Achievement in Uganda: A 'Coleman Report' from a Non- industrial Society."Unpublished PhD. dissertation. Universityof Chicago, Chicago, 1975. Su-Saharan Africa 107

Husen, T., LJ. Saha, and R. Noonan, Teacher Training and Student Achievement in Less Developed Countries.World Baik Staff Working Paper Nlo.310. Washington, D.C.: World Bank, 1978. Jones, M.T. 'Education of Girls in Tunisia: Policy Implications of the Drive for Universal Enrollment." Compaative Education Review, 24:S106-23,1980. KeDly,G.P. "Perspectiveson Women and Education in the Third World." Paper presented at Women and Education in Africa Symposium,University of Wiscunsin,Madison, November, 1988. Kelly, G.P., and C.M. Elliott, (eds.). Women's Education in the Third WorlddComparative Perspectives. Albany State University of New York Press, 1982. Knowles, J. "Non-formalEducation Review Agriculture." Paper presented at the Women and Education in Africa Symposium,University of Wisconsin,Madison, 1988. Kossoudji,S., and E. Mueller, "The Economic and Demographic Status of Female-headed Households in Rural Botswana."Economic Developmentand CulturalChange, 34:831-59, 1983. Lanzas, A., and P.W. Kingston,'English Achievementin Zaire: The Effects of FamilyStatus and Residential Disruption: A Report." ComparativeEducation Review, 25:431-41,1981. Lockheed,M.E., and A. Komenan, "SchoolEffects on Student Achievementin Nigeria and Swaziland."PPR Working Paper. WashingtonD.C.: The World Bank, 1988. Mann, D. "Can We Help Dropouts? Thinking about the Undoable." In School Dropouts Patterns and Policies, edited by G. NatrieDlo.New York: Teachers College Press, 1986. McSweeney,B G., and M. Freedman, 1980."Lack of Time as an Obstacle to Women's Education: The Case of Upper Volta." Comparative Education Review 24:S124-39,1980. Metra Consulting. Handbook of National DevelopmentPlans, Volumes I and 2 (Basic Work and 1988 Supplement 1). London: Graham and Trotman, 1983. Mies, M. 'Towards a Methodologyfor Feminist Research." In Theones of Women'sStudies, G. Bowles and R. Klein (eds.). Boston: Routledge and Kegan Paul, 1983. Morgan, W.R., and J.M. Armer, "Islamic and Western Educational Accommodation in a West African Society A Cohort Comparison Analysis."American SociologicalReview, 53:634-39,1988.

Newman,J.S. Women of the World Sub-SaharanAfrica Washington,D.C.: U.S. Department of Commerce, 1984.

Nkinyangi,J. "SocioeconomicDeterminants of Repetition and Early Withdrawal at the Primary School Level and Their Implicationsfor EducationalPlanning in Kenya."Unpublished Ph.D. dissertation. Stanford University, Stanford, 1980.

- ~. "Material, Pedagogicaland Socio-psychologicalDifferences in Kenya's Primary Schools and their Impact on Educational Outcomes." Developmentand Cooperaton, 12:547-78,1981. "Access to Primary Edncation in Kenya: The Contradictions of Public Policy. Compaive Educatfon Review, 26:199-217,1982. Orubuloye, 0. "Valuesand Costs of Daughters and Sons to Yoruba Mothers and Fathers." In Ser Roles, Population and Developmentin WestAfrica, C. Oppong (ed.). London: Heinemann, 1987. 108 Sub-SaharanAfrica

Ramn,R. "SexDifferences in the Labor Market Outcomes of Education." Women's Education in the Third World Comparative Perspective Editorial by G. Kelly and C. Eliott, pp. 203-27. Albany State University of New York Press, 1982. Robertson, C.C. "The Nature and Effects of DifferentialAccess to Education in Ga Society."Aflica,47:208- 19,1977. "Formal------. or Nonformal Education? Entrepreneurial Women in Ghana." ComparativeEducanon Review, 28:639-58,1984. .Women's------. Education and Class Formation in Africa, 1950-1980."In Women and Class in Africa, C. Robertson and I. Berger (eds.). New York: Africana PublishingCo., 1986. Robertson, C., and I. Berger, (eds.). Women and Class in AfAica New York: Africana PublishingCo, 1986.

Smith, J.E., and P.R. Kunz, 'Polygyny and Fertility in Nineteenth Century America."Population Studies, 30:465-80,1976. Smock,A.C. Women's Educationin DevelopingCountries: Oppoflunities and Outcomes. New York: Praeger, 1981. Stromquist, N. "School-relatedDeterminants of Female Primary School Participation and Achievementin Developing Countries: An Annotated Bibliography."Report No. EDT83. Washington, D.C.: World Bank, 1987. Tembo, L.P. Men and Women in School Textbooks:A National Survey of on Sex Biases in Zambian Textbooksin PHimaryand Junior SecondarySchools and Their Implicationsfor Education in Zambia Lusaka: UNESCO, 1984. Tibenderana, P.K The Irony of Indirect Rule in Sokoto Empire, Nigeria, 1903-1944."The African Studies Review, 31:67-92,1988. Tripp, A.M. "Women and the Changing Urban Household Economy in Tanzania." Unpublished paper. Evanston, Illinois,1988. UNECA. "The Role of Women in the Solution of the Food Crisis in Africa."Arusha, Tanzania, 1984. UNESCO. 7rends and Projectionsof Enrollmentby Level of Education and by Age: 1960-2000(as Assessed in 1982). Paris: Office of Statistics, UNESCO, 1983. StatisticalYewbook Paris: UNESCO, 1987. Weis, L "Schoolingand Patterns of Education in Ghana." CanadianJoumal of Afican Studies, 15:311-22, 1981. Wheldon, A.E., and A.C. Smith, "Gender-based Differences in O-Level Choice and Performance in Swaziland."Bulletin of the SwazilandInstitute for Educational Researc., 7:78-91, 1986. Wondimagegnehu,A., and K. Tiku, "Participationof Ethiopian Women in Education". Processed. Ministry of Education, Addis Ababa, 1988. Wood, M.M.; W.W. Swan, and NJ. Wood, "Supply and Demand Factors Influencing the Educational Dropout Rate in Africa and Asia."Intetational Education, 15:37-4, 1986. Sub-ahiaran Africa 109

World Bank. Educationin Sub-SaharanAfrica: Policies forAdjustmen4 Revitalizationand Expansion.World Bank PolicyStudy. Washington, D.C.: World Bank, 1988. ------WorldDevelopment Report 1989. Washington, D.C.: World Bank, 1989. Wrzesinska,A. "A Young African Girl, Her School and Her FamilyEnvironment." Africa, 35:357-85,1980. MiddleEast and North Africa 111 Chapter 4. Middle East and North Africa

Nagat EM-Saaby

The MiddleEast-North Africa region encompasses affluent oil exporterssuch as SaudiArabia and Kuwait with GNP per capita of $6,200and $14,610,respectively, and low-incomecountries such as Moroccoand Egyptwith respective per capitaincomes of $610and $680. The countriesof the regionare predominantly Muslim,but Turkeyand Tunisiahave relatively liberal laws on womenand familystatus, while Saudi Arabia ha- a muchstricter interpretation of Islamiclaws. Overall,female literacy and educationlevels have tended to be lower in the MiddleEast and North Africa than in Latin Americaand East Asia. The region's educationalsystems and attitudes are shapedby Arab culture dating back thousandsof years, Islam, Ottoman Turk rule, and Christianmissionary influences in the 19thand early 20th centuries. Egyptwas subjectto European influencesbeginning with the Napoleonicinvasion in 1798. In recent decades,the countrieshave undergonerapid expansionof their nonagriculturalsectors. Nevertheless,some Muslim countriessuch as SaudiArabia and Kuwaithave the world'shighest population growth rates--in most cases stillrising--which exerts a strongdrag on GNP growth,except in oil-richnations for as longas the oil lasts. AlthougbIslam supports education of womenand lawsin almostall Arab statesmandate equal educational opportunitiesfor males andfemales, the Arabvalue system is rigid respecting women's sexual conduct; once a woman's'honore is lost, it cannot be regained. Familyhonor dependson conformityof femalesto a "modestycode." The result is sexualsegregation, strict parentalsurveillance, veiling, early marriages,and rigidsex-role socialization. Nevertheless, because of the diversityin interpretationsof the codefrom country to countryand a varietyof developmentlevels, population densities and exposuresto Westerninfluences throughoutthe r._gion,education levels diverge widely. Illiteracy rates of womenrange from40 percentin Kuwaitto 86 percentin Egypt. Primaryeducation levels range fromnear-universal enrollment in Tunisia and Turkeyto more than a third of school-agegirls out of primaryschool in Moroccoand SaudiArabia. The age of educationsystems differ widely as well. Publiceducation for girls in SaudiArabia began only in 1960,while female education in Egyptdates fromthe mid-19thcentury. In Egypt,nevertheless, a fourth of eligibleprimary school girls are stll out of school. SaudiArabia's newer system, on the other hand,has movedfa-t and appearsheaded toward universal enroDlment. Overal, enrollmentratios at the primarylevel havemade impressiveprogress since 1960 (need overal fiure), indudingthe femaleshare which rose from 41 percentto 45 percent. But the numberand qualityof schoolsfor malesexceeds that for females.And the educationgender gap tends to be widerin this regionthan in many other parts of the Third World, SouthAsia excepted,seen in measuresof femaleenrollment shares, percentage of studentswho are female, particularlyat the primarylevel, literacy rates, and numberof yearsof schooling. This chapter focuses on seven Middle Eastern and North Africancountries: Turkey, Jordan, Egypt, Morocco,Tunisia, Saudi Arabia, and Kuwait.All are Arab exceptTurkey, which can be seen as a bridge betweenEurope, Asia, and the MiddleEast, and whichis alsoa membt.rof the Organisationof Economic Co-operationand Development(OECD). PredominantlyMuslim, the sevencountries have a sharedhistory of more than five centuries. Geographically,Egypt, Morocco, and Tunisiaare locatedin North Africa; Turkey and Jordan, in NorthwesternAsia; and Saudi Arabia and Kuwait,on the Arabian Peninsula. Politicaly,Morocco, Jordan, and Saudi Arabiaare monarchies;Kuwait is a semi-monarchy,and Egypt, Tunisia,and Turkeyare republics. SaudiArabia and Kuwaitare key membersof the Gulf Cooperating CounciL 112 Middle East and North Africa Economic and Educational Setting EconomicConditions

Economically,the seven countries are quite diverse (table 4.1). Saudi Arabia and Kuwait are affluent oil- exporters and members of the Organization of Petroleum Exporting Countries (OPEC), and have per capita GNP levels comparable to those of developed countries. The others are considered low-middle- income countries. The countries also differ demographically. Turkey is the most populous, with Egypt a close second, while Kuwait has a small population. Life expectancy,another indicator of development, varies widely among these countries. It is lowest in Morocco and Egypt at 61 years and highest in Kuwait at 73 years. Although Saudi Arabia is a high-income country, it has an average life expectancy,63 years, commonly found among lower-middle-incomecountries. Table 4.i GNP, Population Growth, and Life Expectancy(1980-87)

Population Population Urban GNP Annual Life mid-1987 GrowthRate Population per Capita Growth Rate Expectancy Country (millions) 1980-87(%o) 1987 (%lo) 1987 (US$) 1965-87(%lo) 1986 (Years)

Middle-Income/Lower Middle Morocco 23.3 2.7 47 610 1.8 61 Egypt 50.1 2.7 48 680 3.5 61 Turkey 52.6 2.3 47 1,210 2.6 64 Tunisia 7.6 2.6 54 1,180 3.6 65 Jordan 3.8 3.9 66 1,560 na 66 High-Income Oil-Exporting Saudi Arabia 12.6 4.3 75 6,200 4.0 63 Kuwait 1.9 4.5 95 14,610 -4.0 73 na = not available. Sources: World Bank (1989a)

Education Levels A review of the burgeoningliterature on Middle Eastern and North African women generally finds a dearth of research on women's education. Reflecting the academic background of researchers, most of whom are sociologistsand anthropologists, the focus has been on women's status and roles, labor force participation in the modern and traditional sectors, family,fertility, and related topics. In addition, most of this research has been carried out by non-Middle Easterners. The limited research on female education specificallyhas tended to be conceptual or theoretical, except for a few recent empirical studies. The most recent country data on illiteracyshows that illiteracyrates of females are much higher than for males (table 4.2). For example, in Tunisia in 1984, the illiteracyrete of women is still 50 percent higher. Girlsare less likelythan boys to enter primaryschool, and once enrolled,they are less likelyto complete the cycleor pursue postprimaryeducation (Kaneko 1987). Illiteracyrates vary among these countries, Middle East and North Africa 113 however. Kuwaiti females have the lowest illiteracyrates, and Egyptian and Moroccan women appear to have the highest. National figures also mask intracouintrydifferences in literacy and educational attainment, such as those between urban and rural areas. What l iited data exist suggest much higher illiteracylevels for both men and women in rural areas. Table 4.2 Illiteragr Rates by Sex (&oenenof rekvant population)

Male Female Country/Year Total Rural Total Rural

Middle-Income/Lower-Middle Morocco (1971) 66.4 78.1 90.2 98.7 Egypt (1976) 46.4 55.5 77.6 86.9 Turkey (1985) 12.4 -- 35.7 Tunisia (1984) 39.5 -- 59.4 Jordan (1979) 19.9 -- 49.5 High-Income Oil-Exporting SaudiArabia (1982) 28.9 -- 69.2 Kuwait (1985) 21.8 -- 31.2

Notes: In all countries except rural Egypt, the reference population includes all those who are fifteen years and older; in rural Egypt, it includes those ten years and older. Source: UNESCO StatisticalYearbook 1990.

Female enrollment has increased substantiallyat all educational levelsin the sevencountries, especially since the 1960s (see table 43). Between 1975 and the mid-1980s,the female share of enrollment, averaged across the seven countries, rose from 41 percent to 45 percent at the primary level, from 36 percent to 42 percent at the secondary level,and from 29 percent to 39 percent at the tertiary level. These figures reflect major improvements at the primary level in Egypt (from 38 percent to 43 percent) and Saudi Arabia (36- 44 percent); at the secondarylevel in Tunisia (34-42percent), Jordan (41-48percent), and Saudi Arabia(33- 39 percent); and at the tertiary level in Morocco (19-33percent), Jordan (33-41percent), Saudi Arabia (20- 39 percent), and Turkey (16-33percent). In Kuwait,female students were a majority in higher education in both 1975,at 57 percent, and 1985,at 55 percent. As a result of the recent growth in female enrollments, the gender gap in educational attainment is narrower among the younger age groups. 114 Middle East and North Africa

Table 4.3 Female Share of Total Enrollmentby Education LeveL1975 and 1987 (percent of total emifflnt)

ZbxfiLel SecondLevet ThirdLevel Country 1975 1987 1975 1987 1975 1987

Egypt 38 43 34 41 30 33 Morocco 36 39 34 40 19 33 Tunisia 39 45 34 42 26 36 Jordan 46 49 41 48 33 41 Kuwait 46 49 45 48 57 55 Saudi Arabia 36 44 33 39 20 39 Turkey 45 47 31 36 16 33

Note: The most recent data for Saudi Arabia is 1986. Source: UNESCO StatisticalYearbook 1990.

The overal participation rates of boys and girls also showthe considerable progress all seven countries have made in extendingeducational opportunitiesamong school-agepopulations (table 4.4). At the primary levet Tunisia and Turkey seem to have come close to achieving universal primary enrollment, but Egypt, Morocco, and Saudi Arabia continue to have difficultyin this regard, especially for girLs. Egypt had 100 percent male and 79 percent female gross primary enrollment in 1987,Morocco 85 percent and 56 percent, respectively,and Saudi Arabia 78 percent and 65 percent. In Saudi Arabia, however, achieving universal access is probablyjust a matter of time, giventhat public education for girLsonly began in 1960. In contrast, although female education in Egypt dates from the second half of the nineteenth century, gender disparities in primary education have persisted. Middle East and North Africa 115 Table 4.4 Gross Enrollment Rates at the Primary, Secondary,and Tertiary Levels 1950-1987(pem of ,lvantpopulaon)

First Level Second Level 7ThirdLevel Country Year Male Female Male Female Male Female

Egypt 1950 52 30 27 7 7.81J 1.64/ 1968 85 55 40 20 10.8W 4.0W 1975 89 60 55 31 18.5 8.3 1987 100 79 79 58 28.6 14.8 Tunisia 1950 51 19 11 5 na na 1968 100 83 29 11 na na 1975 116 78 28 15 6.3 2.1 1987 126 107 46 34 7.2 4.0 Morocco 1950 31 11 1 1 0.8'/ 0.19/ 1968 72 36 18 6 2.0W 0.4W 1975 78 45 21 12 5.2 1.2 1987 85 56 43 30 11.9 5.7 Jordan 1950 71 25 7 1 na na 1968 100 85 54 24 1.00 0X/ 1980 105 102 79 73 28.9 24.2 1983 98 99 80 78 41.5 33.2 Kuwait 1950 92 45 4 1 -- 1968 100 85 71 61 0.99/ 2.¢ 1975 99 85 71 61 7.1 11.2 1987 95 92 85 80 11.9 21.0

Saudi Arabia 1950 4 na 0.2 - 0.21a na 1968 44 18 12 2 1.02/ Q1j/ 1980 75 50 36 23 8.8 5.0 1986 78 65 52 35 12.00W3/

Turkey 1950 56 36 7 3 4 5E 1.1w/ 1968 93 61 34 14 4 2.0A/ 1980 103 92 40 19 14.8 3.1 1987 121 113 57 34 .8 5. na = not available. -- = nonexistent.

Note: The net enrollment ratios are actually lower than those shown. The gross enrollment ratios exceeding 100 are the result of the enrollment of over-age students. i1960. k/ 1970. SI 1969. d/ 1984.

Source: Data for first and second levelsare fron UNESCO StadisticalYearbook (1970 and 1990). Data for third level are from UNESCO StatisticalYearbook (1971L1972, and 1990). 116 Middle East and North Africa

Among the seven countries, educational access and participation by females, especiallyat the basic primary level, are higher in countries with high rates of male participation. Kuwait, Turkey, Tunisia, and Jordan, which have near universal male primary enrollment, have come close to attaining it for females also. Those countries experiencing difficultiesin universalizingprimary education for males, notably Morocco, Egypt, and Saudi Arabia, have the lowest rates of female participation in primary education. At the secondary level, the increases in female enrollment ratios in Kuwait--from1 to 80 percent between 1950 and 1987--haveput it far ahead of the other countries. Kuwait appears to have been successful in enforcingcompulsory education at the primary and lower secondary levels. Despite its lower ranking at the primary leveL Egypt surpassed Tunisia, Saudi Arabia and Turkey at the secondary level, with a female enrollment ratio of 58 percent, versus 34 percent in Tunisia and Turkey, and 35 in Saudi Arabia. Jordan, like Kuwait, has been relativelyadvanced in its female enrollment ratios not just for secondary schools (78 percent) but at all educational levels. At the tertiary level,Jordan led the other countries with enrollment ratios of 41.5percent for males and 33.2 percent for females. Egypt was second at 28.6 for males and 14.8percent for females, followedby Kuwait (11.9 and 21 percent, respectively)and Saudi Arabia (12 and 9.3 percent). Tunisia, Turkey, and Morocco all had female enrollment rates below 6 percent. Additionallywomen tend to select traditionally female fields of study, namely,the humanities and social sciences. As in other Islamic countries, however,they are highlyrepresented in medicine, and in Egypt and -.urkey they are also weil represented in the traditionally male field of engineering. Despite the increase in female enrollment at all educational levels.gender disparities persist throughout the education sectors (table 4.5). For instance, although Tunisia has been the second most successful in extending primary education to girls, the gender gap at that level was still relatively wide in 1987 at 19 percentage points. Egypt, which had the third highest female enrollment ratio at the tertiary level after Jordan and Kuwait, had the widest gender gap at that level at 13.8 percentage points. Except in Jordan, the gender gap was lower at the higher level than at either the primary or secondary levels.

Table 4.5 The Gender Gap in Gross Enrollment Rates 1987 percentagepoinb)

FirstLevel Second Level ThirdLevel Rank County Gap Country Gap Country Gap

1 Jordan -1 Jordan 2 Kuwait -9.1 2 Kuwait 3 Kuwait 5 Saudi Arabia 2.7 3 Turkey 8 Tunisia 12 Tunisia 3.2 4 Saudi Arabia 13 Morocco 13 Turkey 6.0 5 Tunisia 19 Saudi Arabia 17 Morocco 6.2 6 Egypt 21 Egypt 21 Jordan 83 7 Morocco 29 Turkey 23 Egypt 13.8 Source: Computed from table 4.4

The lower educational attainment of girls and women has been a result of both low enrollments and high attrition and wastage. A growAngbody of literature shows that girls are more likely than boys to drop out of school before they complete the cycle, especiallyin rural areas. Research on Egypt, Morocco, Tunisia, and several other countries documents this pattern (see Hartley and Swanson 1986 and Cochrane 1986for Egypt; Youssef 1978b and Massialas and Jarrar 1983 for Morocco; and Deble 1980 and Chamie 1983for several other Middie East_rn countries). The reverse is true of Saudi Arabia and Kuwait,where girls have Middle East and North Africa 117

had consistently lower repetition and drop-out rates than boys in primary and secondary education. This pattern has led their governments and educational planners to conclude that the internal efficiencyof girls' education is higher than that of males (Arab Bureau of Education in the Gulf States 1983, Kingdom of Saudi Arabia 1985). Generally, secondary education in Middle Eastern and North African countries is of three types: general academic, teacher training, and vocational/technical. The latter is divided further into industrial, commercial, agricultural, and, for girls, nursing education and schools of domestic sciences. Table 4.6 provides statistics on female secondary education enrollment, distribution among the three main types, and the percentage of girls in total enrollment at both the lower and upper secondary levels (intermediate and secondary proper) in each type of program. Between 1975 and 1985, a remarkable increase occurred in female secondary education enrollment in all seven countries. The bulk of female (as well as male) secondary education enrollment in the seven countries converges toward college-preparatory academic programs, ranging from 73 percent of female secondary students in Egypt to more than 99 percent in Kuwait and Saudi Arabia. The predominance of girls in this education is a result of prestige and abundant supply. Only those with the highest grades are admitted from the intermediate secondary graduating classes; it is considered the "right"education for bright, hard-working, ambitious girls. The proportion of girls who enter the scientifictrack in preparation for college is relatively high. Not all those who enter secondaryschools complete the cycle, however, and not all who complete it achieve the grades required for college admission. Table 4.6 Enrollment of Girls in SecondaryEducation and Their Percentage in Total Enrollment, by Type of Education, 1975 and 1986/87

Total SecondaLv GeneralEducation TeacherEd'n VocationalEd'n Female % of Female % of Female % of Female % of Enrollment Total Enrollment Total Enrollment Total Enrollment Total Country Year (000's) (000's) (000's) (000's)

Egypt 1975 745 34 604 34 14 44 127 34 1987 1,685 41 1,261 40 62 59 362 40

Morocco 1975 167 34 160 34 2 43 5 38 1987 535 40 529 40 A/ 0 5 32 Tunisia 1975 68 34 50 34 0.6 52 18 31 1987 199 42 166 43 3 71 31 34 Jordan 1975 67 41 65 41 / 0 2 29 1987 172 48 158 49 / 0 14 42 Kuwait 1975 49 45 49 46 / 0 0.20 15 1987 122 48 122 48 0 0.09 10

Saudi Arabia 1975 66 33 62 33 4 29 0.20 6 1986 256 39 250 40 6 52 0/ Turkey 1975 550 31 433 32 25 47 91 28 1987 1,183 36 959 38 4 37 220 30

A/ = Magnitude nil. Source: UNESCO StatisticalYearbook (1990). 118 MiddleEast and North Africa Accessof girls to technicaland vocationaleducation in Arab countrieshas been addressedat numerous regionalconferences on educationand at two special UNESCOconferences, one in Algeria in 1964 (UNESCO1964) and the secondin Kuwaitin 1969(UNESCO 1969). The concernsexpressed then persist, and the recommendationsare stillvalid. In vocationaleducation, a large proportionof femaleenrollment is in the domesticsciences. About 50 percent of secondaryeducation enrollment in Turkeyis in this specialty. Female access to vocational educationin both Kuwaitand SaudiArabia is negligible;it is limitedto sewingand tailoringschools and nursinginstitutes. Enrollmentin secondaryteacher training has declinedin Morocco,Jordan, and Kuwait becauseof the upgradingof primaryteacher trainingto the iunior collegelevel. Severalfactors account for the low accessof girlsto technicaland vocationaleducation in MiddleEastern and North Africancountries. Participantsat the 1969UNESCO conference attributed it to dominant beliefs,attitudes, and conceptsconcerning the nature and aptitudesof womenand the typesof workand occupationssuitable for them (El-Ghannam1970). Sex-rolestereotypes limit the supplyof vocational educ ion for girls,and the lowstatus of vocationaleducation depresses the demandfor existingprograms. The attitudethat a woman'splace is in the home prevails,obviating the need for job-relatedtechnical and vocationaleducation; and the stigmaof vocationalschool being a repositoryfor the untalenteddiscourages those who may need it. These attitudes are strongestin SaudiArabia and Kuwait,as reflected in the virtuallack of vocational educationfor girls. Becauseof their uniquesituation, Kuwait and SaudiArabia are not includedin the folowingdiscussion, except to note that in recentyears SaudiArabia has faced mountingpressure from femalesecondary school graduates for entry into collegesand universities.In response,officials have been consideringvocational education to prepare girlsfor workin factoriesfor womenpatterned after the sweat shops of the West, where womencan use their "nimblelittle fingers"in traditionaltextile and garment industries. A seriesof surveyscarried out bythe UNESCONational Commission in variouscountries, including Jordan and Turkey,was the basisof a studyon gender differencesin secondaryschool curricula by UNESCOin 1981. The surveysfound extensive differentiation based on perceivednorms and not individualabilities or interests. This situationwas the major hurdle facinggirls in education(UNESCO 1983). In Jordan, althoughboth boysand girlsfollowed the same courses,girls tended to becometypists or secretaries,while boysnormally tried for jobsin managementor continuedon to the universityor polytechnicinstitutes. The new comprehensiveschools that were introducedexperimentally to integrate general academicand vocationalsubjects still had genderdisparities. In boysschools, a widerange of traditionalmale vocational subjects was offered, whereas the girls school in Amman offered the traditional female subjects-- dressmaking,art, ceramics,child care, and hairdressing.The girlsstudied all four subjectsin the first year and specializedin the last two years. Vocationaltraining followed gender lines as well. In trade trainingschools that were not part of the secondarycycle in Jordan, boys learned traditionalsubjects, whereas girls studiedhome economicsplus twenty-fourhours of 'practical"subjects, including beauty culture, dressmaking, and ceramics(Cameron 1983).In the vocationalschools run bythe UnitedNations Relief and Work Agency, curriculum stereotypes existed as welL A recent studyby the World Bank of alternativemeans of vocationaleducation and technicaltrami4ig made no referenceto girls,an indicationthat they were not considereda target group. InadequatevocAtiona! education places women at a disadvantagein the job market. Severalresearchers have fwundthat in mostof these countries,many women have had to settle for low-paying,low-status jobs becausethey lackedthe necessaryskills (Hatem 1983;Hammam 1981; and Rihani1983). The concenftrntionof girlsin generalsecondary education is reflectedin their distnbutionamong the various academicdisclplines at the postsecondarylevel. They tend to enter traditionallyfemale fields--the humanities,social sciences, and education(table 4.7). Nonetheless,because of valuesthat encouraAewomen to enter occupationsthat serveother women and because of the Muslimaversion to the treatmentof female patientsby male doctors,women are well-representedin medicinein all sevencountries. Further,because MiddleEast and North Africa 119 of their relativelyhigh access to math and scienceat the secondarylevel in manyMiddle Eastein countries, their representationin engineeringand naturalsciences is generallyhigher than in manyWestern nations.

Table 4.7 Females as Percentage of Total Higher Education Erollment, by Field of Study 1987

Field of Study Eypt Morocco T7wa Jordan 7W*q Kuiwai Arabia4

Total Female Enrollment 34 33 37 S0 33 5S 39 Education Science & Teacher Trg 46 28 30 73 45 76 52 Humanities, Religion & Theology 47 97 44 60 42 63 42 Fme and Applied Arts 47 27 54 S8 S2 37 na Law 20 27 34 32 36 58 k Social and Behavioral Sciences 42 31 34 43 31 S3 29 Commerce & Bus. Administration 31 S1 41 39 30 60 na Mass Comm. & Documentation S6 48 63 30 S4 na na Home Economics (Domestic Science) 79 na 43 na 97 na na Serce Trades 79 38 25 na 25 na na Natural Sciences 40 28 38 46 45 54 37 Mathematics & Computer Sc. 27 Oa 25 30 41 63 k Medical Science and Health 39 33 54 S2 40 66 42 Engineering 13 12 12 16 17 31 17 Architecture and Town Planning 17 25 S2 na V Trade, Crafts, & Industry Program iS na na 42 11 na

Transport and Communications 11 13 na 1 39 na Agriculture, Forestry & Fshery 32 22 18 35 26 on 24 Other/Not Specifred 36 37 na 26 0 35 na

*na= not available. Data for Saudi Arabia refers to 1985. h Data included in field of study inmediately above for which data are available. Source: UNESCO S nsicalYerbook (1990).

Factors Affecting Education Quality and Reach As witheducation in anycountry, that for girlsand women in MiddleEast andArab countriesis influenced by a complexset of demographic,sociological, cultural, political, and economicfactors. These include cultural and religiousbeliefs, level of economicdevelopment, level of educationaldevelopment and educationpolicy, family influences, societal factors, and schoolfactors. Dominant Cultural and ReligiousBeliefs The bodyof literaturedeveloped over the past fifteenyears documents changes in the positionof women in various Islamiccountries as a result primarilyof the spread of modern education. Nonetheless, misconceptionsand stereotypesabout MiddleEastern and Muslimwomen persist because of a lack of appreciationof the complexsocioeonomic forces affecting women's education and statusin the MiddleEast and North Africa. In fact, empiricalevider.ce on the effects of cultural traditionsand values on the 120 Middle East and North Africa

education of females in Islamic countries is scarce, except for a few attitudinal studies on Egypt, Tunisia, Morocco, and Jordan. The strongest indictment of Islam as the cause of the disparities in female education in Muslim countries generally comes from Finn, Dulberg, and Reis (1979),who concludedthat "In no major portion of the world is the noneducation of females so much a purposeful part of religious and social custom as it is in the Middle East," and that "educationis contrary to the social pressure for Muslim women to become wivesand mothers." Smock and Youssef (1977) concluded the same. Nevertheless, although the record of Muslim countries relativ- to non-Muslim ones on women's educaticn has been generally wantin , contrary to the Western stereotype, no inherent bias existsin these countries againsteducation for females . Muslimsquote the saying of the Prophet, "the search for knowledge is the duty of every Muslim, man or woman," in support of female education. Within Islamicculture, however,several attitudes and traditions that influence planners and parents inhibit female education. Muslims have a strong concern for the modesty and safety of girls and women. The desire to guard the honor of females is evident in their seclusion and veiling. Antoun (1968), Mernissi (1987), Saadawi (1985), and Abu Zahra (1970) provided some insights into the workings of this system and the limitationsit imposes on the private and public roles of women. Indeed, Arab cultural traditions, rather than Islam er se, have constrained girls education (Othman 1964). The evidence of these traditions include opposition to coeducation and employment of male teachers in girls schools, resistance to sending girls to schools awayfrom home, and pressures on girls to marry at an early age. Prevailingsex-role stereotypes and related divisionof labor in the home and marketplace have also resulted in a lower valuation of female education. Socializationstarts at a very early age and continues during the early years of schoolingand beyond (Safilios-Rothschild1986). From the moment of birth, sons are valued more than daughters, although attitudes vary across classes and communities. Very little research is available on the dynamicsof this process, however. In much of their writings,the Egyptian,Saadawi (1985) and Moroccan, Mernissi (1987) have detailed the sharp distinction between male and female socialization from a feminist perspective and discussed the negative effects on girls' self-esteem and aspirations. In addition, Patai (1973) contended that the Arab world had no "child"-rearingpractices, only "boy"-and "girl"- rearing practices. Generally, Middle Eastern girls are socialized into accepting the predominant sex-role stereotypes, with marriage and raising a family as the ultimate goal. Schools continue to reinforce this differential socializationof gender roles. The prevailingattitude is that a woman should stay home to care for her family and children, while her male guardian--husband,father, brother or older son--shouldsupport her. This divisionof responsibilityobviates the need for women to be educated or to join the labor force to earn a living. Perhaps as a result of this dependence, Middle Eastern women,as in many other countries, are accorded a lower social and economic status than are males. Fear of the greater freedom and independence of thought and action that women gain through education has been a concern during the early history of girls' education, as was found in Egypt in the 1950sand 1960s and in Saudi Arabia in the 1980s. The assertiveness,self-esteem, and belief in gender equality that women may develop through education threatens those who espouse a traditional belief in male superiority. According to Nelson (1968), "the dilemma for both men and women is how to reconcile the man's self- image as dominant authority to the woman's emerging self-image as an equal". Recently, major, although uneven, changes have taken place in attitudes toward women and their role in all Middle Eastern and North African countries. Although observers fear that the emergence of the so- called Islamic fundamentalism signals a reversal of women's roles and a return to the home, the actual

1 A similar view is given by Marshall (1984): "Muslim women have been less involved in traditional agricultural and trading activitiesand have significantlylower rates of literacy,educational achievement,and nonagricultural labor force participation than women in other developing areas at similar stages of industrializaton. This divergence has commonlybeen explained by the religio-culturalinfluence of Islam, which presumably mitigates the impact of social changes precipitated by the development process. While Islam has certainly affected the lifestyles of adherents of both sexes, this constant cultural factor is insufficientexplanation for the considerable range of variation in national levels of female participation within the Muslim region." Middle East and North Africa 121 impact of this trend on women's status, role, and education is yet to be ascertained. Veiled, young girls and women are able to participate as students even on coeducational campuses (El-Guindi 1981). To many Middle Easterners, veilinghas been a way of reconcilingthe "publicuand "private"spheres of women. As Egyptian and other Middle Eastern women venture into the public spheres of education and employment,they have found ingeniousways to bridge the two worlds and accommodate new opportunities within traditional cultural and Islamic values. The existence of traditional attitudes and values calls for caution when talking about reforms for women in education and employment. Terms such as the liberation of women, equality of men and women, economic independence, and the use of education as a means to reduce fertility have to be used carefully, because they have negative connotations, particularly among conservativeparents and educators. Terms such as providing equality of educational opportunities, enhancing the welfare of the family,and acquiring income-generating skills that contribute to family welfare are less threatening and more likely to enlist support. Policymakers,educators and parents are unlikelyto challengea woman's basic right to education. Anything that threatens familystability or the welfare of children, however,including a reduction in fertility, is likelyto be seen as cultural encroachment by the West.

Level of Economic Development Middle Eastern and North African countries, as noted, represent extreme levels of economic development, from very poor to extremelywealthy. The -ven countries reviewedhere have per capita GNP ranging from a high of $14,610 in Kuwait to a low of Ui1Oin Morocco (table 4.1). Several writers have stressed the important effect of national economic development on girls education, givingillustrations from Kuwait (El- Sanabary 1973,1978, 1985;Meleis, El-Sanabary,and Beeson 1979)and Libya and Bahrain (Chamie 1983). The data in tables 4.2, 4.3, and 4.4 corroborate their findings,showing the remarkable progress made by the wealthier Kuwait and Saudi Arabia in reducing the gender gap and extending education to larger segments of the school-agepopulation. Despite their much more conservativesocial climate, Saudi Arabia and Kuwait have spent generously on education, expanding indigenous and expatriate staff and supply of textbooks. Within a relatively short time, they caught up with, and even surpassed at least numerically, other Middle Eastern countries with much longer histories of girls education; for example, the latest enrollment figures put Saudi Arabia ahead of the more liberal Tunisia and Turkey. By comparison, the poorer Middle Eastern countrieshave been unable to provide the schools,teachers, equipment, and facilities needed to educate a considerable segment of their populations,despite compulsoryeducation laws and free schools; for example, the two lowest middle-incomenations, Morocco and Egypt, have had the greatest difficultyexpanding educational opportunities and reducing gender disparities. Thus, insufficienteducational facilities limit children's access to education, and participation rates reflect this scarcity. Demography and economicsinteract to influence the supply and demand for girls' education. The size of the population and its growth rate determine the educational effort required to reach the target age groups, and economic resources limit the amount countries can devote to educational development. Ironically, the most populous states are also the poorer ones that lack the resources to educate their large and ever- growing school-age populations. Egypt, for example,with a population of about 50 million (see table 4.1) that is growing at a rate of 2.7 percent annually and with per capita GNP of $680 in 1987, has been struggling to satisfy the growing demand for education. Universalizationof primary education, especially among girls, continues to be a major problem, and the quality of education leaves much to be desired. Morocco is also handicapped by a relatively large population of 23.3 million and a per capita GNP of only $610. In contrast, Kuwait,with a population of just 1.9 million and per capita GNP of $14,610,has, as noted, made remarkable progress. SaudiArabia, with a much larger populationthan Kuwait,has made significant progress but is taking much longer to reach the educational level and quality achieved in Kuwait. Because of generally high birth rates and low life expectancies, the population of Middle Eastern and North African countries is youthful. About 40 percent of the population in Egypt, Kuwait, Morocco, and Tunisia is under 15 years of age. In Jordan and Saudi Arabia, 48 and 45 percent, respectively,fall within this age group. Morocco's populationis expected to reach 32 million by the year 2000 (World Bank 1989). Egypt's . opulation is expected to reach 67 million in year 2000 and 99 million in 2025. The expansion of 122 MiddleEast and North Africa this very youthfulbase has seriousimplications for the burden of expandingeducation, in generaLand femaleeducation, in particular.

Level of Educational Developmentand Education Policy The levelof educationaldevelopment as used here refersto the maturationof an educationalsystem and societyas revealedby the history,range, and type of educationalinstitutions and their distnbution,as weU as the fiteracyrate and educationalattainment of the adult population.Some MiddleEastern countries haveadvanced educational systems that go back more than a century,others have less developedand more recent ones. Generally,the more developedand diversifiedthe educationalsystem, the better are the educationalopportunities for females. Oil has transformedthis relationship,however, giving the wealthy states an edge over the poor ones. The oil-richcountries may not have as much educationaldiversity as the more developedsystems do, but they are catchingup quickly. Educationtends to generateits own demand. Educatedpeople aspireand appreciateeducation for their childrenand often push them to acquireat least a levelsimilar to their own. Schoolingnorms becomea tradition, and the social demand for education perpetuates itself from generation to generation (Psacharopoulos1977). The old conceptof educationas a prelude to governmentemployment, inherited from the colonialera, has had a depressingeffect on the supplyof girls educationalfacilities. Despite laws for compulsory educationand proclamationsabout the right of all citizensto education,governments have designed educationfor job preparation,thus accordinga lowerpriority to the educationof femalesbecause of their low levels of labor force participation. In the 1970s,this conceptwas replaced by a new and more sophisticatedone: 'the manpowerplanning" approach to education. Governments,with the help of educationalplanners and internationalconsultants, shifted their attentionto preparingyoung men for high- and middle-leveljobs required for industrializationand modernization. The result was emphasison secondaryand higher educationto the relativeneglect of primary mass education. Governmentsalso emphasizedvocational and technicaleducation for males,while girls had to contentthemselves with home economicsclasses that preparedthem for domesticresponsibilities. The educationalpolicies of MiddleEastern governments can be categorizedas generaland gender-specific, or gender-focused(Klein 1987). General policiesapply to all childrenregardless of gender. Intentional gender-specificpolicies may be directed toward "gender differentiation"or toward "gender equity," dependingon their bases and effectson the educationof girlsand women. All MiddleEastern and North Africancountries have easilyidentifiable general educational policies that applyequally to malesand fQmales.Delineating the elementsof intentionalgender-specific policies requires a detailedassessment that is beyondthe scopeof this paper. Where they exist,such policiesare directed towarddifferentiation rather than equity. To date, no MiddleEastern countryhas issuedgender-equity policiessuch as the "affirmativeaction" policies of the UnitedStates and England. Where gender-specific policiesare not explicitlyexpressed, they are often implicitin educationalpractices.

CompuLqEducation Compulsoryeducation legislation in the MiddleEast and NorthernAfrica has not guaranteedequal access of girls to education. The first of the sevencountries to enactsuch laws were Egypt(1923) and Turkey (1924). Egypthas encounteredmajor difficultiesin universalizingprimary education, while Turkeyhas succeededbecause of a deliberatepolicy of enforcingcompulsory education laws. Althoughboth countries set target dates for achievinguniversalizadon-1970 and 1972,respectively--Egypt kept extendingthe date whileTurkey made it a nationalcommitment on socialand politicalgrounds. The Turkishgovernment saw primaryeducation as fundamentalto democracyand a keychannel for promotingmodernization (Szyliowicz 1973). MiddleEast and North Africa 123 In the 1970s,many economists, and the World Bankas welt believedthat primaryeducation was not as importantas higherlevels of education(see Harbisonand Myers1964; World Bank 1974). Reflectingthe thinkingof the 1970s,with its lowpriority for primaryeducation, Szyliowicz criticized the Turkishpolicy because of its massivefiscal and human resourcerequirements and a fear that it wouldlead to further deteriorationin educationalquality. To the contrary,it provedeffective in reducingilliteracy rates (to 26 percentfor malesand 55 percentfor females)and raisingthe educationallevel of the population.Egypt, on the other hand,continues to sufferfrom high illiteracyrates (70 percentfor males and 86 percentfor females),and about 25 percentof school-agegirls are not in school. Hartleyand Swanson(1986) noted that compulsoryeducation laws were not enforced in Egyptianschools and that they were 'probably unenforceable"because of inadequatefacilities. In short,compulsory education laws are meaninglessunless enforced and supportedby adequatefinances and facilities.Otherwise, they can onlyserve as a shieldagainst discriminatory allegations. In orderto raise the educationalstandard of females,Egypt must commititself truly to expandaccess of girls to basic primaryeducation, to improvethe qualityof their education,and to reducewastage. These goalscannot be accomplishedwithout political support (El-Ghannam 1970 as quotedin Chamie1983; Silliman 1987).

Adniions Pocies

All sevenMiddle Eastern and North Africancountries have liberalopen admissionpolicies, especially for generalsecondary education and mosthigher educational institutions. Entrance is basedon studentscores in highlycompetitive national examinations. Although, theoretically, these policiesapply to all children regardlessof gender, classor other factors,they do not guaranteeequity, because educational motivation and achievementare stronglyinfluenced by students'socioeconomic status. Furthermore,open admission cannot compensatefor the lack of girls' schoolsin smalltowns and rural and remoteareas. Also, open admissiondoes not apply to all types of educationbecause of gender-specificpolicies that restrict or prohibit,explicitly or implicitly,female accessto certain types of secondaryvocational and technicaleducation or to certainfields of highereducation. These policiesare based on belief in innate gender differencesand on prevailingnotions about appropriate male and female educational and occupationalspheres. Among the seven countries,Saudi Arabia has the most explicitgender-specific differentiationpolicies, and it is the onlyIslamic country that has completelyseparate systems of maleand femaleeducation. Nevertheless,it continuesto expandwomen's curriculum options in highereducation.

FreeE 1Uin

All sevencountries provide free educationto both sexes from primary through the tertiary,and even graduate,levels. Moreover,some providescholarships and low-costdormitory facilities at the university leveL In principle,this policyshould provide access to the lowersocioeconomic strata, but free education is not enough,especiaUly at the primaryleveL Parentsof millionsof children,mostly female, still cannot affordthe other directand indirectcosts of schooling:clothing and shoes,books and schoolsupplies, and fees of privatetutors, now a necessityin most educationalsystems because of crowdedclassrooms and decliningeducational quality. Schoolingalso has opportunitycosts-that is, the incomeand other benefits of productiveor learningactivies at home that are foregonein favorof schoolattendance. Free or publiclysubsidized postprimary education (especially at the universitylevel) can run counter to expansionof primary education. Also, it favorsthe urban bourgeoisieand helps perpetuate the elitist orientationof education(Psacharopoulos 1977; Psacharopoulos and Woodhall1985; Coombs 1985; Todaro 1985).

Emphais on GeneMiAcadmc and Hh*erEducation The free and open admissionpolicies, as wellas emphasison certificationand degreesfor modern-sector employment,have created a dysfunctionalimbalance in Middle Eastern and North Africaneducational 124 Middle East and North Africa

systems that has had important implicationsfor female education. Two manifestations of this imbalance are the emphasis on general versus vocational and technical education, and the emphasis on higher education at the expense of basic primary education. Access to higher educattion,nonetheless, has opened new opportunities for women, as documented by statistics and research showing the advanced position of Middle Eastern women in higher education compared to those in other developing countries.2 Examination of the data on the number of students in higher education per 100,000inhabitants highlightsthe relative advantage of Middle Eastern women at this level (table 4.8). In 1987, relative to their respective populations, Kuwait and Jordan had more females in higher education than either Japan or the United Kingdom. Much of this success in expanding higher education was achieved through costly subsidies and at the expense of basic primary education. Thus, higher education has created an elite group of women and men and has widened the gap between them and illiterate females (El-Sanabary 1973, 1985).

Table 4.8 Enrollment in Higher Education per 100,000 Inhabitants, 1975, 1980, and 1987

Male Female 1975 1980 1987 1975 1980 1987

Kuwait 635 739 1,110 1,009 1,330 1,758Ja Jordan 594 1,313 1,877 313 1,183 2,096 Egypt 1,821 2,372 2,485 808 1,i03 1,086 Saudi Arabia 555 882 1,237 154 403 917 Turkey 1,348 805 1,322 269 246 700 Morocco 426 892 1,214 97 267 60WJ Tunisia 536 696 714 191 297 434 France 2,101 na 2,394 1,845 na 2,395 Germany F.R. 2,168 2,447 3,273 1,241 1,566 2,123 United Kingdom 1,716 1,924 2,099 920 1,056 1,735

Japan 2,773 2,820 2,636 1,284 1,333 1,507 USSR 2,045 na 1,772 1,803 na 1,876

Note: Data for Jordan is for 1985, and for Saudi Arabia, 1986. na = not available. Source: UNESCO Staistical Yearbook(1990).

2 Cochrane, Mehra, and Osheba (1986) noted that in the Egyptian educational system, higher education claimed 17 percent of educational enrollment as compared to an average of 3.7 percent for developing countries. The relativelyhigh proportion of Middle Eastern women in higher education was noted also by Youssef (1974, 1978b), El-Sanabary (1973), and Meleis, El-Sanabary, and Beeson (1979). Middle East and North Africa 125 Family Influence As the basic social unit, the family has ultimate authority over its members, especially the young and female. Decisions regarding education, choice of career, marriage, and even childrearing are a family affair. Despite the commonalities, major differences exist in socialization, expectations, and amount of support parents provide for their daughters educational and occupational aspirations, depending on the family's socioeconomicstatus and specificcharacteristics. Parents' demand for education of their children is shaped not only by their attitudes and values but also by their assessment of the cost and outcome of educational options. Under certain circumstances,they may feel the anticipated benefits do not justify the expected costs. For instance, alternative income-generatingactivities cr marriage will prevail unless parents are assured of receiving a comparable return for their investmentin their daughters' education.

Family Income A family's socioeconomic status influences a daughter's education directly through financial and moral support for schooling or indirectlythrough a set of interveningvariables such as health, physical,cognitive, and psychologicaldevelopment, as well as her own motivation,aspirations, and expectations. The family's income determines its ability to support a child financiallythrough schooling. Generally, girls and women from upper and middle-incomefamilies are more likelythan less fortunate girls to enter school and progress to the tertiary level and even to pursue graduate studies. Wealthier families are able to enroll their daughters in public or private schools, at home or abroad, and provide them with financial and moral support. For instance, before Saudi Arabia opened its first public girls school in 1960,many wealthySaudi families sent their daughters to public and private schools in Egypt and Lebanon. They were also the first to take advantage of universityeducation for girls when it became available at home (El-Sanabary 1988). Throughout the Middle East and North Africa, girls are a majority of the students in private schools and col:eges, proof that well-off families place a high value on their daughters' education and willsupport them in expensivei.istitutions. Professional Egyptianparents working in the oil-rich Gulf states are able to provide their children with such educational opportunities as private schools and extensivetutoring for the British general competencyexamination (GCE) for college entrance. Many such children are able to bypass the secondary education cyclecompletely by taking that examination and entering an Egyptian university or the American University of Cairo right after the intermediate education level. In Egypt, there is a whole new generation of these young people. In contrast, many working-class families cannot even support their children's education in free public schools. In addition, they often need their children at home to care for their siblings. The need for domestic and farm labor in urban slums and rural areas is a major barrier to female education. Despite educational expansion and tuition-free schools, education remains beyond the means of large segments of the population. Those from a working-class background are also more likely to enroll in vocational education, nursing, and teacher training institutes than in the general stream at the secondary level or in professional education at the tertiary level (Howard-Merriam 1979;El-Sanabary 1983). These non-academic secondary schools provide easy access to paid jobs. By comparison,professional higher education, while prestigious, takes longer and is costly. Female students in engineering, medicine and other professional programs tend to come from wealthier families (El-Sanabary 1985;Papanek 1985;Howard-Merriam 1979, 1981;and Moore 1980). For instance, Egyptian engineers tend to have an "elite social origin," and female engineers come from an even more upper class background than do men (Moore 1980).

3 Sixty-fivepercent of the female engineers under the age of 35 came from the urban bourgeoisie and aristocracy--thetop 11 percent of the Egyptian population--as compared with 41 percent of their male cohorts (Moore 1980). 126 Middle East and North Africa Rum Residec

Approximatelyhalf of the population of Middle Eastern and North African countries lives in rural areas--53 percent in Morocco and Turkey, and 52 percent in Egypt--and some countries have relativelylarge nomadic groups. Girls and women in rural areas and among the nomadic population are educationallyhandicapped by their gen Ier, residence, and class. Their enrollment rates are consistentlylower than those of urban populations. The concentration of schools in the more affluent urban areas, cultural traditionalism among rural inhabitants, and the perceived irrelevance of school education to agricultural activities place rural children at a disadvantage compared to urban children.

The girls in a rural communitythat lacks a school may either commute daily to one in a neighboringtown or may board, where possible. The daily commute to school (either on foot or by undependable and often primitive transportation) may be costly and risky and is not alwaysan option for many young village girls. Boarding facilities are not always a practical solution because they can be expensive and because social stigma may prevent young girls from living away from home. The lower participation rates of girls in rural areas reflect both low rates of access and high rates of attrition. In Morocco, attrition and repetition occur at all grade levels in primary education, especially during the fifth (the final) grade. Youssef(1978b) reported research findingsindicating that among students who completed the primary level, only 10 percent did so without repeating, 26 percent repeated once, and 42 percent repeated twice. Dropping out is also a serious problem in Morocco; the rate was estimated at 21 percent for boys and 28 percent for girls during the first year of primary school, and half of the rural children who entered primary school never completed the cycle. Similarly, Hartley and Swanson (1986) found that attrition rates were much higher in rural than in urban areas and among girls than among boys in Egypt.

Paetal Education and Epectations The educationallevel of a girl's parents is probably the most important factor in determiningher educationalchances. Educatedparents are more likelyto have accessto resourcesand information,as wellas an appreciationfor the value of education. A studyof a sampleof about 500students in Kuwait Universityfound that the majoritycame from familiesin whichboth parents knew at least how to read and write; only 14 percent had illiteratefathers and 28 percent illiterate mothers (Al-Thaquib1975). Cochrane,Mehra, and Osheba (1986)found that, holdingincome constant, parental educationhad the most influenceon educationalaspirations for childrenin both rural and urban areas in Egypt. The higher the educationallevel of the parents,the greater the expectationsfor educationof their sonsand daughters, and this effect was larger for daughters' than for sons' education (table4.9). In rural families,this intergenerationaleffect implies a tremendousincrease in educationfor daughters.A studyin Jordan found the educationof fathersto be systematicallyassociated with the schoolingof daughters(Stuart 1981).

4 The rural-urbandisparities were noted by Szyliowicz(1973) for Turkey,El-Sanabary (1973) for Libya, Algeria,and Syria;Youssef (1976-77) for Morocco;and Massailasand Jarrar (1983)for the Arab countries in general. Middle East and North Africa 127 Table 4.9 Effect of Parents' Education on the Education Level They Want for Daughters and Sons: Egypt

Father'sAmiraions for Mother'sAspiratons fwr Area Daughters Sons Daughters Sons

Urban Areas 0.22** 0.10** 0.11** 0.05 Rural Areas 0.48** 0.22** 2.13** 0.98*

Note: These regression coefficientsindicate the additional yeirs of schooling parents would have wanted for their son or daughter if they themselves had one more year of schooling.

d Statistical significanceat 5 percent level. ** Statistical significanceat 1 percent level. Source: Cochrane, Mehra, and Osheba (1986).

In one of the largest surveys of Egyp:ian students and their families in the southern and western parts of Cairo, researchers found that parents placed a high value on their daughters' education and had to make economic sacrifices to enable them to continue their studies. Most of the parents, 94.5 percent, intended to allow their daughters to pursue an education, and 84 percent of those wanted them to have a university education in order to enter prestigiousjobs (Khattab 1984). For others, education of their uaughters was compensationfor their own lack of schooling. In many families,the tradition of education was a deciding factor. Among the parents of limited means, 82 percent considered the cost of private tutors a burden on the family budget but were still willing to provide them to improve the scores of their daughters in the national examinationsand to increase their chances of gaining accessto universities. The girls also had high aspirations for their own education; most wanted to become doctors or engineers, the two most prestigious occupations. In an Egyptian fertility survey,parents were asked what level of education they desired for their children.5 a.,erafl, 75 percent wanted a university education for their sons and more than 50 percent had that aspiration for their daughters. The regional differences were very sharp, however (Hallouda and others 1983,as quoted in Cochrane 1986). People in urban areas generally held high aspirations for both sons and daughters; in rural upper Egypt, interest in a daughter's education decreased dramatically (Cochrane, Mehra, and Osheba 1986). In a study of two rural communities in Egypt and Tunisia, Larson (1988) found that parents strongly emphasized their daughters' education as leading to an office job and an easier life. Similarly, a study of a sample of 100 nomadic, urban, and rural women in Saudi Arabia found that the majority of the urban women interviewed wanted their daughters to complete their education utpto the universitylevel (cited in Allaghi and Almana 1984). This findingwas true even for recently settled illiterate women, although some did not know the difference between the various educational levels.

Poverty continues to stand in the way of many families and their aspirations for their children. Khattab (1984) noted that in Egypt, lack of financial means and family poverty influenced parents to enroll their daughters in vocational education so that they could work after graduation and help support their families.

5 This survey providedmuch of the data for the published research on parental effects on female education in Egypt. 128 Middle East and North Africa

More than half of the forty-sixcases in this group indicated financial need as the reason their daughters did not pursue higher education, while early marriage was reported in only six cases. A mother's education has a particularly strong influence on her daughters' education. Educated mothers can help their children with their schoolwork,especially during their early years of schooling. They also provide more positive reinforcement of their children's educational and occupationalachievement. Their standards and expectations for their daughters are different from those of uneducated mothers, and their daughters usually have greater confidence in their abilities to pursue higher levels of education (Bach and others 1985). Al-Thumali (1984) found that, in Saudi Arabia, the higher the educational level of mothers, the greater was their influence on their daughters' academic plans. A study of the gender-role orientation of Arab university students in Lebanon, Kuwait,and Egypt found that the daughters of educated mothers held less stereotypicalattitudes than did those who had uneducated mothers. A mother's work experience also had a positive effect on the attitudes of her sons and daughters (Lorfing and Abu Nasr 1985). Basson (1981, 1982) found that, even within the same family,attitudes may differ toward girls' education. In Jordan, many girls drop out of school to help with domestic tasks or because villagers see twelve years of education as unnecessary for marriage. Often girls continue their schooling under harsh family circumstances. Althoug!hone daughter is taken out of school to help at home, another in the same family is sent to a teachers' college and her marriage delayed several years. Like her unmarried brothers, she is expected to contribute cash incomteto the household. The decision could depend on the daugh. -r's birth order, school performance, or attitude toward school; and a highly motivated daughter may be difficultto dissuade from going to school. A family could also see tangible benefits from her education in the form of a job, income, or better marriage prospects.

Fami Size Because family size is associated with other factors, including family income and parental education and occupation,especially that of the mother, evidenceof its effect oil girls' education is scarce and inconclusive. It is suggested,however, that the smaller the family size, the greater the chances of daughters entering and advancing in school. Small families, even those of modest means, may not have to choose whether to educate a daughter or son. A study of parents' aspirations for their children's education found that larger families with children under 13 years of age had lower aspirations for the education of their children, especially daughters (Cochrane, Mehra, and Osheba 1986). In another study, also in Egypt, the number of children in the family affected both home praise and self-confidence(Bach and others 1985). These findings are consistent with those of Rosen (1961),who argued that smaller familysize allowed parents to commit more time to each child, thus enhancing the children's verbal and cognitive development and consequentlytheir educational attainments (cited in Bach and others 1985). Despite the generally positive effects of a small family on schooling,family size may be insignificantif the family is wealthy or, in poor families if elderly family members can relieve a girl from domestic responsibilities. In affluent Kuwait, amonig a large sample of students at Kuwait University,Al-Thaquib (1975) found that 67.3 percent came from familiesof seven or more people and 19.8percent from families with four to six members. This effect was also established by research conducted in Botswanaand Taiwan (Chamie 1983).

Women'sPa#coaton unthe Labor Force One of the main functions of education is to prepare young people for their adult roles. Thus, women's roles in the family and the paid labor force have important implicationsfor their education. The level and pattern of female participation in the work force are reflected in the goals and orientation of women's education. Both educational policymakers and parents base their decisions, in part, on the perceived linkages between education and employment. Despite the growing numbers of Middle Eastern women who work outside the home in formal and nonformal occupations, women's rates of participation are low compared with other developed and Middle East and North Africa 129

developing countries. For example,in 1975,the percentage of economicallyactive females in the modern sector ranged from 3 percent in Jordan to 8 percent in Morocco. Moreover, women are often concentrated in traditionally female fields of employment, such as education, medicine,and social welfare (Youssef 1974, 1976-77;Azzam 1979;Hammam 1986). In practice, Middle Eastern women have played and continue to play a vital role in the agricultural and nontraditional work force, either as unpaid family workers or small- scale business owners (Basson 1981;Hatem 1983;Hammam 1986;Larson 1988). Major variations exist across Middle Eastern countries both in the types of positions open for women and attitudes toward their employment. Employment conditions range from the quite liberal and often mixed work environments in Egypt to the strictly gender-segregated ones in Saudi Arabia.6 Traditional attitudes continue to prevail in most of these countries, favoringoccupations with minimal intermingling of the sexes. Although the potential labor force participation of women feeds back into the family's decision regarding education options for daughters, no systematic study has tried to measure the magnitude of this effect or to identify which families are more likelyto respond to labor market opportunities for women. In Saudi Arabia, by law, women's employmentis limited to teaching, socialwelfare, and medicine. This puts severe restrictions on the educational choices for women. This is an extreme case, however,which does not apply in the other six countries. Moreover, although this employment restriction shapes the choice of field by those who reach higher education, it may or may not affect the educational attainment of those who did not continue beyond secondary education.

School Factors

Although school characteristics, such as type of school and curricula, affect female access to school and performance, their impacts have not been systematicallytested. Empirical evidence exists for the effect of distance to school, but very little is available on other school factors. Most studies tend to examine the effect of school characteristics in isolation (through simple tabulations), thus preventing an assessment of the importance of those characteristics relative to each other and to family or community factors.

School Location

As pointed out earlier, the availabilityof a school in the local communityincreases enrollment of females. Cochrane, Mehra, and Osheba (1986) found that, in rural areas of Egypt, distance to a secondary school is negatively associated both with the aspirations of parents for their children's education and with the probability of a child attending school. Interviewswith parents in Tunisia, Morocco, and Egypt indicated a reluctance to send their daughters to distant schools because the commute could entail morally or physicallydamaging incidents (Allman 1978, 1979;Basson 1981). This reluctance applies whether the girl has to commute or live in boarding facilities. Some reluctance is also due to the fact that daughters may be needed at home to help with domestic or farm work.

School Type and Physical Characteristi&s

Because of prevailing social and religious values, Islamic countries oppose coeducation, especially at the secondary level. The reason is their concern over the behavior of adolescent girls and boys in a mixed enviromnent. Attitudes and policies vary, however, across countries and socioeconomic groups. Saudi Arabia prohibits coeducation beyond kindergarten, while in Turkey and Tunisia coeducation is prevalent at all levels. In the other four countries, the general pattern is single-sexschools at the secondary level and both coeducational and single-sexones at the primary and tertiary levels. Coeducational foreign and private schools, in which female students outnumber males, are found in all these countries. Generally, all coeducational schools have both female and male teachers and administrators, while single-sex schools

6 For an overviewof the employmentof Arab women,see Azzam, Abu Nasr and Lorfing (1985), chapter 1. 130 MiddleEast and North Africa usuallyhave teachers and administratorsof the samegender as ,he students. None of the sevencountries has womenteachers in boys' intermediateor secondaryschools, but severalallow men to teach in girls' schools. How does the prevalenceof single-sexschools affect female access to educationand the quality and outcomeof their educationalexpeaience? Recent research,especially in Englandand the United States, failsto provideany simple answer. Afterreviewing several studies on the subjectof coeducationin primary basic educationin developingcountries, Chamie (1983) conduded that the negativeimpact that is found results,in part, fromthe difficultyof providingseparate facilities and of recruitingfemale teachers for girls' schools;that is,the burdenis on governmentsrather than on students.Areas where educational authorities have to provideseparate but equal educationalfacilities for males and females,particularly in sparsely populatedareas and villages,often lack girls' schoolsor schoolswith more than a few primarygrades. Attemptsto solvethe problem of scarce educationalfacilities in sparselypopulated areas by adopting coeducationmay be counterproductive.The reluctanceof someArab governmentsto ratifythe UNESCO ConventionAgainst Discrimination in Educationmay have stemmedfrom its inmlusionof a clause urging memberstates "to adopt the practiceof coeducationin first and secondlevel education as one means of ensuringequality of accessto education"(El-Sanabary 1973). Single-sexschools are not necessarilydiscriminatory, according to UNESCO(1966), 'so longas the same numberof places is offeredto pupilsof each sex, and the conditionsunder whichinstruction is givenare equal." In manycountries in the region,this is a costlyproposition. Only in SaudiArabia and Kuwaithas this optionbeen implementedon a large scale. In these two countries,more girlsthan boysgraduate from secondaryschools, and they are closeto achievingthe samelevel of participationas males in their strictly gender-segregatedcolleges (El-Sanabary 1988). The successof Kuwaitand SaudiArabia stemsfrom their oil wealth,which counterbalances cultural values and gendersegregation. It is probablyfair to saythat the issue in most countriesis not whetherschools are single-sexor coeducationalbut rather whetherschools existat all. Manyof the girls'schools in the poorer of these countriesare not onlyinferior in generalphysical condition but also lack specializededucational and recreationalfacilities such as libraries,laboratories, boarding options,and cafeterias. Weli-supervisedboarding facilities,for instance,are especiallyimportant in attractinggirls from remote areas. As part of its massiveeffort to universalizeprimary education in the early 1970s,the governmentof Turkey set up regionalboarding facilitiesfor children from sparsely populatedareas. This measure shouldbe effectivewith girlsin secondaryschools and colleges,although not with youngerones. Boardingfacilities are essentialat teachertraining and nursinginstitutes. Single-sexschools frequently offer a narrowerrange of curriculumoptions, which usually varies from school to school. The offeringsdepend in large part on the availabilityof same-genderteachers with training in variousfields. Becausegirls and boys alike have to competefor the same national examinations,the possibilityof curriculumchoice along gender lines is reduced. Weste- researzhhas documentedthat girls in single-sexschools are more likelyto make less stereotypicalcare hoicesbecause of the emphasison competitionand excellencein all subjectsand identificationwith stro.&role modelssuch as femalemath and scienceteachers and femaleprincipals. In MiddleEastern countries,girls who choosescience subjects have an equal or better than equal chance of successin these subjectsto that of boys. Girlshave consistently achieved higher examination grades than boys, especiallyon the General SecondaryEducation Examination in severalAi b countries,including Jordan,Saudi Arabia and Kuwait.In Jordan,a studyof two rural secondaryschools showed that 56 percent of the girls and 46 percentof the boyspassed the GeneralSecondary Education Exnamination, results that reflectedthe scoresfor the wholecountry in 1983(Layne 1984). In SaudiArabia, a surveyof 1,992students, 222 teachersand 36 principalsdocumented that girlsoutperformed boys academically in both scienceand literary subjects (Endargeeri 1986). Similar results were obtained in Kuwait,where girls attained significantlysuperior results compared with boys in all sciencesubjects but lowerresults in mathematics(Al- Methenand WiLkenson1988). Among hypotheses offered to explainthis outcomeare: 1) Girls havemore time to study than boys, who have more personal freedom; 2) Girls need to excel in these national examinationsin order to gain accessto universities;and 3) Girls who make it to secondaryschools are Middle East and North Africa 131

generally a more select group than boys. On average, they would tend to have better family background and greater abilitythan the average male student.

As in other areas of research on women's education in Islamic countries generally, the effects of coeducation and single-sexschools on achievementare terra incgnlita. A study of the effect of single-sex schools on fear of success and sex-role orientation among a sample of 626 Jordanian community college students found no difference in achievement motivation between males and females. The students in coeducationalinstitutions, however, revealed more sex-typedorientations (Al-Nhar 1986). One study of the education of Muslim girls in Ghana concurred with the more recent fndings in Western nations: teachers tended to ignore girls in coeducationalschools and put them in the back of the classrooms,where they kept quiet.

Because the effects of single-sexschools on curriculum choice and achievement appear to be positive, no valid reason exists to question their academic viability. Instead, the chaDlengeis to ensure quality in both girls' and boys' schools.

Sterotypes in School Tatbooks

Much has been written about sex-role stereotypes in school textbooks and their effect on educational and occupational aspirations and choices. Two studies on this subject, one in Arab countries (Abu Nasr, Lorfing, and Mikati 1983)and another in Egypt (Mohamed 1985),corroboracted the many research findings with a slight difference.7 In Arabic textbooks, the most important roles cited for women were mother and littlc girL while the role of wife was mentioned less often and mostly in the Saudi Arabian textbooks. Women in other roles (such as widows,older daughters, fighters, dream woman, fantasy woman, princess, historical figure, slave, witch, aunt, cousin, and neighbor) figured in less than ten percent of aUlroles mentioned. Working women were portrayed as engaged in traditional feminine occupations--forexample, teaching, nursing, pediatrics, dressmaking,administration of girls' schools (headmistresses),textile factory work, domestic help, and agriculturalwork. Reference to working women was most frequent (22 percent) in the People's Democratic Republic of Yemen, where they were presented as shouldering the dual burden of domestic and agricultural work. In Egypt and Tunisia, women as workers made up 16 percent and 11 percent, respectively,of the references. Although the Tunisian textbooks referred to mothers engaged in income-generating activities in cases of pressing economic need, the divisionof labor was clear: mothers were baking cakes at home for their sons to seoi in the market.

The textbooks stressed a number of desired "feminine"traits, includingweakness, sensitivity, submissiveness, dependency, and self-sacrifice. They were seen as deriving their identity and status from conformity to gender-based role expectationsas mother, dutifulwife, and obedient daughter. Mohamed (1985)concluded from a study of Arabic elementary reading textbooks that they did not reflect the changingroles of women in Egyptian society, nor were women present in equal proportion to their numbers in the population (they were only 20 percent of the characters portrayed in these books). These textbooks perpetuate the predominant sex-role stereotypes and potentially limit girls' educational and occupational aspirations and choices.

Avaiabity and Qwaliy of Teacers

Where schools are gender-segregated, the availabilityand quality of female teachers are key issues that affect female enroDlment,achievement, and persistence. Despite their predominance in teacher training institutes (ITl) and teachers' colleges in the countries studied, Middle Eastern women are a minority in

7 The Arab country study was conducted for UNESCO by researchers from the Institute of Women's Studies in Arab Countries, located in Lebanon. The researchers analyzeda sample of seventy-nineArabic readers, fift-two of which were from Lebanon and the rest from Egypt, Tunisia, Kuwait, Saudi Arabia, Qatar, and the People's Democratic Republic of Yemen. All the textbooks reviewed were primary-level except the Lebanese ones, which included intermediate-levelbooks. 132 Middle East and North Africa

their teaching force, especially at the postprimary level (table 4.10). The number of female primary and secondary school teachers grew substantiallybetween 1975 and 1987. At the primary level, their numbers more than tripled in Tunisia, Morocco, and Saudi Arabia, more than doubled in Jordan, Kuwait,and Egypt. Turkey, which by 1987had achieveduniversal primary enrollment, showed a only modest increase in female teachers. Despite this progress, only in Jordan and Kuwait were female primary teachers a majority, and only ;n Kuwait did women make up 52 percent of secondary tcachers, with Jordan a close second at 49 percent. Despite the narrowing gender gap in the teaching profession, women were just about one-third of all secondary teachers in the remaining five countries in 1987.

Table 4.10 Number and Percentage of Female lfachers at the Primary and Secondary Levels 1975 and 1987

PrimarvEducation SecondarvEducation 1975 1987 1975 1987 Female % of Female % of Female % of Female % of Country Teachers Total Teachers Total Teachers Total Teachers Total

Egypt 56,323 48 115,949 49 20,962 27 85,271 34 Morocco 7,725 21 28,791 35 na na 22,091 29 Tunisia 5,597 24 18,180 42 2,475 28 8,457 30 Jordan 5,719 51 12,860 67 3,027 39 10,558 49 Kuwait 3,471 55 6,950 69 4,564 49 10,340 52 Saudi Arabia 10,568 31 41,235 46 3,360 24 19,579 43 Turkey 86,205 41 91,870 42 41,414 35 53,630 37 Total 175,608 39 315,835 50 75,802 34 209,926 39 na = not available. Note: Data for Morocco (secondary education) is for 1988 and for Egypt (both levels) 1986. Source: UNESCO StatisticalYearbook (1990).

The same factors that depress women's educational access have tended to restrict female access to and persistence in teaching. These factors vary from one country to another and from one region to another in the same country. They include cultural attitudes; financial constraints; increasing devaluation of the teaching profession, especiallyat the primary level; difficultyin recruiting and retaining female teachers in rural and sparsely populated areas; lack of mobilityof women because of family responsibilitiesor cultural considerations; and the added difficultiesof married women teachers. Despite the importance of female teachers for the quantitative and qualitativedevelopment of girls' education, no research has addressed the factors that influence supply or the problems these teachers and their schools encounter because of their special needs and circumstances.

In most Middle Eastern and North African countries, the marked educational expansionover the past few decades has not been matched by sustained measures to train teachers and administrators. Most countries have teacher shortages, especially in math and science, foreign languages, and even native languages, resulting in overcrewdedclassrooms and high pupil/teacher ratios, reliance on male teachers in girls' schools Middle East and North Africa 133

in some countries, or simply having classeswith no teachers. The nature and intensity of the problem vary across countries. North African and Arabian Gulf states suffer particularly from shortages of indigenous teachers. In Morocco, for instance, replacing European teachers with Moroccans, especially at the upper secondary level,has been a challengesince independence. In 197-78, the ratio of Moroccans to all women teachers was less than one-third at the secondary level and two-thirds at the intermediate level (Cameron 1983).

Heavy reliance on expatriate teachers is a problem in all oil-exporting Gulf states such as Kuwait and Saudi Arabia (Meleis, El-Sanabary and Beeson 1979; El-Sanabary 1988). In 1979-80,in Saudi Arabia, only 15 percent of all female teachers at the intermediate and upper secondary levels were Saudis. These figures had been expected to rise to 46 percent and 42 percent, respectively,by 1984-85,but whether this actually haprened is not known (Kingdom of Saudi Arabia 1980). "Saudization"of primary teaching staff in girls' schoolswas realized by the mid-1980s, except in some rural areas where the output of local teacher training institutes has been insufficientto meet local needs and recruiting urban women to work in rural areas has been difficult. At the intermediate and secondarylevels, Saudi female teachers are in short supply because the teachers colleges do not produce enough graduates to mnuetthe demand, especiallyfor foreign language and science teachers. At the same time, thousands of women graduate from colleges and universities with specializationsthat do not match existing needs. Reliance on expatriate teachers who do not understand local cultural traditions may hurt educational quality because of communicationproblems between teachers and students or because of the tendency of expatriates to compromiseeducational quality for the sake of their own work security (Birks and Rimmer 1984).

specialPblem of Fmal Teahe

T'he need to juggle the double burden of work and domestic responsibilitiesaccounts for the reported high rate of absenteeism and dropout (in SaudiArabia and Jordan, for instance) and reduced productivityamong female teachers. For example, Rumaihi (1986) quoted a Saudi government report issued in 1981as saying that an average of three women a day, both Saudi nationals and contract teachers, left their jobs in primary education. Where child care is unavailable or inadequate, most women quit to look after their children. Little has been done to address this problem, and the measures that have been undertaken may be creating new and more serious problems. Before the 1950s,only unmarried women could become teachers in Egypt. The government eventually abolished that policy and also issued maternity benefits. In Jordan, because of the high absenteeism of married teachers for maternity leave and other family reasons and because of the difficultyof finding substitutes, especiallyin remote areas, the Ministryof Education in 1965prohibited the hiring of married females as teachers and pressured those already teaching to resign (Al-Tall 1979). This measure only exacerbated teacher shortages, and the regulation was canceled in 1967. Solving a serious problem by discriminatorymeasures will worsen the teacher shortage, violate the human rights of married teachers, and discourage girls from enrolling in teacher education programs.

Most countries provide maternity leave and other benefits for childbirth and childrearing. Egypt, ;vr instance, allows a working mother up to six years of leave (no more than two years at one time) for child care, with no loss of seniority and other benefits. What proportion of female teachers in Egypt take advantage of these benefits and what provisionsthe Egyptiangovernment makes for substitute teachers are unknown. In SaudiArabia, female teacher absenteeism and drop-out from teacher education programs are common, but no satisfactory solutions have been found. The Saudi government gives female teachers a six-weekmaternity leave that may be combined with a six-month sick leave upon the recommendation of a physician. It also permits female teachers to resign for marriage or child-rearing reasons without giving six-months advance notice, as is required of all other employees. This policy wastes the slender supply of female teachers and aggravatesthe continuingreliance on foreigners. Several Saudiwriters have suggested more equitable solutions such as part-time teaching,job-sharing, and expansionof on-site child care. Some women advocate a longer maternity leave, similar to that in effect in Egypt (El-Sanabary 1987). 134 MiddleEast and North Africa Teacher/Stdent Irniction Recenteducation studies have highlighted the importanceof teacher-studentinteraction to pupilmotivation, achievement,and persistence.Teachers often perpetuate sex-role stereotypes directly and indirectly through whatthey teach, their behavior,their interactionswith pupils, and their assumptionsabout the differentskills and abilitiesof girls and boys (Whyte1986). Femaleteachers may inspiregirls to high achievementand accomplishmentor direct themtoward conformity with prevailing domestic ideals. Havingacademically and pedagogicallyqualified female teachers is importantbut not sufficient;they must alsounderstand sex-role stereotypes and their potentialeffects. Femaleteachers, if adequatelytrained, can identifygirls at risk of droppingout and providethe specialcare and encouragementneeded to keep them in school.

Femake Admins0utos Female administratorsplay a keyrole in girls' schoolsin all MiddleEastern countries,except in Tunisia, where they are very few. Throughmore efficientmanagement and leadershipand support of female teachers and students,they can help improvethe qualityof girls' education. Unfortunately,many are unpreparedto deal with the problemsof absenteeism,drop-out, and lack of motivationthat particularly affectfemale students and teachers. Withproper trainingand support,female administrators could play an importantrole in deali , with these problems,thereby reducingattrition and improvingeducational quality. They can be sensitizedthrough printed materialsand in-servicetraining about these issues. Femaleschool administrators, like their male counterparts,receive very little,if any, preservicetraining, are often overworkedand underpaid;and lack adequateresources. Manyare lessqualified academically tha'l their male colleagues.In Jordan, for example,female principals in the mid-1970swere more likely than men to be graduatesof secondaryschools or teacher traininginstitutes and lesslikely to have had a collegedegree. Thesedifferences reflect the relativeproportions of men andwomen who continuebeyond secondaryeducation and the different qualificationsof teachers who are promoted to administrative positions.As withteachers, women principals are lesslikely to participatein in-serviceprograms than are their male counterparts,who are more likelyto learn of these programsand take advantageof them becauseof men's networksand greater accessto resources.

OtherSchool Factors No studiesare availableof other school-relatedfactors that mightaffect the state of women'seducation. Most schoolsin rural areas lack basic instructionalmaterials, even a chalkboardand chalk. Textbooks are unavailableor inefficientlyhandled. In Egypt,where textbooksare distributedannually to students free of charge, manyultimately end up as wrappingpaper in grocerystores. Audiovisualmaterials are unheardof, exceptin someschools in wealthycountries such as Kuwaitand SaudiArabia. Ironically,with externalassistance, modern educational technology such as computersis beingintroduced on a limitedscale in some rural schools. Suchtechnology is not accessibleor feasiblefor most schools,however. What is needed are simple,inexpensive, and readilyavailable media such as radio,posters, film,and instructional toys that can be producedeasily and cheaplylocally. In the classroom,the emphasison rote learningand theoreticalconstructs with no! elevanceto the realities of students' lives detracts from th&appeal of attendingschool. This educationalapproach targets the hardworkingand highly motivated students, ignores individual differences, and disregards children who need special assistanceand support. Those most likelyto suffer are girls who are poorlymotivated and malnourishedor who have other demandson their time. The congestionin classroomsin most countries and the poor qualificationsof manyteachers, however, may leaveno alternativeto memorizationof facts to be recalledduring poorly conceived and writtenexaminations. Another deterrent to effectiveeducation is automaticpromotion, which permits students to move up withouthaving acquired basic skills. Somecountries have adoptedthis policyto reduce repetitionat the MiddleEast and NorthAfrica 135 primary level. Schoolingunder these circumstancesbecomes an exercisein futility,and discouraged childrenare then withdrawnfrom school. Most of the countriesthat have this policyalso have a serious student attritionproblem. Lack of educationalguidance and counselingservices in the schoolsoften leadsto attritionor premature specialization.As UNESCO(1983) noted, guidance is frequentlylacking or is providedtoo late to influence pupils'attitudes, educational and careerdecisions. Consequently, students and their parentsmake decisions basedon prevailingperceptions of women'sroles and limitedknowledge of educationaland careeroptions. The result couldbe lossof human talentand resources.Furthermore, female students at risk of dropping out go undetectedand are permanentlylost to the system.

ConcludingRemarks Despite the considerableprogress by Middle Eastern and North African countriesin improvingfemale accessand educationalattainment, major problemsremain and call for expedientsolutions. The studies reviewedhere illuminatethe issues related to femaleeducation and its socio-economiccontext. They highlightthe effects of economicdevelopment and income distribution;religious and cultural values; demographicpatterns; sex-role stereotypesand the divisionof labor in the home and marketplace; educationalpolicies; socio-economic background of parents and their educationallevel; and finallythe availabilityand quality of schools and their physicalcharacteristics and instructionalprograms. This research,however, does not weighthe relativesignificance of the variousdeterminants or their scientific validity. Most MiddleEastern and North Africangovernments recognize their enormouseducational problems, includinggender disparities, and have undertaken various educational reforms--but they face tough decisions becauseof inadequateresources. Ironically,they often pursueWestern models that haveproved unsuited to local needs and circumstances,sometimes with unintendedresults. For instance,the adoption of automaticpromotion policies, as noted earlier, has led to deteriorationof achievement;and the extension of compulsoryeducation to eight years before achievinguniversal access to the six-yearcycle has meant more educationfor some and less or none for others,mostly females. Very few of the educationalreforms, if any, are specificallygeared to women. Generally,progress in femaleeducation has been a by-productof educationalexpansion. Overall educationalpolicies have had mixedresults for femaleeducation. The reforms,of whichfemales have been the unintendedbenefciaries, have includedexpansion of educationalfacilities in rural areas, albeit slowly,diversification of school curriculaand increasedaccess to technicaland vocationaleducation (more so in some countries than others);school reorganization, including emphasis on basicprimary education and the introductionof some comprehensivehigh schools that teachmarketable skills; improvements in teachingmethods and curriculum relevance;upgrading the levelof primaryschool teacher training from the secondaryto the junior college level,and provisionfor somein-service training for teachersand administrators.Much more stillneeds to be done to enhancegender equity in educationto benefitnot onlywomen but the wholeof society. Educationalreform modelsand patternswill depend o.i localcircumstances and needs. Each countrywill haveto clarifyits goalsand policies for femaleeducatiot, and set prioritiesfor educationalobjectives. Poor countrieswill need extensiveexternal financial assistance to achievethese objectives,while richer countries may need consultationin identifyingproblems and mappingout solutions. Regionaland international cooperationmay prove very effective. 136 Middle East and North Africa References

Abu Nasr, J., I. Lorfing, and J. Mikati, Identificationand Elimination of Sex Stereotypesin and from School Textbooks:Some Suggestionsfor Action in the Arab World. Paris: UNESCO, 1983. Abu Zahra, N. "On the Modesty of Arab Muslim Villages:A Reply."American Anthropologist(Lebanon), (72), 1970. Ahdab-Yehia, M. "Women,Employment and FertilityTrends in the Arab Middle East and North Africa." In The Fertilityof WorkingWomen: A Synthesis of Intemal Research, Stanley Kupinsky (ed.). New York and London: Praeger (Praeger Special Studies in International Economics and Development), 1984.

Al-Kadhi, A.B. "Women's Education and Its Relationship to Fertility: Report from Baghdad." In Women and the Family in the Middle East, Elizabeth W. Fernea (ed.). Austin, Tex.: University of Texas Press, 1985. Allaghi P., and A. Almana, "Surveyof Research on Women in the Arab Gulf Region."In Social Science Research on Women in the Arab Countries,UNESCO, pp. 14-40. London and Dover, N.H.: Francis Pinter Publishers, 1984.

Allman, J. (ed). Women's Status and Fertilityin the Muslim World. New York and London: Praeger, 1978. Social-.....--. Mobility, Education and Developmentin Tunisia. Leiden, The Netherlands: EJ. Brill, 1979. Al-Methen, A.E., and WJ. Wilkenson, "In Support of a SociologicalExplanation of Sex Differences in Science and Mathematics Achievement: Evidence from a Kuwaiti Study of Secondary School Certificate ExaminatioDs."Research in Science and TechnologicalEducation, 6(1):91-101,1988. Al-Nhar, T. "Coeducationalversus Single-SexSchooling, Sex-Role Orientation and Fear of Success:Results from Jordan." Ph.D. dissertation. Universityof Pittsburgh, Pittsburgh, 1986. Al-Nouri, Q.N. "Modern Professionalismin Libya: Attitudes of University Students,"Intemational Social ScienceJournal, 27(4): 691-702,1975. Alrabaa, S. "Sex Division of Labour in Syrian School Textbooks,"Intemational Review of Education, 31(3): 335-48, 1985. Al-Tall, A.Y. Education in Jordan. Islamabad: National Book Foundation, 1979. Al-Thaqwub,F.T. "The Attitude of Kuwaitis Toward Women in Contemporary Society."In Studies of the Status of Women in Kuwait and the Arab Gulf. Proceedings of the Regional Conference on Women in Contemporary Society, Kuwait, April 1975. Kuwait: The Women's Cultural and Social Organization, 1975. Al-Thumali, SA. "Family, Peer, and School Influences on Saudi Arabian High School Seniors' Undergraduate College Plans and Academic Majors." Ed.D. dissertation. University of Houston, Houston, 1984. Antoun, R. "On the Modesty of Women in Arab Moslem Villages: A Study in the Accommodation of Tradition."American Anthropologist(Lebanon), 70(4), 1968. Arab Bureau of Education in the Gulf States.A Study of EducationAttrition in the Arab Gulf States.Riyadh: Arab Bureau of EDucation in the Gulf States, 1983. Middle East and North Africa 137

Azzam, H.T. "The Participation of Arab Women in the Labour Force: Development Factors and Policies." World Employment Programme Research, Population and Labour Policies Programme, Working Paper No. 80. Geneva: International Labour Organisation, 1979. Azzam, H.; J A. Nasr, and I. Lorfing, "An Overview of Arab Women in Population, Employment and Economic Development."In Women, Employment and Developmentin the Arab Worl4 Julinda Abu Nasr, N. Khoury and H. Azzam (eds.), Chapter 1, pp. 5-37. Berlin, New York and Amsterdam: Mouton Publishers, 1985.

Bach, R., S. Gadallah, H.A.S. Khattab, and J. Gulick, "Mother's Influence on Daughters' Orientations Toward Education: An EwyptianCase Study."Comparative Education Review, 29(3), 1985. Balderston, J., md others. Malnourished Childrenof the Rural Poor: The Web of Food, Health, Education, Fertilityand AgriculturalProduction. Boston, Mass.: Auburn House PublishingCompany, 1981. Basson, P. "Women and Traditional Food Technologies:Changes in Rural Jordan." Ecology of Food and Nutrition (11) (May), 1981. "Domestic--.-- Productivity in Male and Female Headed Households of Rural Jordan."Ecology of Food and Nutrition, 12:75-78,1982. Benson, C.S. (principal investigator). Education and Nutrition:Perfomnance and Policy. A University of California at Berkeley project funded by the USAID. Berkeley: University of California, 1980. Birks, J.S., and JA. Rimmer, "DevelopingEducation Systems in the Oil States of Arabia: Conflicts of Purpose and Focus." University of Durham Occasional Papers Series No. 21, Manpower and Migration Series No. 3. Durham, England: Center for Middle Eastern and Islamic Studies, 1984. Bowman, MJ., and A.C. Anderson, "The Participation of Women in Education in the Third World." ComparativeEducation Review, (Special Issue): S13-S32,1980.

Bremer, MA., and L. Pauli, Wastagein Education:A WorldProblem. Geneva: UNESCO, 1971. Cameron, J. Educational Handbook of Educational Systems, Section B: Nofth Africa and the Middle East. Chichester, New York, Brisbane, Toronto, Singapore: John Wiley and Sons, 1983.

Chamie, M. "National, Institutional and Household Factors Affecting Young Girl's School Attendance in DevelopingSocieties." The International Center for Research on Women and the Office of Human Resources, U.S. Agency for International Development, Washington, D.C., 1983 Cochran, J. Education in Egypt. London, Sydney and Do ver, New Hampshire: Croom Helm, 1986.

Cochrane, S. Fertilityand Education: WhatDo WeReally Know? Baltimore and London: The John Hopkins University Press for the World Bank, 1979. Cochrane, S. H., K. Mehra, and I.T. Osheba, "The Educational Participation of Egyptian Children." EDT Discussion Paper No. 45. Waskington, D.C.: World Bank, December, 1986. Coombs, P.H. 7he World Crisisin Education: The Viewfrom the Eighties.New York and Oxford: Oxford University Press, 1985. Deble, I. The School Education of Girls. An International Comparative Study on School Wastage Among Girls and Boys at the First and Second Levels of Education. Paris: UNESCO, 1980. Doan, R.M. "Class and Family Structure: A Study of Child Nutritional Status in Four Urban Settlements in Amman, Jordan." Ph.D. dissertation. Ithaca, New York: Cornell University,1980. 138 Middle East and North Africa

Egypt. Central Agency for Public Mobilisation and Statistics and World Fertility Survey. International S.atistics Institute. Egyptia Fertility Suivey 1980. Vol. m: Socio-Economic Differentials and Comparative Data from Husbands and Wives. Cairo: Central Agency for Public Mobilisation and Statistics and World Fertility Survey, 1983. El-Ghannam,M. Educaion in the Arab Region Viewedfrom the 1980 MarrakeshConference. Educational Studiesand Documents.Paris: UNESCO, 1970. El-Guindi, F. "Veiling Infitah with Muslim Ethics: Egypt's Contemporary Islamic Movement." Social Problems, 28(4) (April), 1981. El-Sanabary, N. "A Comparative Study of the Disparities of Education Opportunities for Girls in Arab Countries." Ph.D. dissertation, University of California, Berkeley, 1973. ------"Female Education and Manpower Needs of Arab Countries." In Issues in Human Resource Developmentin Arab Countres. Kuwait: National Ministry of Culture. In Arabic, 1978. 'Access of Girls and Women to Technical and Vocational Education in Arab Countries." In Women and Work in the Third Worldt The Impact of Industrializationand Global Economic Interdependence,Nagat El-Sanabary (ed.). Proceedings of two conferences at the University of California, Berkeley, May 1982 and April 1983. Berkeley, California: University of California- Berkeley, Center for the Study, Education and Advancement of Women, 1983. =------Continuityand Changes in Women's Education in Arab Countries."In Women and the Fanily in the Middle East, Elizabeth W. Fernea (ed.). Austin, Texas: University of Texas Press, 1985. ------. "Women's Education and Work for Children and Development." A paper prepared for the Conference on Childhood and Development, Riyadh. Ministry of Planning, Riyadh, 1987. "Educating------. the Second Sex Saudi Arabia's Educational Policyfor Women and Its Implication." A paper presented at the Annual Conference of the Comparative and International Education Society, Western Regional Conference, Sacramento, October, 1988. Endargeeri M.S. "The Impact of Selected Social and Familial Factors on the Academic Achievement of Female Students in Saudi Arabia." Ed.D. dissertation. Universityof Southern California, 1986.

Finn, JD., L Dulberg, and J. Reis, "Sex Differences in Educational Attainment: A Cross-National Perspective."Harvard Education Rc=,w, 49 (November), 1979. Fimn,JD., J. Reis, and L Dulberg, "SexDifferences in Educational Attainment: The Process."Comparatve Education Review, 24(2, Part 2) (June):S33-S52,1980. Fuller, B. "What Factors Raise Achievement in the Third World."Review of Educational Research,57(3) (Fall):255-92,1987. Haddad, W.D. "RegionalReview of Alternative Modes of Vocational Training and Technical Education." Educational Training Series, Education and Manpower Development Division,Europe, Middle East and North Africa Region. Washington, D.C.: World Bank, December 1986. Hallouda, AM., SZ. Amin, S. Cocbrane, and Farid, S.M. The Egptian FertilitySurvey 1980: Socio- Economic Differentialsand Comparatve Datafvm Husbands and Wives.Cairo: Central Agency for Public Mobilisation and Statistics, 1983. Hammam, M. "Labor Migration and the Sexual Divisionof Labor."MERIPReports, 95 (March/April), 1981. Middle East and North Africa 139

------. "CapitalistDevelopment, Family Division of Labor,and Migration in the Middle East." In Women's Work Developmentand the Divisionof Labor, Eleanor Leacock and Helen I. Safa (eds.), pp. L58- 73. Boston: Bergin and Garvey Publishers Inc, 1986. Harbison, F., and CA Myers, Education, Manpower and Economic Growth. New York, Toronto and London: McGraw Hill Book Co., 1964.

Hartley, MJ., and E.V. Swanson, "Retention of Basic Skills Among Dropouts from Egyptian Primary Schools."Discussion Paper. Education and Training Series. Report No. EDT40. Washington, D.C.: World Bank, December, 1986. Hatem, M. 'Women and Work in the Middle East: The Regional Impact of Migration to the Oil-Producing States." In Women and Woik in the 7hird World The Impact of Industrializationand Global Economic mnterdependence,Nagat El-Sanabary (ed.), pp. 13140. Proceedings of two conferences at the Universityof California,Berkeley, May 1982and April 1933.Berkeley, Calif.: University of California- Berkeley, Center for the Study, Education and Advancementof Women, 1983.

Howard-Merriam, K "Women,Education, and the Professions in Egypt." ComparativeEducation Review, 23(2)(June), 1979. ------"Egypt'sOther Political Elite." Westem Political Quwterl 34, 1981.

Jones, M.T. "Of Girls in Tunisia: PolicyImplications of the Drive for Universal Enrollment." Comparative Education Review, (June), 1980. Kaneko, M. The Educatonal Compositionof the World'sPopulaton: A Database. Education and Training Division. WashiDgton,D.C.: World Bank, May 1987.

Khattab, HAS. "Female Education in Egypt: Changing Attitudes Over 100 Years." In Muslim Women, edited by Freda Hussain. New York: St. Martin Press, 1984. Kingdom of Saudi Arabia. Ministry of Planning. The Third DevelopmentPlan, 1980-1985.Riyadh: Ministry of Planning, 1980. -- The-. Fourth DevelopmentPlan, 1985-1990.Riyadh: Ministry of Planning, 1985.

Klein, SS. "The Role of Public Policyin the Education of Girls and Women."Educational Evaluton and Poliy Analysis, 9(3) (Fall):219-30,1987.

Larson, B.K. "Women's Work and Status: Rural Egypt and Tunisia Compared."A paper presented at the Middle East Studies Association Meeting, Los Angeles, California, November 5, 1988.

Layne, L "Educational and Social Hierarchy in Rural Jordan." A paper presented at the Annual Meeting of the American AnthropologicalAssociation, Chicago, 1984. LeVne, RA. "Influencesof Women's Schooling on Maternal Behavior in the Third World." Comparatve Education Review, 24(2, Part 2):78-105,1980. "Women's Schooling, Patterns of Fertility and Child Survival." Educational Researcher, 16(9)(December), 1987. Lorfing, I., and JA Nasr, "Sex-RoleOrientation dAArab UniversityStudents." In Women, Employment and Development Julinda Abu Nasr, et al. (eds.), pp. 131-43.New York: Mouton Publishers, 1985. Maher, V. Women and Propenyin Morocco:Their ChangingRelation to the Processof Social Straation in the Middle Atlas. Cambridge Studies in Social Anthropology, No. 10. London, New York Cambndge University Press, 1974. 140 Middle East and North Africa

"Women-.....-. and Social Change in Morocco."In Women in the Muslim World, Lois Beck and Nikkie Keddie (eds.), Chapter 4. Cambridge, Massachusetts: Harvard University Press, 1978. Marshall, S.E. "Politics and Female Status in North Africa: A Reconsideration of Development Theory," Economic Developmentand CulturalChange, Vol. 32, No. 3, pp. 499-524,April 1984.

Massialas,B., and S. Jarrar, Educationin theArab World.New York: Praeger, Praeger Special Studies,1983. Meleis, A.I., N. El-Sanabary, and D. Beeson, 'Women, Modernization, and Education in Kuwait." ComparativeEducation Review, (February):115-24,1979. Mernissi F. Beyond the Veil: Male-Female Dynamics in Modem Muslim Society. Revised edition. Bloomington, Indiana: Indiana University Press, 1987. Mohamed, F.E. "Sex-Role Stereotyping in Arabic Elementary Reading Textbooks in Egypt." Ph.D. dissertation, Jniversity of Pittsburgh, Pittsburgh, 1985. Moore, C.H. Images of Development:Egyptian Engineers in Search of Identity. Cambridge, Massachusetts: The MIT Press, 1980. Nagi, M.H., and E.G. Stockwell,"Muslim Fertility: Recent Trends and Future Outlook."Joumal of South Asian and Middle Eastem Studies, VI(2)(Winter):48-71,1982. Nelson, C. "ChangingRoles of Men and Women: Illustrationsfrom Egypt."Anthropological Quarterly, 41(2) (April), 1968. OECD. Girls and Women in Education. Paris: OECD, 1986. Othman, A. Observationson the CulturalAttitudes Toward the Education of Women in Arab Society, Considerationsfor Planning. Sirs el-Liyyan, Egypt: UNESCO, Arab States Training Center for Education for Community Development (ASFEC). In Arabic, 1964. Papanek, H. "Class and Gender in Education-Employment Linkages." ComparativeEducation Review, (August):317-46,1985.

Patai, R. The Arab Mind New York: Charles Scribner and Sons, 1973. Psacharopoulos,G. "How Equal Is Free Education?"Comparative Education Review, 21(1) (February), 1977.

Psacharopoulos,G., and M. Woodhall,Education for Development:An Analysis of Investment Choices.New York: Oxford University Press for the World Bank, 1985. Rihani, M. "Women and Work in Morocco." In Women and Work in the Third World: The Impact of Industrializationand GlobalEconomic Interdependence,Nagat El-Sanabary(ed.). Proceedings of two conferences at the Universityof California,Berkeley, May 1982and April 1983. Berkeley,California: University of California-Berkeley,Center for the Study, Education and Advancement of Women, 1983. Rosen, B.C. "Family Structure and Achievement Motivation." American Sociological Review, 26 (August):574-85,1961. Rumaiha, M. Beyond Oil-Unity and Developmentin the Gulf, translated by James Dickins, p.122. London: Al Saqi Books, 1986. Middle East and North Africa 141

Saadawi, N. "Growing Up Female in Egypt." In Women and the Family in the Middle East, Flizabeth Warnock Fernea (ed.). Austin, Texas: University of Texas Press. Translated frowmthe Arabic by Fedwa Malti-Douglas, 1985. Safilios-Rothschild,C. "SexDifferences in Early Socializationand Upbringing and Their Consequences for Educational Choices and Outcomes." In Girlsand Women in Education:A Cross-NationalStudy of Sez Inequalitiesin Upbringingand in Schools and Colleges.Paris: OECD, 1986. Silfiman, J.M. "Basic Education for Women: The Promise Fulfilled? A Comparative Analysis of Five Countries (Nepal Egypt, Indonesia, Tanzania, Peru)." E&D. dissertation. Columbia University Teachers Colleg., New York, 1987. Simmons, J., and L. Alexander, "Determinants of School Achievement in DevelopingCountries: A Review of the Research." Economic Developmentand CulturalChange, 26:341-57,1978. Smock, A.C. Women's Educationin DevelopingCountries: Opportunities and Outcomes. New York: Praeger, Praeger Special Studies in Comparative Education, 1981. Smock, A.C., and N.H. Youssef, "Egypt: From Seclusion to Limited Participation." In Women:Roles and Status in Eight Countries,edited by Janet Zollinger and Audrey Chapman Smock, Chapter 2. New York and London: John Wiley and Sons, 1977. Stuart, M.F. "Developing Labor Resources in the Arab World: Labor Activity Effects from School Attendance and Socio-EconomicBackground Among Women in the East Jordan Valley."Ph.D. dissertation, University of S -uthern California, 1981 Szyliowicz,J.S. Education and Modemization in the Middle East. Ithaca, New York and London: Cornell University Press, 1973. Todaro, M.O. "Educationand Development."In Economic Developmentin the Third World (third edition), Chapter 11. London and New York: Longman, 1985. UNESCO. Access of Girls to School Education in the Arab States. General background document prepared by the UNESCO Secretariat for the Meeting of the Experts on the Access of Girls to School Education in the Arab States, Algiers, Algeria, April 6-16, 1964. UNESCO/ED/FAR/3. Paris: UNESCO, February 25, 1964. ------. Access of Girlsto SecondaryEducation. Paris: UNESCO, 1966. Access of Girls to Technicaland VocationalEducation in Arab Countries.A working document submitted by the UNESCO Secretariat for the Meeting of the Experts on the Access of Girls and Women to Technical and Vocational Education in Arab Countries, November 1-7, 1969, Kuwait. ETF/AR/3. Paris: UNESCO, September 12, 1969. Equality---.----. of Educational Opportunityfor Girls and Women. Paris: UNESCO, 1983. ------.Statistical Yea,book. Paris: UNESCO, Various years.

Whyte, J. 'The Development of Sex Stereotyped Attitudes Among Girls and Boys: Different Models of Their Origin and Their Educational Implications."In Girls and Women in Educaion, OECD, pp. 57- 88. Paris: OECD, 1986. World Bank. Education: Se?ctorWorking Paper. Washington, D.C.: World Bank, 1974. ---. "Education in Sub-SaharanAfrica: Policy for Adjustment, Revitalizationand Expansion."World Bank Policy Study. Washington, D.C.: World Bank, 1987. 142 Middle East and North Africa

------. World DevelopmentReport 1989.Washington, D.C.: World Bank, 1989. Youssef, N.H. Women and Work in DevelopingSocieties. Population Monograph Series No. 15. Berkeley, California: University of California, 1974. ------. Education and Female Modernism in the Muslim World." Joumal of Intemational Affairs, 30(2):191-209,1976-77. 'The Status and Fertility Patterns of Muslim Women." In Women in the Muslim Word4 Nikki Keddie and Lois Beck (ed.). Cambridge, Mass., and London: Cambridge University Press, 1978a. ------. "Women and Their Professional Future: Assessment of Training Needs and Training Programs in Morocco." Office of Women in Development,USAID, June. Mimeo. 1978b. Latin America 143 Chapter 5. Latin America

Ines Bustillo

Girls and women in most Latin American countries, unlike other Third World regions, are not disadvantagedin educationcompared with boys and men. However,inequities remain in a few countries plaguedby pervasivepoverty, in pocketsof the educationsector, particularlyat the higher levels,and in rural areas of some countrieswhere indigenous Indian populationsare not integratedinto the education systembecause of povertyor linguisticdiscrimination. For example,in Haiti whereannual income averages $360per person, illiteracylevels exceed 62 percentfor malesand 67 percentfor females. In Bolivia,with a large, poor Indian populationand $580GNP per capita,24 percentof men and 49 percentof women cannot read. By contrast, in Argentina, which underwent early industrializationand educational developmentand has an incomeof $2,390per capita,less than 6 percentof the populationis illiterate. Althoughwomen trail men slightlyin abilityto read in nost LatinAmerican countries, illiteracy tells more about past educationdisparities by sex,while school paticipation and achievementrates of womencohorts are better measuresof the current situation. And exceptfor the abovementionedexceptions, girls of the region are as likelyto attend school more likelyto continueon in school and have higher achievement levelsthan boys. The biggestinfluence on women'seducation has been the remarkable expansionof Latin American educationalsystems since 1960, with gross school enrollment of girlsrising from 71 percentto 104.1percent in 1985. Educationhas been a highpriority in the regionand publicexpenditures on educationhave risen as a share of GNP in manycountries. Another factor in women'srelatively high education level is Latin America'slarge urbanpopulation share, more than 85 percentin Argentinaand Uruguay. The regionhas one of the highestrural-to-urban migration rates in the world. In manycountries, more womenthan men migrateto the cities(IDB 1980-81,p. 133),where schoolaccessibility and qualitytends to be higherthan in rural areas. In addition,expanded employment opportunities for womenin LatinAmerica--reflected in their growinglabor force participation--have increased incentives for women'seducation. The servicesector, whichis more open to hiringwomen than other sectorsof LatinAmerican economies, is growingas fast as the industrialsector, whichpredominantly hires men. The same situationexists in East Asia, whilein Sub-SaharanAfrica, industrial sector growth outpacesthtt of the servicesector. But these factorsalone cannotaccount for LatinAmerica's high female education level. Other developing regionsalso have undergone fast-paced educational expansion, urbanization, and growth in female workforce participationin the past 25 years,while their gendergaps remainsignificantly wider than LatinAmerica's. Additionalsocio-cultural forces have shaped Latin American attitudes about educating females. A possible explanationis that manyof the region'sindigenous Indian cultures had a relativelyegalitarian attitude towardwomen. Peruvian women during the Incaperiod, for example,participated in agriculturaland other productivework on equal terms withmen, and someof these socialpatterns can be seen in today'speasant communities.In fact, althoughmore femalesare illiteratein Peru than males,the gap is narrowerin rural agriculturalareas (Gill 1990). In the rural Bahiaarea of Brazil 30 percentof sonshave had no primary education compared with only 21 percent of daughters. Colonialrule and accompanyingChristian missionaryinfluences forced different economic,political and social structures on the region, and the Christianinfluences may have been relativelyless constrictingof women'sfreedom than, for example,the Moslemor Confucianinfluences on peoplesof other parts of the world. LatinAmerica's progress in expandingwomen's education and eliminatingthe gender gap in most places shouldnot leavepolicy makers complacent. Additional measures are neededto raiseliteracy and education levelsfurther acrossboth genders,and to eliminatethe pocketsof inequityfaced by womenand the rural poor. Indeed, female-headedhouseholds make up a growingshare of these poorer segmentsof Latin Americansocieties. 144 Latin America

This study reviewsthe literature on the iactors influencingeducational attainment and achievementof girls and women in Latin America and assessu. policies designed to improve their situation. It addresses three questions:

* What are the effects of the general expansion of education in Latin America on the schoolingof females? * What are the effects of supply factors, such as school characteristics, and of demand factors, such as family socioeconomiccharacteristics, on the educational achievement and attainment of females? * Which policies will be most effective in addressing the educational needs of girls and women in Latin America?

Any comprehensivestudy of educational policyand status in Latin America faces several difficulties. First, the data for some countries are limited. Cross-countrycomparisons are made whenever the data permit. A second problem is that most of the research on this topic has focusedon the determinants of enrollment in primary education, generally neglectingachievement or performance,and issues in the postprimary levels. Finally, most of the educational studies have considered either both sexes together. As a result, their conclusions and policy recommendations have been general and not directed toward women's education.

The Economic and Educational Setting Economic Conditions

The population of Latin America exceeded 400 million in 1987, 70 percent of it urban, due to heavy migration to citiesstemming in part from women's search for better socioeconomicopportunities. Table 5.1 presents selected indicators for countries of the region. Fertility rates have declined substantially for all countries, but current and projected rates remain high for many of these nations (Bolivia, Ecuador, Guatemala, and Honduras, for example). In Cuba, Chile, and Uruguay, however, total fertility rates com ire favorably to those of developed nations. One of the most outstanding changes in the status of women in Latin America is their increased participation in the labor force. Between 1950and 1980,their labor force participation rate grew faster than that of men. The share of women in the labor force rose from 17.9percent in 1950to 26.1 percent in 1980 and is projected to exceed 27 percent in the year 2000 (IDB 1987). Female employment continued to increase in the services sector, which by 1950 employed more than half of working women and by 1980, more than 65 percent. The industrial sector has attracted few women. In fact, the participation of women in this sector declined from 22.6 percent in 1950 to less than 20 percent in 1980. Table 51 Compaative ndicators for Lati Ameicn and Carba Comties

Total Total Labor Force Laor Force Fetility Rate GNPper Totl Population Urban P

Antigua & Barbuda 2,540 1.9 0.08 na 29 Argentina so 38 9 2 na a na 2,390 3.0 31.10 8s 19 52 28 Bahamas 10,280 13 34 na na 19.2 2.9 0.24 62 36 47 47 5 17 Barbados 5,350 1.8 025 na na na 83 35 53 43 10 21 na na Belize 1,240 5.0 0.18 na na 10 45 37 17 na na na Bolivia 580 61 6.70 S0 14 Brazil 49 23 46 20 na na na 2,020 35 141.40 75 20 53 27 31 Chile 1,310 2.7 27 na na 13.0 1250 85 17 48 27 16 25 na Colombia 1,240 3.2 29.50 69 na na 16 48 25 34 24 7.0 2.6 na Costa Rica 1,610 33 2.60 45 1S Cuba 53 22 31 12 4.5 25 16.4 na 1.8 10.70 na na na na Dominica 1,440 na na na na na 3.1 0.08 na 22 40 35 36 11 Dominican Republic 730 3.8 na na na 6.70 58 8 51 14 46 15 7.0 Ecuador 1,040 4.3 9.90 55 3.0 na 12 49 19 39 20 7.8 2.7 na El Salvador 860 4.9 4.90 44 17 Grenada 49 25 43 19 na na 215 1,340 33 0.10 na 23 41 42 Guatemala 950 29 18 na na na 5.8 8.40 33 9 48 15 57 17 Guyana 390 3.1 .80 na na 15.0 27 18 53 25 27 26 6.5 4.8 Haiti 360 4.7 6.10 29 22.4 Honduras 36 S0 43 70 8 6.1 2.9 na 810 5.6 4.70 42 10 49 18 61 Jamaica 940 16 6.2 4.8 21.6 2.9 2.40 51 43 51 46 31 16 Mexico 1,830 3.6 8.1 33 33.0 81.90 71 18 48 27 37 29 na Nicaagua 830 5.5 350 na na 58 14 47 24 47 16 na na Panama 2,240 3.1 2.30 54 19 na Paraguay S0 27 32 18 5.7 2.7 na 990 4.6 3.90 46 14 53 21 49 Peru 1,470 4.1 21 8.2 2.9 2L4 20.20 69 15 47 24 40 18 73 St Kitts St Nevis 1,700 3.0 0.04 33 22.5 na na na na 33 29 na na St Lucia 1,400 3.8 0.14 na na 20 4Ow na 30 9 na na na St Vineent & Genadines 1,000 3.0 0.11 na 11 Suriname 20 36 30 na na na na 2,270 3.7 0.42 65 18 45 29 Trinidad & Tobago 4,210 20 20 na na na 2.8 1.20 67 23 54 33 10 39 Uruguay 2,190 2.6 4.6 32 na 3.00 85 23 55 31 16 29 na Venezuela 3,230 3.8 18.30 83 na na 18 49 27 16 28 7.0 2.6 22.6

- not available 1987-World Bank Atlas. NostM/ recent estimate-Social Indicators of Development. / Population Change and EconomicDevelopment data.

Souce World Bank. World Development Report 1989. 7TheWorld Bank Atas; Washington,D.C. World Bank (1988). SocialIndicators of Developmen4Washington, D.C., World Bank (1987). Inter-American DevelopmentBank, Econonic and Social Progress in Latin America (1988). Jamison, B. Women of the WorldkA Chanfook for DevelopingNadons Washington,D.C., USAID (1985). Populaion Change and EconomnicDevelopmen4 Washington, D.C., World Bank (1984). 146 LatinAmerica A growingnumber of LatinAmerican women are heads of households;estimates range from 15 to more than 40 percentof all households.These. female-headed homes represent an increasingshare of the low- incomepopulation. They have less accessto basicurban senices,enjoy fewer opportunities for improving their income-generatingcapabilities, and fall more often belowthe povertyline (Whiteand others 1986).

Education Levels Since1960, educational systems in LatinAmerica have expanded remarkably. Both the supplyof schools and the numberof individualsdemanding educational services have increased. The overallimpact of the emphasison educationhas been substantial.Between 1960 and 1980,the annualaverage rates of increase in enrollmentswere 4.0 percent for primaryeducation, 8.7 percent for secondaryenrollments, and 11.1 percentfor higher education. And between1980 and 1988,enrollments continued to grow at an annual averagerate of 1.5percent at the primarylevel, 3.6 at the secondarylevel and 4.6 at the tertiarylevel (table 5.2). The timing and characteristicsof this educationalexpansion differed from countryto country. Countriescan be groupedaccording to some general similaritiesin their educationalachievements and developmenthistory. Table 5.2 Average Annual Growth Rates of Enrollment, 1960-1988

Pimaav Level Seconda Level TerliarvLevel Total Females Total Females Total Females

1960 - 70 5.5 5.6 9.5 9.6 11.1 13.0 1970 - 80 3.2 3.2 8.9 9.4 11.6 13.8 1980- 88 1.5 1.2 3.6 4.0 4.6 5.5

Source: UNESCO StatisticalYearbook 1990and UNESCO1989.

Group I--Argentina,Costa Rica, Chile,and Uruguayhad illiteracyrates closeto 10 percentin 1980and relativelyhigh per capita incomes. Modernizationof their economiesand educationalsystems took place early. As an example,by 1960,illiteracy rates in Argentinaand Uruguaywere alreadybelow 10 percent. Today,most childrenin these four countriesattend primaryschools, and manycomplete the seconday and tertiary levels. Overall,women in this group have relativelyhigh levels of educationalattainment and relativelylow rates of illiteracy. Group II--Colombia,Mexico, Peru, Ecuador,and Venezuelahdd illiteracyrates between10 and 20 percent in 1980.Although expansion of their educationalsystems occurred after that in the firstgroup of countries, it proceededswiftly as part of a growthprocess that emphasizedrapid industrialization.The educational structuresof these countrieshave been describedas undergoinga "mutation"whereby the numberof people enrolledat the tertiary level expandedconsiderably, while manyothers did not even completeprimary school(ECLAC 1988). Women,as well as men, have been subjectto the mutationpattern. Latin America 147 Group IlI--Bolivia,Haiti, and the CentralAmerican countries of El Salvador,Honduras, and Guatemala had illiteracyrates excn.eding20 percentin 1980and per capita incomesamong the lowestin the region. Haiti is the extremecase withthe lowestincome per capitaand more than half of its populationilliterate. .t countriesare alsoless urban,have lower than averagelabor force participationrates for women,and hanwA high percentage of the labor forcein agriculture.Education levels in this third groupreflect the low or unevendegrees of economicdevelopment. Group IV--Cubaand Nicaraguahave stronglyemphasized preschool and basic educationfor both girls and boys,as well as national literacycampaigns, despite their underdevelopedeconomies. Nicaragua's illiteracyrate, however,as measuredin 1971(most recent data) exceeds40 percent; its GNP is $830per capita, placingit amongthe lowermiddle-income group of countries. In contrast,in Cuba, less than 2 percentof the populationaged 10-49is illiterate. In general, a close relationshipexists between a country'seconomic development and the educational profileof its population.Countries with earlier industrialization and relativelyhigh per capitaincome have lowerilliteracy rates and highereducational attainment. In turn, lowerlevels of economicdevelopment are associatedwith lower levels of educationalattainment. Countriesthat haveexperienced rapid and uneven developmentshow substantial accomplism ents at the higherlevels of educationbut stillsuffer relatively high illiteracyamong the youngerand the rural populatini. As a whole,Latin American countries have improvedthe generallevel of educationof their populations. The educationalgains of females,their improvedaccess to formalschooling, and their increasededucational attainmentare outgrowthsof thisgeneral expansion in LatinAmerican educational systems. Their advances havebeen more the result of nationalsocioeconomic changes than of gender-specificeducational policies.' This does not mean that countrieshave been unawareof the issues related to integratingwomep- into development. Governmentshave assignedpriority to women'sissues in plans and programsof a global nature;however, specific strategies and policiesfor implementationwere rarely defined. Even edaication projects undertakenby internationaldevelopment agencies were found to give little considerationto strategiesto benefitwomen (Stromquist 1986). Indeed,the scopeof most formaleducation projects has been generalrather than gender-specific. At the regionallevel interestin the advancementof womenhas been expressedin manyforums. Four regionalconferences explored ways to integratewomen into the economicand socialdevelopment of Latin America--Macuto,Venezuela, in 1979,Mexico City in 1983,Havana in 1984,and GuatemalaCity in 1988. In the early1980s, the Projecton Educationfor LatinAmerica and the Caribbeanand subsequentmeetings and workshopsanalyzed and proposed recommendationsto address the situationof girls and women. Recommendationsmade to govemmentswith respect to educationhave encompassed coeducation, revision of textbooksand educationalcurricula, and adulteducation programs, with an emphasison womenin rural areas. In manycases, however, implementation of these recommendationshas met v..h institutionaland financialconstraints.2 Despite the progressin educationalopportunities in LatinAmerica, not all womenhave benefited equally from this growth. Althoughthe main issues of femaleeducational attainment relate to incomeand/or economicdevelopment, gender-related differences in attainment,particularly in highereducation, persist. Markeddisparities are found in the literacyand educationalattainment rates amongand withincountries. Table53 showsthat, except in CostaRica, the DominicanRepublic and Uruguay, illiteracy rates for women in about 1980were higherthan for malesin all countriesfor whichdata are available.The countrieswith the lowestilliteracy rates--Argentina, Costa Rica, Chile,and Uruguay--werealso amor3 those with the

1 For a reviewof womenas a subjectof publicpolicy in Latin America,see ECLAC(1988). 2 See, for example,the resolutionsfrom the Secondand Third RegionalConferences on the Integration of Womeninto the Economicand SocialDevelopment of LatinAmerica in Macuto,Venezuela, November 1979,and MexicoCity, 1983(ECLAC 1988). 148 Latin America

smallestdifferences in male and female illiteracy.3 Two of the countries with the largest gender difference- -Guatemala and Bolivia--wereamong those with tLe highest overall illiteracyrates. Indeed, the gender gap in illiteracy is widest in countries with large indigenous populations and low levels of income.

Illiteracy rates among rural females are considerably higher than those of their urban counterparts. In Bolivia,Brazil and Nicaragua, the female illiteracyrate is at least 20 percentage points higher in rural than in urban areas. The countries that industrialized early (Argentina, Uruguay, and Chile) generally have narrower urban-rural differentialsfor females, but rural women have a substantial disadvantagein countries with rapid but more recent expansion (Peru, Bolivia,and Brazil). Rural-urban differenzes are especially evident in areas with large indigenous populations. For linguistic, cultural, and economic reasons, widespread education for rural women remains a challenge.

Gender differences in educational attainment are small (Table ;.4). But, this regional average hides the cross-country differences as shown in appendix table 5.1. Less than 5 percent of Latin American females have finished higher education. Uruguay,Cuba, and Panama, and Peru lead the other countries, with nearly 7 percent of women having completed this level. Table 53 Illiteracy Rates by Sex ( mWages for most ecen year)

Male Female Country Year Total Rural Total Rural

Argentina 1980 5.7 14.2 6.4 15.1 Bolivia 1976 24.2 373 48.6 68.5 Brazil 1985 20.9 44.7 23.4 48.0 Chile 1982 8.5 20.9 9.2 23.2 Colombia 1981 13.6 -- 16.1 -- Costa Rica 1984 73 -- 7.4 -- Cuba 1981 1.7 -- 2.1 -- Dominican Republic 1981 31.8 -- 30.9 -- Ecuador 1982 12.8 21.7 19.4 33.1 El Salvador 1980 26.9 39.0 33.2 45.5 Guatemala 1973 46.4 '. 9 61.5 77.6 Haiti 1982 62.7 -- 67.5 -- Honduras 1974 41.1 52.1 44.9 56.8 Mexico 1980 13.8 -- 20.1 -- Nicaragua 1971 42.0 63.8 42.9 67.0 Panama 1980 13.7 23.6 15.1 28.9 Paraguay 1982 9.7 -- 15.2 -- Peru 1981 9.9 -- 26.1 -- Uruguay 1985 5.6 11.1 4.5 7.4 Wnezuela 1981 13.5 -- 17.0 --

Note: The percentages are for the population over 15 years old except: in Cuba, the rate pertains to the population aged 10-49;in El Salvador, 10 and above; and in the Dominican Republic, 5 and above. Source: CELADE 1987and UNESCO StatisticalYearbook 1986.

3 These rates reflect the cumulative changes in educational policy that began prior to the 1960s. Latin America 149 Table 5.4 Distibution of Adult Populationin Latin Amenca by Highest Educational Attainment, about 1980(percent)

None nary Secondary Tertiary Total

Males 20.2 46.5 28.0 5.4 100 Females 243 43.5 27.9 4.2 100

Source: Kaneko1987.

As with literacyrates, placeof residencemakes a significantdifference in educationalattainment for both genders. Countriesfor whichdata are availableshow significantdisparities between urban and rural women. Rural womenin Bolivia,Guatemala and Haiti are particularlydisadvantaged relative to urban women. In Haiti, for example,nearly 25 percent of women in urban areas have attained secondary education,compared with less thac two percentin rural areas.

Fmwy Eucation

Most Latin Americancountries have achievedimpressive increases in primary schoolenrollment rates (table5.5) and fewdisparities remain between the accessof boysand girls. Overall girlshave a somewhat smallerprobability than boysof enrollingin primaryeducation, but in all countriesexcept Bolivia, Brazil, Cuba,and Nicaragua,they constitute at nearlyhalf of total primaryschool enrollments. Again, the levels of enrollmentof rural childrenare lowerthan thoseof urbanresidents in all countries.Repetition, attrition, and late entryinto schoolare pervasive,particularly among rural childrenand lowersocioeconomic groups. l1O LatinAmerica Table 5.5 Grm Enrolment Rates in PrimaTyEducadon, around 1960and 198P/

(i197) Arond 190 Around 1987 Females Total Women Total Women as % of Total

Argentina 98 99 110 110 49 Bolivia 64 50 91 85 47 Brazil' 95 93 99 97 47 Chile 109 107 102 103 49 Colombia 77 77 114 115 50 CostaRica 96 95 98 97 48 Cuba 109 - 104 100 47 DominicanRepublic 98 98 101 103 50 Ecuador 83 79 117 116 49 El Salvador 80 77 79 81 50 Guatemala 45 42 76 70 45 Haiti 46 42 83 80 48 Honduras 67 67 106 108 50 Mexico 80 - 118 116 49 Nicaragua 66 66 99 104 52 Panama 96 94 107 105 48 Paraguay 98 90 103 101 48 Peru 86 74 122 120 48 Uruguay 112 105 110 109 49 Venezuela l00 100 107 107 49

-- = non-eistent. i/ Totalenrolhment as a percentageof the populationin the relevantage-group. The definitionsvary accordingto the ages and yeas of primaryschooling estabUshed by legislationin each country. / Most recentdata is for 1980. Source: UNESCO Stascal Yeabook (1975,1990).

In the majorityof countries,the primaryschool survival rate of girl was skightlyhigher than that of boys. This couldreflect a loweropportunity cost of femaleschooing relative to males,or the fact that females performbetter in schooland are less ikely to drop out for academicreasons. Accordingto UNESCO (1980)estimates, the averagesurvival rate up to the fourthgrade is 73 percentor belowfor the region as a whole,and in some cases,25 percentof the childrenare lost between'he first and secondyear of school, a fgure that runs as high as 50 pecent in rural areas. In sixcountries in 1984,less than 40 percentof a cohort reachedthe fifthgrade. However,primary completion rates differwidely among countries of the region. Aside from Cuba at 99 percent,the highestsurvival rates are found in Group I countries: Chile at 84 percent,Costa Rica at 79percent, Argentina at 71percent and Uruguayat 96. Groupm countries- -thosewith low levels of economicdevelopment-exhibited the lowestpercentages of a cohort reachingthe fifthgrade. LatinAmerica 151 Table 5.6 Grade Repetition at the PrimazyLeel, Around 1987

Total Femakes

Argentina na na Bolivia na na Brazil 20 na Chile 7 6 Colombia 17 17 Costa Rica 10 9 Cuba 8 na DominicanRepublic 17 na Ecuador 6 6 El Salvador 8 7 Guatemala 4 na Haiti 9 9 Honduras 15 15 Mexico 10 na Nicaragua 15 14 Panama 11 9 Paraguay 9 8 Peru 19 19 Uruguay 10 8 Venezuela 9 na

na = not available Source: UNESCOStatstical Yearbook (1990).

Grade repetitioncontributes to the low levelsof completionat the primaryleveL Repetitionrates are high, reaching20 percent in BraziL Repetitionin the first grade is much higher than in other grades, usuallyexceeding 25 percent. Limiteddata suggeststhat girlstend to repeat less than boys (Table 5.6). In poor and rural areas, childrenstartng schoolare older, their schoolattendance is lower, and their repetitionand dropoutrates are higherthan for childrenin urban sectors. Indeed,urban retentionrates are more than doublethose in rural ueas. In Paraguay,for example,although 48 percentof firstgraders in urbanareas completedgrade 6, only'5 percentof firstgrade pupils in rural areas did so. Althoughthe trend in repetitionand dropoutrates has been downward,the numbershave remainedhigh. Secondwy Edcain Secondaryschool enrollmentalso has grown dramaticallysince 1960, with women's enrollmentrates exceedingmen's overalL In 1960,women's enrollment rate in secondaryeducation varied between6

4 See Wmkler(1980). He also reportedthat the readingachievement of urban pupilswas doublethat of rural ones. 152 Latin America

percent and 38 percent in the region (table 5.7). By 1987the range had risen to between 17 and 92 percent. In 13 of the 20 countries, female enrolment rates exceeded male enrollment rates. In most countries, women constituted at least half of total secondary school enrollment and in no country did the percentage of women in total enrollment fall below 45 percent. As in the case of primary education, enrolment rates are tied to the country's level of income and recency of educatido1l expansion; thus, with some exceptions, enrollment is higher for countries in Group I. The distribution of women in secondaryeducation is heavilyweighted toward general education as opposed to teacher training or vocational and technical programs (table 5.8). A comparison of enrolment distributions between 1970 and 1987,however, shows interesting patterns. For example, there was a steep decline in the proportion of female students in teacher training programs, particularly in Colombia,Cuba, Ecuador, Mexico,Panama, Paraguay, and Venezuela. In Cuba and Ecuador, this decline was accompanied by a large increase in female enrolment in vocational and technical education; in the other five countries, the share of general education rose. Table 5.7 Rates of Enrollment in Secondary Education, around 1960 and 1987 (peeN of appropyiate We gwoup)

Around 1960 1987 Total Females Total Females

Argentina 32 33 74 75 Bolivia 11 10 37 34 Brazil 11 10 34Ja na Chile 24 24 68d' 69 Colombia 12 11 56,/ 49 Costa Rica 21 21 41 45 Cuba 14 na 88 79 Dominican Republic 13 14 51S/ na Ecuador 12 10 56 53 El Salvador 11 10 27J 26 Guatemala 7 6 18a' na Haiti 4 na 19_/ 16 Honduras 8 7 32k' 36 Mexico 11 8 53 53 Nicaragua 7 6 43 48 Panama 29 32 60 63 Paraguay 11 11 29 29 Peru 18 13 65s/ 57 Uruguay 37 38 60w' na Venezuela 21 21 54a/ 49 na = not available.

A/ 1980 k/ 1984 £1 1985 O/ 1986 Source: UNESCO Stadsltcal Yearbook 1975 and 1990. Latin America 153 Table 5.8 Distribution of Female SecondaiyEnrollment by Type of Education, 1970 and 1987

General Teacher Vocadonal/ Education Training Technical County 1970 1987 1970 1987 1970 1987

Argentina 48 451 0 0 52 49 Bolivia na na na na na na Brazil 74 na 14 na 12 na Chile 71 83s/ 0 0 29 17 Colombia 62 760' 13 4 25 20 Costa Rica 93 78 0 0 7 22 Cuba 82h/ 69 102/ 4 8h/ 27 Dominican Republic na na na na na na Ecuador 80 62 7 1 13 37 ElSalvador 64 28 0 1 36 71 Guatemala 71 64W 15 20!' 14 17Q/ Haiti na na na na na na Honduras 75 64 9 8 16 28 Mexico 74-/ 84 0 20 16 Nicarsgua 87 71 6 8 7 21 Panama 62 72 4 1 34 27 Paraguay 83 94 11 0 5 6 Peru 82 96!' 0 ' 18 4 Uruguay 83 92!' 0 0!' 17 8!/ Venezuela 66 95 4 0 31 5 na = not available. Note: Percentages shown may not add to 100 because of rounding. a/ 1980 h/ 1975 £/ 1938 Ot 1986

Source: UNESCO StatisticalYewbook 1975 and 1990. 154 LatinAmerica Table 5.9 Rates of Enrollment in Terdaiy Education Around 1960and 1984 (prent of appmpuiaeqW gmup)

1960 1987 Countiy Total Women Totdal Women

Argentina 10.9 7.2 40.8 44.1 Bolivia 3.6 1.6 na na Brazil 1.6 0.9 lo.9!' 11.2 Chile 4.2 3.1 17.8 16.0 Colombia 1.7 0.6 139 13.4 Costa Rica 4.8 4.3 na na Cuba na na 22.6 26.0 DominicanRepublic 1.3 0.7 lO.(V 8.9 Ecuador 2.6 0.9 36.5 27.5 El Salvador 1.1 0.4 17.7 14.8 Guatemala 1.6 0.3 4'3/ 2.0 Haiti 0.4 0.1 1.e-1 0.6 Honduras 1.1 0.4 9.5 7.2 Mexico 2.6 0.9 15.7 12.1 Nicaragua 1.2 0.4 8.4 9.2 Panama 4.6 4.1 25.9b/ 30.2 Paraguay 2.3 3.2 8.8 7.6 Peru 4.1 2.8 21.5 15.6 Uruguay 8.0 6.5 47.2 50.8 Venezuela 43 2.8 26.5 25.5 na =- -nt available a/ 1975 t/ 1985 S/ 1980 f 1988 Soure: UNESCO Statsikal Yewbook 1975 med1990.

IWfIg! EdAcatfim Female enrollmentsin higher educationhave also increasedsubstantially in the region,but these rates remainslightly lower than men's in a majorityof countries(Table 5.9). The data show that by the mid- 1980s,in six countries,more than 25 percent of femalesof the appropriate age were enrolled at the universityleveL Two and one-halfdecades earlier, female enrollmentrates exceeded5 percent onlyin Argentinaand Uruguay. As expected,greater gender equality exists among countries in GroupsI and IV. The largestgender gaps in favorof men are found in countrieswith relatively low levelsof development. In Ecuador, althoughwomen benefited from the expansionof tertiaryeducation, the gap with respect to men was relativelywide. Althoughthe share of womenin most fieldsof studyhas increased,they remain highlyconcentrated in traditionallyfemale areas such as educationand health sciences. Table 5.10 providesan overallpicture LatinAmerica 155 of women'sparticipation at the tertary levelio the arts and sciencesin 1975and i9825 In mostcountries, femalesaccouoted for 35-64percent of enrollmentsin the arts, but less than 35 percentof enrollmentsin the sciences. A more detailedbreakdown by field showsthat educationscience, a predominantlyfemale field in the 1970s,saw an increasein the concentrationof womenin the 1980s. Women have made considerableprogress in the social sciences. In three out of the twehe countriesfor which data were available--Uruguay,Panama, and Colombia-womenconstituted more than half of enrollments. The medicalsciences, too, have attracteda large number of femaleenrollments; by 1982, femaleparticipation in thisfield exceeded 50 percentin ten LatinAmerican countries. The naturalsciences and agriculturehave attractedfewer women, although in manycountries the trend has been towardhigher participationby women. However,often more teling than the share of women in an overalltertiary education field are the concentrationsof womenat moredisaggregated levels. For example,in the medicalsciences, women were concentratedin nursingrather than medicine. In the educationalsciences, women were primarilybeing trained for positionsin primary school systemsrather than in the more prestigiousand higher paid secondaryand tertiarysystems. The concentrationof womenin certainfields, and in categorieswithin those Selds,is true for all countriesin the region.

AM includeseducation and social sciences,while ss is made up of natural and medicalsciences as well as agriculture. 156 Latin America Table 5.10 Distributionof Counties by Percentage of Females in Terdazy Education by Broad Field of Study, 1975and 1982

Less 35 pecent 35-50 ercent More than 51 rcent 1975 1982 1975 1982 1975 1982 Arts Peru Guatemala Uruguay Paraguay Brazill/ Uruguay Guatemala Haiti Dominican Ecuador Panama Panama Honduras Republic/ Mexico Argentina Nicaragua Haiti Ecuador Honduras Chile Chile Paraguay El Salvador Colombia Colombia Peru Argentina Honduras Mexico Sciences Dominican Colombia Paraguay Uruguay Republic'/ Honduras Uruguay Panama El Salvador Chile Argentina Paraguay Brazilg1 Haiti Panama Argentina Chile Peru Nicaraguan/ Peru El Salvador Colombia Mexico Honduras Ecuador Mexico Guatemala Ecuador Haiti Guatemala

Note: The countries are cliffied within each broad field of study by descending order of percentage of females. Data are for 1975 and 1982 or nearest available years. ^/ Data are available for only one of the two years.

Source: UNESCO (198).

Factors AffectingEducation Quality and Reach Primay Educaton Most studies of the determinants of achievement and attainment in education can be grouped into two categories: those that focus on the availabilityand quality of education, and those that consider the demand for schools and education servicesby students and their families. Teacher quality,availability of textbooks, content of education and other school-relatedfactors affect the supply of education. Family characteristics, LatinAmerica 157 such as income and educationof parents, as well as the cost (price) of schoolingto parents, affectthe demand.6 Althoughprevious research on the determinantsof school participationemphasized such factors as availabilityof schools,distance to school,and costs of education,more recent analyseshave focusedon characteristicsof the decision-makingunit. Few of these studies,however, use this frameworkto identify school-relatedfactors in Latin Americathat explaingender differences in enrollmentor attainment.And even fewer systematicallyanalyze the differencesin returns on investmentin the educationof girls and boys.7 For instance,of the studiessummarized in table5.10, only Bowman and Goldblatt(1984), Irwin and others (1978),King and Bellew(1988), and Schiefelbeinand Farrell (1980)estimated separate equations for males and femalesto allowfor differencesbetween the sexesin the determinantsof achievementand attainment. Studiessuch as Armitageand others (1986),Birdsall (1985), Klees (1979), Psacharopoulos and Arriagada (1989),Tienda (1979), and Wolfeand Behrman(1984) treated genderas an explanatoryvariable--one that wasfound to be statisticallysignificant in severalcases. Thesestudies assume the marginaleffects of factors to be the samefor both sexes. Three of the studiesfound that girlshad a higherattainment level than boys even after controllingfor variousindividual characteristics. These investigations,however, can onlyguess at the explanationsbehind the gender effect. The rest of the studieslisted in table 5.10 did not at all differentiatebetween boys and girls.

6 For a model of schoolingchoice along this framework,see Birdsall(1985). 7 For a comprehensivereview of the effectof schoolinputs' on achievementin the Third World,see Fuller (1987). For Latin America,see Schiefelbeinand others (1978),Schiefelbein (1987) and Muelle-Lopez (1984). 1S8 Latin America Table 5.11 Summary of Studies of Determinants of Educational Achievement and Attainment

Auithor(s) Countnies OutcomeMeawred MaiorConcusons Armitage and Brazil Comprehensiveexam Provisionof simple quality-enhancinginputs such as textbooks,other others (1986) instructional materials, and teacher upgrading can increase student achievement. Birdsall (1985) Brazil Years of schooling In both urban and rural areas, the elasticitiesof demand with respect to public inputs are high. In urban areas, the positive effect of school inputs is greater for children from poorer and less-educated families. In rural areas the effect is greater, if anything, for children from relativelyhigh-income families. Bowman and Mexico School attainment Work constraints on children are associated with lower schooling Goldblatt (1984) attainment and over-age students. The negativeeffect of low income is greater on the schoolingof girls than boys. Clark (1981) Guatemala School attendance Income-earningand housekeepingactivities of children reduce school attendance for some children. Flp and Bolivia, Colombia, Language and Preschool education helps to narrow the gap in achievement of Schiefelbein Chile, Argentina mathematics tests children from different socioeconomicbackgrounds. Rural children (1982) benefit the most from preschool education. Heyneman and Bolivia, Paraguay, Science achievement Schooland teacher qualityare the predominant influenceson student Loxley (1983) El Salvador, Peru, Reading and learning. Colombia, Brazil, mathematics Mexico, Argentina Irwin and others Guatemala School attendance Parents appear to make schooling decisions based on accurate (1978) perceptions of their children's inteDlectualdevelopment. In the case of girls, early intellectual ability predicts length of attendance. Jamison and Nicaragua Mathematics Textbooks and radio-based instructional programs have significant others (1981) achievement effects on achievement. King and Bellew Peru School attainment The influence of family characteristics on educational attainment of (1988) children has lessened over time. The relative impact of parents' education differs for sons' and daughters' schooling. Place of residence is important to females' years of schooling. Simple school inputs, such as textbooksand desks, are effective in raising schooling levels. Klees (1979) Mexico Language and The use of televisionin secondary schools has a significant positive mathematics effect in mathematics and Spanish achievement. Psacharopoulos Brazil Schoolparticipation, Household resources and the demand for child labor have positive and Arriagada attainment, and and negative effects respectively,on children's school participation. (1989) drop out Parental occupation is the strongest predictor of grade attainment. Boys are less likelyto enroll in school, attain less schooling,and have a significantlyhigher probability of dropping out than females. S&hiefelbeinand Chile Achievement and No evidence found of systematic discriminationagainst women within FarreU (1980) attainment the educationalsystem. Becauseof anticipatorysocialization. women score significantly lower than men on university admission tests. Educational attainment is more important for women than men: for the same ype of job, female applicants tend to need higher levels of education than males. TMenda(1979) Peru Labor force Children are more than twice as likely to be economicallyactive if participation they Ive in rural areas as urban areas. Social background has the strongest influence on the labor force activity of younger children; individual characteristics, such as age and school enrollment, are more important for teenagers. Wolfe and Nicaragua Schoolattainment Rural boys receive less schoolingthan rural girls,possibly because of Behrman (1984) higher opportunity costs for boys in agricultural worL Family characteristics, such as income and parental schooling-particularly maternal schooling-have a significant impact on child schooling. LatinAmerica 159 Sclool-Rdated Factors A questionthat has receivedconsiderable attention is how the schoolitself influencesachievement and attainmentof students,after accountingfor familycharacteristics. One reviewof the researchdone in the late 1970sby Simmonsand Alexander(1978) suggested that school-relatedfactors--independent of family characteristics--hada small effect on achievementin developingcountries. A later studyby Heynemanand Loxley (1983) contradictedthese findings. They examinedthe impact of school facors and family characteristicson studentachievement in sciencein twenty-ninecountries and foundthat, for the nine Latin Americancountries in this study, school factors explaineda significantproportion of the variance in achievement.For example,more than 80 percentof the variancein achievementin Braziland Colombia couldbe attributedto schoolquality. In a recent comprehensivereview of the literatureon school factors and achievement,Fuller (1987) concludedthat "muchof the empiricalwork 'suggests' that the schoolinstitution exerts a greater influence on achievementwithin developing countries compared to industrializednations, after accountingfor the effectof pupilbackground" (pp. 255-56).Although the evidenceindicates that the more developeda society is, the greater the influenceof familycharacteristics on attainmentand achievementand the smallerthe effectof schoolfactors, even in the relativelyadvanced developing country of Chile,Schiefelbein and others (1983)found schoolfactors to be more importantthan familycharacteristics. Someevidence also suggeststhat disadvantagedchildren are morelikely to respondfavorably to an increase in the qualityand availabilityof schools. Birdsall(1985) found that rural childrenand childrenfrom low- incomeand less-educatedurban householdsin Brazilcould benefitsubstantially from improvementsin school-relatedinputs. Usinghousehold data fromthe 1970census, she estimatedhigh elasticities of demand withrespect to the availabilityand qualityof schoolsin both urban and rural areas.8 The percentagechange in yearsof schoolingassociated with the percentagechange in the availabilityand qualityof schoolswas virtually the samefor rural and urban8- to 11-year-olds(1.09 and 1.08,respectively). Rural childrenin the 12-to 15-year-oldgroup weremore responsive(0.95) to a changein supplythan are thosefrom corresponding urban households (0.18). Withinurban areas, the resultsshow that as the income and educationof parents rise, differencesin the availabilityand qualityof schools diminishes. The implication,according to the author,is that improvementsin schoolsupply would benefit children from poor householdsmost. A dummyvariable for gender was significant(at the 5 percentlevel) only for urban8- to 11-year-olds,indicating a small,positive effect. Beingfemale increased by 0.13the completedyears of schooling. Researchersalso have tried to identifythe particularschool inputs most effectivein raisingachievement. The evidencefor Latin Americasuggests that simplechanges in schoolinputs couldbe quite effectivein reducingrepetition and dropoutrates and also in raisingchildren's school performance. The schoolinputs most relevant to raising schooling attainment are textbooks and instructionalmaterials, teachers' characteristics,the contentof education,and other supplyfactors.

Textbooksand InsructionalMaterials Althoughrobustness of results differed,most studies suggestthat textbookshave a significantpositive effecton children'sachievement. Someindicated that textbooksmight have a larger impacton children

8 As a m 'asure of the availabilityof schooling,the author used the sum of the incomesof all school teachersin an area, dividedby the numberof childrenaged 7 to 13 in that area. The mean educationof all teachersin the area was used as a measureof the qualityof the schools. 9 The studiesare not strictlycomparable because they often measured achievement using different outcome variables. One studydid not find a strongpositive relationship between the availabilityof textbooksand educationin Ecuador (see Muelle-Lopez1984). 160 LatinAmerica from lowerincome families, oni rural students,and on femalestudents. For example,Brazilian children with parents who lackedschooling were almostthree times as likelyto pass primary schoolif they had used two or more textbooks,compared with those childrenin the same situationwho had no textbooks. In the case of studentswith parentswho had completedprimary school, 73 percentof all childrenwith at least two books passedprimary school, compared with 61 percentof those withno books.10 A more recent research for Brazil(Armitage and others, 1986),lends support to the overallimportance of textbooksin the educationof children. Theyfound that the use of textbooksin rural Brazilin 1983had a positiveand significanteffect (7 to 8 points)on achievementscores. Overall,they reported that simple quality-enhancinginputs, such as textbooks,instructional materials, and teacher upgrading,could inc.ease achievement. The results of their cost-effectivenessanalysis showed that providingwriting materials, textbooksand drinkingwater, and improvingthe levelof the teacher'sown presecondary school education and trainingseemed to be the mostcost-effective means of raisingachievement levels. Other investments, such as furnitureand buildings,were foundto be muchless effectiveon a per-dollarbasis. When consideringstudent characteristics(1981 and 1983second and fourth graders),the authors found that whilebeing male had a negativeeffect, but the effectwas significant only for 1983second graders and it was small. Sex differences in achievement became highly significant,however, when broken down by performancein Portugueseand mathemnatics;girls consistently did better than boysin Portugueseand less well in mathematics. Research on Nicaraguan children suggests that textbooks can be relatively more effective in rural than urban schools. Jamison and others (1981) examined whether the availabilityof textbooks and radio instruction improved the mathematics achievementof a sample of urban and rural Nicaraguan first graders. They found that textbooks had a significantpositive effect on achievement.1" Their availabilityincreased student post-test scores by about 3.5 correct items. The reported differences in the effectiveness of textbooks between rural and urban schools suggest that the availabilityof mathematics textbooks could compensate for the disadvantage of being in a rural school. For example, on an exam administered before and after the textbooks were made available, the urban controlgroup averaged4.5 additionalcorrect items, while the rural group averagedan additional6.1 items. The authors concluded that textbooks not only improved school quality but also reduced the gap in achievement between urban and rural schools. As in the case of Birdsall'sstudy, the evidencesuggests that improvements in school-related inputs could benefit more needy children substantially.

King and Beliew (1988)found the presence of books to have a positive and statisticallysignificant influence on the education attainment of girls in Peru. Reading or math books, or both, raised schooling levels by nearly one-half year for both boys and girls. Separate analysesby birth cohort showed significantestimates for the impact of textbooks on attainment for each of the six cohorts: the estimates increased for both sexes in the younger cohorts up to cohort born in 1955-59. For example, the effect of having books for every student increased the years of schoolingby 0.69 for the 1940-44female birth cohort, while the introduction of these books added 1.14years for the 1955-59birth cohort. These cohort regressions also indicated that having a textbook had a larger effect on females than males; for example, the impact of textbooks on the 1955-59male cohort was 0.92years of schooling. The larger effect of textbooks on females could mean that "becausethere was less of a push witbin the family to educate girls, the quality of the learning process was more important in determining how many years of schoolinggirls attained" (King and Bellew 1988, p. 41).

10 See Fuller (1987) for a review of these 6tudies.

11 The impact of radio instructional programs is discussed below. Latin America 161 Tea.er Caracteristic

Evidence pertaining to teacher quality and its impact on achievement is not as conclusive as that for textbooks and instructional materials. In a review of nineteen studies assessing the impact of teacher characteristics, Schiefelbein (1987) noted that only eight reported that teacher's education had positive effects on achievement. The greatest differences were obtained when comparing teachers with some educational training and those with none. Increments in the length of training seemed to have decreasingly marginal returns in terms of student achievement. In addition, only four out of twelve studies noted the experience of teachers as statisticallysignificant. Although teacher upgrading had an insignificanteffect on achievement,Schiefelbein suggestedthat upgrading through stages in good schoolsor in discussiongroups (as opposed to traditional classes in selected topics) may have a positive effect on achievement. A study of rural Brazil (Armitage and others, 1986) used four indicators of teacher quality--teachers' basic cognitive skills (years of schooling),years of experience, motivation to tcach (salary as a percentage of the regional minimum wage), and in-servicetraining. Of the four, the years of schoolingof the teacher had the most consistently positive and significant impact on student achievement, albeit not a large one; one additional year of schoolingby the teacher raised student achievementonly 1.5 points. Although the effect was not large, the authors suggestthat this finding is important, given that teachers in rural areas generally had low levels of educational attainment. For example, 30 percent of the teachers in the sample from northeast Brazil had four or less years of formal education. Teacher upgrading through in-service teacher training programs seemed promising, and preliminary estimates indicated a large positive impact--a 6- point achievement gain--in the sample of 1983second graders.

Although teachers' salaries are positivelyrelated to student achievement,the effect is small, and the authors cautioned against relyingonly on increasingthe salaries of current teachers to improve student achievement. The coefficienton years of teaching experience was insignificant.

The Brazilian cost-effectivenessresults indicate that the achievement pay-off to each dollar for teacher education appears to be higher for completing primary education than for secondary schooling. Four complete years of primary education, or upgrading through in-service training, is cost-effectivecompared to raising the level of teachers' formal education to three years of completed secondary education.

In addition, the authors indicate that certain teacher behaviors could have a differential impact on the achievement of girls relative to boys. They pointed out, for instance, that female teachers hold sex- stereotypical attitudes with respect to female capability in mathematics and use teaching techniques that are likelyto depress girls' achievement in this area.

Contentof Educatio

Several studies indicate that textbooksare &s agent for the transmissionof gender stereotypes in the region (for example, see Braslavsky 1984). Textbooks portray women as housewivesand mothers, roles in which they are seen as passiveand incapableof makingdecisions. When women are portrayed in the marketplace, they are placed in jobs traditionallyassociated with the female nurturing stereotype (for example, teaching, nursing, and domestic service).

Although ample evidenceindicates that textbooks, by devaluingthe roles and status of women, strengthen the negative stereotype of females in society, little research has addressed the impact of those stereotypes on girls' achievement and attainment. If information absorbed at a young age shapes attitudes and aspirations, it would probably affect future performance at the secondary and post-secondary levels. An issue of equal concern in research is the overall relevance of the educational content--textbools and curricula together--to children from rural areas and indigenous cultures. How the education these girls receive through formal schooling relates to their future employment opportunities and i;ves ;- a question. 162 Latin America Other SupplyFactors Other schol-related factorsthat affecteducation are furniture,number of teachers,the schoolcalendar, radio instruction,and the presenceof additionalgrades in the primaryschool, to name but a few. For example,Jamisoki and others (1981)determined that radio-basedinstructional programs had significant positiveeffects on achievement.In a Nicaraguanexperiment, instruction by radio increasedstudent exam scoresby 14.9correct items compared to an increaseof 3.5correct items from the introductionof textbooks. The authorsattributed the greatereffect of radioinstruction on achievementto the uniformlyadministered radio lessonscompared to "the more inconsistentapplication of the textbooktreatment (in the hands of teacherswith relatively low levels of education)"(p. 565). More importantly,either textbooks or radio can reduce the urban-ruraldifferences in schoolquality. The Kingand Beliew(1988) study for Peru assessedthe effectsof the availabilityof furniture,the number of grades offered,and the numberof teacherson schoolingattainment. The resultspertaining to all birth cohortscombined indicated that the provisionof a desk and chairfor each student,holding other factors constant,raised the attainmentof malesabout 0.4 yearand 02 yearfor females(the latter is not statistically significant).Each additionalgrade offeredin a schoolincreased the educationalattainment of Peruvian boysby about 0.8 year, whileit affectedgirls by 0.5 year. The authorsinterpreted the differencebetween boys and girls as a reflectionof the relativelyweaker demand for female schooling. With respect to increasingthe numberof teachersin a school,the largest effectwas obtainedgoing from one to three. The impactof the number of gradesoffered in a primaryschool was positive,and it increasedacross the cohorts. For example,for femalesborn between1950 and 1954,adding one more grade to the number of grades offeredwould have increasedattainment by about one-halfyear, whilefor those born in the 1960s,the effectwould have been 1.1 years The improvementsin the availabilityof schools,thus, had increasedthe attaimmentof youngercohorts.

Fuamily-relaed Factors Most of the studiesof factors affectingthe demand for educationhave examinedthe impact of family influences,such as parents'education and incomeand the workconstraints placed on children,on student achievementand attainment. The evidencesuggests that parents'education and their abilityto pay have strong,positive effects on schoolingattainment. The workconstraints placed on childrenare likelyto have gender-specificeffects. Someevidence indicates that the opportunitycost of girls' schoolingis lowerthan that for boys.

Parents' Education Studiesfor Nicaraguaby Wolfeand Behrman(1984), for Peru by Kingand Bellew(1988), and for Brazil by Birdsall(1985) suggest a strongrelatonship between parents' education and children'sparticipation in schools. Wolfe and Behrman'sestimates indicate that parental educationhad a significanteffect on children'sschooling, with diminishing returns at the higher levelsof parents' education. The maximum impact occurredafter eight or ninegrades. The mother'slevel of educationhad a greater impacton her children'sschooling than did the male's in the household. Relative to a householdwith the mean characteristics,every additional year of a mother'sschooling increased the child'sschooling by 0.12grades; an additionalyear of educationfor the male in the householdhad about one-thirdthis impact.12 King andBellew (1988) determined that the schoolinglevels of both parentshad a positiveand statistically significanteffect on the educationalattainment of Peruvianchildren, both boys and girls, althoughthe impactof each parent's educationdiffered for boysand girls. The father'seducation had twiceas large an

12 The authors explainthat the extent to whichthe larger coefficientsfor women'sschooling reflected tastes, genetics,or householdproductivity effects remain unclear. LatinAmerica 163 impacton his sons' schoolingas did the mother's education,while the effectsof either parent's education on a daughter'sschooling were equivalent(strong and positive). Theycalculated that the elasticityof the sons'education with respect to their father'seducation, evaluated at the samplemeans (0.19), is more than twicethe elasticityof the mother'seducation (0.09). Thesetwo elasticitymeasures are approximatelyequal (0.19and 0.21)with respect to the educationallevels of daughters. The authorsexplained these resultsby noting that educated mothers partly counterbalancedthe father's preferenceto send sons rather than daughtersto school. Kaufman(1989) found similar results in Brazil.in the urban and rural areas of Sao Paulo,Salvador and Bahia: exceptfor Salvador,still a *traditional"city, the impactof mother's education on raisingchildren's grade levelsand reducingtheir over-agewas greater than that of fathers' education. WhenKing and Bellew(1988) analyzed the effectof parents'schooling in eachof the sixcohort regressions, they once again found positiveand statisticallysignificant effects, although these effects diminishedover time. The impactof parents'schooling was larger for the older cohortsthan for the youngerpeople of both genders. For example,for femalesborn between1950 and 1954,the elasticitieswith respect to mother's and father's education,at their mean values,were 0.19 and 0.22;these same elasticitieswere just 0.11 and 0.13 for the 1960-66cohort. Accordingto the authors,these findingscan be attributedto the effectsof increasedavailability and qualityof schoolsin more recent yearscr to the higherreturns associatedwith education. Birdsall(1985) also estimated a significantrelationship between parents' education levels and their children's yearsof schoolingin Brazil. Usingher reportedresults, elasticities calculated at the mean valuesindicated that a mother'syears of schoolgenerally had a strongerimpact than the fathers'on the number of years of schoolingtheir childrencompleted. This relationshipwas particularlytrue in the urban areas of Brazil. For urban childrenaged 8 to 11, the elasticitymeasures we-e 1.123for mothers'education and 0.107for fathers',while in rural areas the elasticitieswere 0.139and 0.112,respectively. In a study in rural Guatemala of the relationshipbetween school achievementof children, family characteristics,and preschoolingcognitive test performance,Irwin and others (1978) found that early intellectualability influenced girls' length of attendance.In addition,parents' perceptions of the intellectual abilityof their childrenwas found to have a positiveinfluence on both schoolenrollment and lengthof attendance.

Abilityto Pay Parents' abilityto pay for educationalso appears to be positivelyrelated to their children'sschooling attainment. Wolfe and Behrman (1984)reported for Nicaraguathat higher predictedearnings for the womanin the householdand other householdincome (generally represiating the earningsof the malein the household)were both significantiyassociated with greater educationattainment by the children. For the nationalsample, the calculatedelasticity measure with respect to householdincome, evaluated at the mean,was 0.832. With respectto other householdincome, the authorscalculated a larger responsiveness of children'sschooling in the central metropolisas comparedto the less urban and rural areas of the country. Evidencefrom Peru alsoindicates the importanceof familyincome in the attainmentof higherschooling levels. Kingand Bellew(1988) used parents' occupation as a proxyfor incometo examinethis relationship. Childrenof farmershad feweryears of schoolingthan did childrenof parentswith white-collar jobs. For example,they foundthat if a motherhad a white-collarjob, her sonswere in school0.6 yearslonger, while her daughtersattended 1.1 years longer. Sons and daughterswere in school1.1 and 1.4 years longer, respectively,if the father had a white-collarjob. Childrenof motherswho were not in the labor force tended to have more years of schoolingthan did childreuof mothers in farm-relatedoccupations. The authorsexplain this result by notingthat nonfarmmothers who were not employedwere likelyto have higherfamily incomes than thosefrom farm families. Furthermore, Farrell and Schiefelbein(1985) reported that all of the childrenof upper-classfathers in Chilehad completedprimary school, compared with 70 percentfor those withmiddle-class fathers, 48 percentfor childrenof urban laborers,only 18 percentfor childrenof rural laborers. 164 Latin America

In a study of Mexico, Bowman and Goldblatt (1984) indicated that low income was more detrimental to the schooling of girls than to that of boys. Visible returns to schooling,which the authors measured by the proportion of white-collar workers with high incomes, were not significant in the girls' educational attainment but were important for that of boys. Therefore, among lower-incomehouseholds, demand for girls' education could be weaker than that for boys'.

Value a£ Childen's Time In many parts of Latin America, children are required to work to contribute to the:r family's income, and these work constraints can affect their educational attainment. The nature of these ,.ork constraints seems to vary between boys and girls. Moreover, the demand for children's labor is generally higher in rural areas and among low-incomehouseholds. In Tienda's (1979) study of economic activity of children in Peru, rural children who were enrolled in school were nearly six times more likelyto work than urban children. One conclusion that can be drawn is that rural children could more easily combine school and work activities because of the seasonal nature of many agricultural tasks. Daily absences from school also probably facilitatedthis combination. AlthoughTienda found children's employment to be negativelyassociated with school enrollment, the net effect was more important for teenagers than for 6- to 13-year-olds. Gender was an important determinant of labor force participation rates only among teenage children. The author reported no difference in the probability that boys and girls aged 6 to 13 would become economicaly active, but older males (14- to 18-year-olds)had a labor force participation rate higher than that of females. The relationship between low attendance and work constraints for Guatemalan children was explored by Clark (1981)uEing a time-use survey. She found that a relativelyhigh percentage of primary school children were not attending school. For some children, especiallyfor older boys, income-earningand housekeeping activitiesexplained the low school attendance. Work constraintsdid not explainall the cases, however. For some children, parents assigned a low value to the expected return from investing in schooling.13 W'ith respect to the impact of work constraints on the school attainment of girls relative to boys, a lower opportunity cost for girls' schooling could explain why in some cases girls receive more schooling than boys. 'Nolfe and Behrman (1984) found this situation in rural Nicaragua, but they found no gender difference in schooling for children in Managua. They attributed this result to the higher opportunity costs of male schooling in rural areas, in that farm tasks generally require a boy's physical strength. In addition, the Wolfe and Behrman study explores the effect on the schooling of boys of mothers who had expressed a desire for male children.14 The national results showed that these boys did not have significantlyless schooling than girls, as was the case for other boys. In Managua, boys with mothers who preferred sons atLainedconsiderably more schooling than did other children. The authors concluded that this male preference compensated "for a systematictendency for boys to otherwisebe slightlyless schooled than girlse and may also have reflected the evidence that the expected returns on male schooling were greater than those for girls. Schiefelbeinand Farrell (1980)found similargender differences in Chile. They attributed them to the fact that foregone earnings for older boys were higher than those for girls. Consequently,lower income families were more likelyto withdraw the boys from school.

Bowman and Goldblatt (1984) provided evidencethat children's work also explained over-age studenits in Mexican schools. In their study, over-age referred to students in the first grade who were 10 or more years old.

14 This preference was indicated by a positive response to the question; "If you had four daughters and no son would you have another baby in hopes of having a son?" LatinAmerica 165 In an analysisof the educationof Brazilianchildren, Psacharopoulos and Arriagada(1989) found similar evidence. They found that boys are less likelyto enroli in school,tend to have lower attainmentlevels, and have a significantlyhigher probability of droppingout of schoolthan girW.The authorssuggest that these findingscould be attributedto higheropportunity costs and labor demandfor boys relativeto girls.

Policy Initites Manystrategies to improveattainment and achievementof childrenhave been suggested,and severalhave been implemented.Among the manyschool-related strategies that havebeen tried are early educational stimulation;radio programs;changes in schoolcalendars and promotionprocedures; and provisionof textbooks,school lunches, and transportation.15Some of these progm seem to be relativelysuccessful in improvingthe educationalattainment and achievementof girls.' In additionto providingand increasingthe use of textbooks,a change in the stereotypesof womenin textbooksand other instructionalmaterials could create strongerrole models that mightmotivate girls towardhigher school achievement. Mexico, for one, has removedthe traditionalgender stereotypes from textbooks.School curricula in rural areas couldbe designedto be more relevantto the lifestyleof peasant and indigenouschildren so as to increasethe returns from schooling. In addition,textbooks could b combinedwith radio instruction,which has provedeffective for the educationof ruralNicaraguan students. 1

The wayteachers relate to studentsand, in particular,their attitudestoward female students also influence achievement.In a studymentioned earlier, in the case of Brazil,teachers (predominantly female) believe girls are less capablein mathematicsand consequentlyfail to use teachingtechniques that mightimprove girls' achievementin math (Armitageand others, 1986). Changingteachers' views through their training programscan made a difference. In Uruguay,Argentina, Cuba, CostaRica, Bolivia,and Guatemala,the schoolcalendar has been adapted to providetime for childrento help with agriculturalwork and to allowfor extremeweather conditions. Schiefelbein(1987) found the schoolcalendar changes to havebeen effectivein improvinggirls' attendance. Whether similar adjustmentsin days and times for schoolwould benefit girls who mainlyengage in houseworkremains to be researched. Preschooleducation and early interventiontargeted toward poor childrencould also be explored. Filp and Schiefelbein(1982) reported resultsfrom Argentina,Bolivia, Colombia, and Chileshowing that rural children of lower socioeconomicstatus reaped the greatest benefits from preschooleducation. Their participationin preschoolappears to have had a positiveeffect on their age of firstgrade enrollmentin Argentinaand Chileand promotionto the secondgrade in Argentinaand Colombia. Negble effects were found,however, for Bolivianchildren. For peasantand indigenouschildren whose transition to primaryeducation may be difficult,early childhood interventionseems particularly promising. In a reviewof earlychildhood intervention programs in Latin America,Halpern (1986)concluded that "it is the early childhoodteacher's tendencyto recognizeand

's For a descriptionof these experimentssee, for example,Schiefelbein and others (1978),Braslavsky (1984),and Schiefelbein(1987). 16 Althoughinteresting experiments have been conductedin nonformaleducation, they are outsidethe scopeof this report.

17 Televisionprograms such as "PlazaSesamo" have provedto be quite effective,too. 8 Chilehad a policyof automaticpromotion. 166 LatinAmerica respondto individualchildren's learning needs that, if adoptedin primaryschools in LatinAmerica, might have the mostfar-reaching conseqr,ences for children'sprimary school careers' (p. 215). Manyof the familycharacterist:-; that determinethe demandfor educationare difficultto modifyin the short run. Two possibilitiesc -ld be exploredfurther, however. First, parent-trainingprograms and encou,agementof parents' participationin the school systemhave been found effectivein improving children'sschool experience. In Chile,for example,parents' involvementin the actual constructionand managementof schoolsincreased their interestin their children'seducation (Schiefelbein and others1978). This participationalso increasedtheir awarenessof their children'sintellectual ability, which encouraged the childrentoward greater achievement. Other family-orientedmeasures to explorewould include programs to reducethe amountof time girlsare expected to devote to housework. Clark (1981) and Chamie (1983) recoLtmendedthe provisionof communityday-care facilities for preschoolsiblings so that girls with prohibitivework constraintscould attend school. In summary,both school-relatedand familyinfluences have been found to have a strong impacton the achievementand attainmentof childrenin primaryeducation. Althoughmost of the studies havebeen done for countriesin GroupsI and Im, even in the case of Chileand Argentina(Group I), whichhave higher income levels and more balanced educationalsystems, school factors seem to have been the predominantinfluence on students'education. Evidence from somecountries suggests that improvements in schoolfactors can havea greaterimpact among lower income and rural children. Evensimple changes in schoolinputs, such as the provisionof textbooksand changesin schoolcalendar, have proveneffective in raisingattendance student performance. Expansionof schoolsand improvementsin their qualityalone are not enoughto close the gap between the attainmentlevels of rural residentsand thoseof more advantagedurban groups. Thesemeasures must be complementedwith policies that changeparents' attitudes toward their children'seducation, as well as withefforts to ease the constraintsthat workoutside of schoolplaces on schoolwork.

Secondaiy and Higher Education The number of secondaryau higher educationinstitutions has grown dramaticaflyin recent decades throughoutmost of LatinAmerica. For example,in 1958,Venezuela had sixuniversities and one pedagogic institute;by 1984,it had nineteenuniversities, seven pedagogic institutes and forty-sixother institutionsof highereducation (Psacharopoulos and Steier 1988).The data for LatinAmerica show considerable growth in private-sectorinstitutions and an increasein the shareof studentsattending private institutions. In 1955, only 14 percent of total enrollmentswere in the privatesector, by 1965,private school enrollments had reached20 percentand by 1975,34 percent. The growthin publiceflucation was alsoextraordinary, with enrollmentsjumping from 350,000in 1955to uver2 millionin 1975.9 The growthof postprimaryinstitutions and enrolmentswas concentrated in urbanareas. In manycountries, rural secondaryschools are scarce,and in some countriesuniversities exist only in the capital city. As a result, the greater availabilityof secondaryschools and universitieshas benefitedmainly the demand of urbanwomen for education.Various studies have addressed the educationaladvantage that urbanresidents have overrural counterpartsin the regionas a whole.20For example,in the case of Peru, Kingand BeDiew (1988)showed that being an urban residentat age 13 added about one year to schoolinglevels.

19 Levy(1985) noted the exceptionsof Cuba and Uruguay,where private universities have not opened. 20 See, for example,Klees (1979), Fernandez (1986), and Velasquez(1987). Latin America 167 Famiby Sodo -onoinkChmarati Socioeconomicstatus is a goodpredictor of secondaryand postsecondaryschooling access and attainment in LatinAmerica. Schiefelbein(1987) reported that graduatesfrom high schooltended to be fom the upper 25 percentof the socioeconomicdistribution of each cohort. For example,Farrell and Schiefelbein (1985)estimated that the distnrbutionof studentsfinishing secondary school was: 77percent for the upper class,49 percent for the middleclass, 21 percentfor childrenof urbanworkers, and 4 percentfor children of agriculturalworkers.

Selectionprocesses of educationalsystems have tended to discriminateagainst students from lower socioecononiclevels. In Braziland Colombia,university entrance examinations have caused many families to investin good,fee-paid private high schoolsas a way to gain access to good, fireepublic universities (Schiefelbein1987). In the end, studentsfrom upper-incomelevels nave been the ones to attend good publicschools, rather than the verystudents that the free educationwould benefit the most. Disimation in the educationalsystem according to socioeconomiccharacteristics also affectswomen. A 1982study of five metropolitanareas-Bogota, San Jose, Panama City, Lima-Callao,and Caracas- showedthat the proportionof womenwith thirteen yearsor more of schoolingwas substantiallylarger amonghigher per capita incomegroups. In Caracas,for example,47.7 percent of women25 to 34 years old in the highestincome category had 13 or more years of educationas comparedto 1.7percent for woomenin the lowestincome category (Braslavsky 1984). In addition,UNESCO (1981) reported that amongthe womenwho enrolledin higher educationin Venezuelain the last few decades,the highest proportionattended fee-paidprivate institutions.

Edcadon anudLabo Mm*& Linkag Even where gender equalityseems to prevailin education,women find themselvesat a disadvantagein enteringthe labor market,as suggestedin the casesof Chile and Uruguay. The educationalsystems of these two countries are among the best developedand most efficientin the region. The expansionof educationalopportunities for womentook place early as comparedwith most other countries,and today womenhave acceptableaccess to semiprofessionaland higherlevel occupations. Howevr, it is at this stage that womenare at a disadvantagerelative to men. In particular,differences arise in the linkage betweenthe formal schoolingsystem and the labor market. Schiefelbeinand Farrell's 1980longitudinal study of a cohort of youngChilean adults found that the first noticeableeducational difference between men and womenoccrred in their performanceon the university admissionstest, withwomen scoring significantly lower than men. The authorsexplained this result on the basisof "anticipatorysocialization" among women. Becausewomen enrolled in traditionalyfemale fields of study,which required lower test scoresfor universityadmission, they were underless pressure to perform well on the exam. The studyalso showedthat the lengthof time spent in schoolwas more importantfor womenthan for men in gettinga job. Schiefelbeinand Farrell (1984)reinforced this claim by notingthat the labor market for women seems to have been more predeterminedby tradition than that for men, a situationthat

21 In Chile,by 193049A percentof primaryand 43 percentof secondarystudents were female (Schiefelbein and Farrell 1980).In Uruguay,women already represented half of secondarystudents by about 1920(Piotti 1988). 168 Latin America

explained the larger importance of educational attainment for women, as well as the smaler importance of occupational aspirations.22

The evidence from Uruguay is similar. Although women have equality of access to the universityand even enjoy a bias in their favor, the percentage of men who graduate exceeds that of women. Piotti (1988) explained the lower school survivalrate of women by the fact that when some got married or had a baby, they dropped out of school or took more years to complete their degrees. Women also found themselves at a disadvantagein the labor market; and as in the case of Chile, educational attainment proved to be more important for women than for men for similar occupations.

Policy Evaluation

As in the case of primary education, the expansion of women's opportunities at the secondary and tertiary levels was a consequenceof the general expan-.on of educational systemsand economiesin many countries. Moreover in almost all countries, despite the advances, the gap between men and women's educational attainment has been wider at the tertiary level than at the lower levels, and in those countries where industrialization and urbanization started late.

Various strategies have been proposed to improve the access,achievement, and attainment of disadvantaged groups within Latin America. Few of the studies, however, have focused on strategies targeted toward woomen.The Organization of American States (OAS) (1985) made two general recommendations for the region that addressed the needs of rural womenand encouraged women to enter diverse,and typicallymale- dominated, fields of study. For rural areas, the OAS recommended increasing the availabilityof secondary schools. Other researchers have proposed ways, such as tele-education, to improve student performance in regions with little access to schools?3

Greater availabiity of schools would lower the cost of education to women. Nevertheless, even free education is not sufficient to ensure access by lower income groups. An important limiting factor is that of foregone family income, a factor that calls for measures such as scholarships,subsidized student loans, or funds to cover the foregone income. Whatever the financing scheme, it should be targeted to the secondary level of education, where social selectivityis most crucial.24

In addition, strategies that modify attitudes toward the education of women are required. Mainly in the case of the lower socioeconomic,rural groups in Latin America, women are restricted to marriage, fertility, and housework. Consequently,neither parents nor daughters see the benefits of additional years of forrmal sc'ollng.

The OAS (1985) suggested establishingof public and private fellowshipprograms for women in new fields of study with some promise of future employmentto encourage women's interests in diverse fields of study. In addition, it suggested establishingemployment agencies specificaly for women to increase their access to job markets under equal conditions, along with strengthening organizations of women workers and professional women. Given the precarious financial situation and many urgent problems facing Latin

22 The authors also found that, for females, educational attainment was more important than educational achievement as far as the level of the first job was concerned. For both genders, although more so in the case of males, educational quality had a strong direct effect on occupationalattainment.

23 Mexico's use of television at the secondary level, Telesecundaria, has had a positive effect on student achievement. For an evaluation of Mexico's experience, see Klees (1979).

24 The financing of education in Latin America and its efficiencyand equity implications have been much debated. For a comprehensive review of the main issues, see IDB (1978), Schiefelbein (1987), and Woodhall (1983). Financingschemes could be a source of discriminationagainst women; for example,see Jimenez and Tan (1987). Latin America 169

America in the 1980s,the success of these and other strategies will require concerted commitment and effort by the population, govermnents, and national and international organizations.

Summaryand Conclusions

In recent decades, Latin America's supply of schools and demand for educational services has expanded substantially. As a result, the educational attainment of girls and women improvedthroughout the region, although with differences in gains across and within countries.

The educational profile of the Latin American population and that of women in particular is closelyrelated to a country's level of income and the characteristics of educational expansion. Countries where modernization of their economies and educational systems took place earlier have lower illiteracy rates, higher levels of educational attainment and greater equalityin the education of men and women. Countries with more recent economic and educational growth show a considerable increase in the number of people enrolled at the higher levelsof education, while illiteracyrates among youngerpeople in rural areas remain relativelyhigh. In these countries, the attainment of womenis still below that of men, especiallyat the postsecondary level of education. Fmally, countries with lower levels of income have higher illiteracy rates, lower levels of educational attainment and a wider gender gap at the upper level of education. While the gender gap has narrowed greatly or even closed over time in most countries,a large gap remains between urban and rural residents. In countries with large rural and indigenous populations,vast groups of women are illiterate. Full primary education for rural children has not yet been achieved. Moreover, the advantages of secondary and post-secondaryeducation generally accrue to urban women from middle and upper socioeconomic levels.

The review of the empirical literature on the determinants of educational attainment and achievement of girls and women and the assessment of the policies designed to improve their schooling indicate the followingfindings and guidelines for action:

- Both school-related factors and family characteristics have a strong impact on the educational achievement and attainment of children. The impact of changes in school factors is greater among children from lower income and rural backgrounds.

* The school inputs that have a positive impact on achievement and attainment of girls are textbooks and instructional materials, teacher characteristics and other factors such as adding grades to a school. Consequently,among the school-relatedpolicies suggested to raise schoolinglevels are a greater provision and use of textbooks,suppression of traditionalgender stereotypes in reading materials,revision in school curricula to make them more relevant to the lifestyles of rural children, implementation of teaching techniquesto foster girls' achievement,adaptation of the schoolcalendar to accommodatehousework and income-earning activities,and pre-school and early intervention programs targeted to the needs of rural students.

* Parents' income and years of schooling have a strong and positive impact on children's education, with the mother's educational level seeming to have a greater impact. The mother's influence also seems to be stronger in the case of daughters than sons, a fact that counterbalances the weaker demand for girls' education linked to the father's education level.

* The relationships between work constraints and enrollment, attendance and attainment are uncertain. However, there is some evidence that the lower opportunity cost of girL schooling explains why they receive more education than boys. To the extec. that work constraints affect school participation, adapting the school calendar to agricultural tasks and the provision of day-care facilities for pre-school- aged siblingswould reduce the need for children to work, and reduce the opportunitycosts of education.

* At the secondary and post-secondary levels, selection processes within the educational systems discriminate against students from lower socioeconomiclevels. Women with access to higher education 170 Latin America are concentratedin fieldsof studyassociated with women's traditional roles in society.Policies suggested to address these two issues are the provisionof scholarshipsto lower income groups and the establishmentof fellowshipprograms for womenif diversefields of study. * Even though equalityin the educationof men and women has been achieved,women remain at a disadvantagein enteringthe labor market from schooL Educationalattainment is found to be more importantfor womenthan for men when enteringsimilar occupations. * Fmally,any future successesin improvingthe educationallevels of less-advantagedgirls and womenin LatinAmerica will dependnot onlyon the designof educationalpolicies and strategiesthat target the specificneeds of females, but also on a firm commitmenton the part of educators, government authorities,and internationalorganizations to achievemore in education. Latin America 171 References

Armitage, J., B. Joao, R. Harbison, D. Holsinger,R. and Leite. "SchoolQuality and Achievement in Rural Brazil."EDT Discussion Paper No. 25. Washington, D.C.: World Bank, 1986. Behrman, J. "Schoolingand Other Human Capital Investments:Can the Effects Be Identified?"Economics of Education Review, 6(3): 301-305,1987. Behrman, J. and N. Birdsall. "The Quality of Schooling:Quantity Alone is Misleading."AmericanEconomic Review, 73: 928-46, 1983. Behrman, J. and B. Wolfe. "The SocioeconomicImpact of Schooling in a DevelopingCountry." Review of Economic and Statistics, 66(2): 296-303,1984. "Labor------. Force Participation and Earnings Determinants for Women in the Special Conditions of Developing Countries."Journal of DevelopmentEconomics, 15, 259-288,1984. Behrman, J., B. Wolfe, and D. Blau. 'Human Capital and Earnings Distribution in a DevelopingCountry The Case of Pre-revolutionary Nicaragua."Economic Developmentand CulturalChange, 34: 1-29, 1985. Birdsall, N. "A Cost of Siblings:Child Schoolingin Urban Colombia."Research in Population Economics, 2: 115-1, 1980. "The------. Impact of School Availabilityand Qualityon Children's Schoolingin Brazil."EDT Discussion Paper 82. Washington, D.C.: World Bank, 1982. "Public-.------Inputs and Child Schooling in Brazil."Journal of DevelopmentEconomics, 18: 67-86, 1985. ------and- M.L. Fox. "Why Males Earn More: Location and Training of Brazilian School Teachers. Economic Developmentand CulturalChange, 33: 533-56, 1985. Bowman, M. J. "An Integrated Framework for Analysis of the Spread of Schooling in Less Developed Countries." Compaative Educaton Review, 28(4): 563-83, 1984. and CA. Anderson. "The Participation of Women in the Third World." ComparativeEducation Review, 24(2): S13-S32,1980. Bowman, MJ. and P. Goldblatt. "SchoolAttainments in a Development Perspective: Transition Patterns in Mexico."Working Paper Series #84-6. Chicago, Illinois: Universityof Chicago, 1984. Braslavsky, C. Mujer y Educacion: DesigualdadesEducativas en America Latina y el Canbe. Santiago: OREALC, 1984.

CELADE. BoletinDemogrfico. (Santiago) 20(39), 1987. Chamie, M. National Institional and Household FactorsAffecting Young Girl's School Attendance in DevelopingSocieties. Washington, D.C.: International Center for Research on Women, 1983. Clark, C. 'Children's Economic Activities and Primary School Attendance in Rural Guatemala." Paper presented at the annual meeting of the Society for Research in Child Development, Boston, Massachusetts, 1981. 172 LatinAmerica Cohn,E., and R. Rossmiller."Research on EffectiveSchools: Implications for LessDeveloped Countries. CompaiwtiveEducation Review, 31(3): 377-99, 1987. Conde, E. "Educationand High-levelEmployment: The Case of Latin AmericanWomen." Statistical Bulletinof the OAS, 8(1-2):1-31, 1986. CostaRica. Educaciony Formacionde la Mujeren Costa Rica, 1980. Deble,I. TheSchool Education of Girls.Pairs: UNESCO,1980. Dettmer,J. 'Nicaragua:La RevolucionSuperior en la Educacion."Revista Latinoamnericana de Estudios Educativos,13(1): 103-30, 1983. ECLAC. El Desatrollode Amenica Latina y Sus Repercusionesen la Educacion . Cuadernos de la CEPAL No. 41. Santiago: ECLAC, 1982. ----. Five Studies on the Situation of Women in Latin America. Santiago: Estudios e Informes de la CEPAL,1983. Mujeres------.Jovenes en America Latina. Santiago:ECLAC, 1985. The------. Decade for Women in Latin America and the Caribbean.Santiago: ECLAC, 1988. Ehrlich, M. "ParentalInvolvement in Education:A Reviewand Synthesisof the Literature."Revista Mexicanade Analisis de la Conducta, 7(1): 49-68, 1981. Elissetche,M. "UPE in a Latin AmericanPerspective." Intemational Review of Education,29(2): 1955- 1958,1983. Farrell, J. and E. Schiefelbein."Education and StatusAttainment in Chile:A ComparativeChallenge to the WisconsinModel of StatusAttainment." Comparative Education Review 29(4):490-506. 1985. Fernandez,H. "Women'sEducational Situation in Peru."INIDE, Lima,1986. Filp, J. and E. Schiefelbein."Efecto de la EducacionPreescolar en el Rendimientode Primer Grado de Primaria El Estudio UMBRALan Argentina,Bolivia, Colombia y Chile" [The Effect of Pre- S c h o o 1 Educationon Achievementin First Grade in PrimarySchool: The UMBRALStudy in Argentina, Bolivia,Colombia and Chile).Revista Latinoamericanade Estudios Educativos, ^'t): 9-42,1982. Finn, J., J. Reis, and L Dulberg."Sex Differences in EducationalAttainment: The Process."Comparative Education Review, 24(2): S33-S52,1980. Fuler, B. 'Raising SchoolQuality in DevelopingCountries: What InvestmentsBoost Learning?" EDT DiscussionPaper, 2. Washington,D.C.: World Bank, 1986. 'Is PrimarySchool Quality Eroding in the Third World?"Comparative Education Review, 30(4): 491-506,1986. .WWhat School Factors Raise Achievement in the Third World?"Review of EducationalReseachm, 57(3):255-92, 1987. Fuller, W., C. Manski,and D. Wise. "NewEvidence on the EconomicDeterminants of Postsecondary SchoolingChoices." Journal of Human Resources, 17(4): 477-98, 1982. Latin America 173

Gomez Campo, V.M., and M. Pena. "La Educacion Superior en Colombia: Analisis de Tres Modalidades Recientes."Revista Latino Amencana de Educacion, 17(2): 47-77, 1987. Halpern, R. "Effects of Early Childhood Intervention on Primary School Progress in Latin America." ComparativeEducation Review, 30(2): 193-215,1986. Hansen, D. "Age Variations in the Formation of Educational and Occupational Career Goals of Brazilian Youth: A Cross-Cultural Test of the Wisconsin Model." ComparativeEducation Review, 2i: 37-50, 1977. Hernandez, A. "Formal Education and the Dominican Woman."Santo Domingo: Autonomous University of Santo Domingo, 1986. Heyneman, S. and WA. Loxley."The Effect of Primary School Quality on Academic Achievement Across Twenty-NineHigh and Low Income Countries."Ametican Journalof Sociolog, 88(6): 1162-94,1983. IDB (Inter-American Development Bank). Econom.ncand Social in Latin America. Washington, D.C., 1980-81. ------The Financingof Education in Latin America Washington, D.C.: IDB, 1978. Altemativas---...... de Financiamientopara las PrioridadesEducativas de America Latina. Washington, D.C., 1980. Economic--..------and Social Progressin Latin America. Washington, D.C., 1987.

Irwin, M., and others. "The Relationship of Prior Ability and Family Characteristicsto School Attendance and School Achievement in Rural Guatemala." Child Developmen4 49: 415-27, 1978. Jallade, J.P. "Education and Economic and Social Development in Latin America." In The Financingof Education in Latin America. Washington, D.C.: IDB, 1978. Jamison, D., B. Searle, K Galda, and S. Heyneman. "ImprovingElementary Mathematics Education in Nicaragua: An Experimental Study of the Impact of Textbooksand Radio on Achievement."Journal of Educational Psychology,73(4): 556-67, 1981. Jiimenez,E. "The Structure of Educational Costs: Multiproduct Cost Functions for Primary and Secondary Schools in Latin America." Education and Training Report, 8. Washington, D.C.: World Bank. Jimenez, E. J.P. and Tan. "Selectingthe Brightest for Post-SecondaryEducation m Colombia: The Impact of Equity." Education and Training Report, 61. Washington, D.C.: World Bank, 1987. Kaneko, M. "The Educational Composition of the World's Population: A Database." Education Training, Report, 29. Washington, D.C.: World Bank, 1987. King, E., and R. Bellew. "EducationPolicy and SchoolingLevels in Peru." Draft. Washington, D.C.: World Bank, 1988. Klees, S. "Televisionas an Educational Medium: The Case of Mexican SecondaryEducation." Comparative Education Review, 23: 82-100, 1979. "The---. Role of Public Policy in the Education of Girls and Women." Educational Evaluationand PolicyAnalysis, 9(3): 219-30, 1987. Komenan, A. "World Education Indicators an Annex to Improving the Efficiency of Education in Developing Countries."Education and Training Report, 88. Washington,D.C.: World Bank, 1987. 174 Latin America

Lembert, M. "The Impact of Mother's Beliefs, Expectations and Attributions on Children's Primary School Dropout: The Case of Low SocioeconomicStatus Families in Urban Mexico."Paper presented at the American *EducationalResearch Association, 1986. Levy,D.C. "LatinAmerica's Private Universities:How SuccessfulAre They."Comparative Education Review, 29(4): 440-59, 1985.

Levy, M. 'Determinants of Primary School Dropouts in Developing Countries." ComparativeEducation Review, 15: 44-58, 1971. Lewin, H. "The Urban Woman of Rio de Janeiro in the University System."Rio de Janeiro: Cesgranrio Foundation, 1976. Lynch, P. "School and Family Predictors of Achievement and Dropout in Elementary Schools of a Developing Country."Paper presented to the American Educational Research Association, 1975. McMahon, W. "Potential Resource Recovery on Higher Education in the Developing Countries and the Parent's Expected Contribution."Economics of Education Review, 7(1): 135-52,1988. McGinn, N., E. Schiefelbein,and D. Warwick."Educational Planning as Political Process: Two Case Studies from Latin America." ComparativeEducation Review, 23: 218-39, 1979. Ministerio de Educacion y Cultura. "LaIdeologia Patriarcal: el Rol de la Educacion."Montevideo: Instituto de la Mujer, 1988. Molina, D. "Educacion,Trabajo e Igualdades de la Mujer en Colombia."Empleo y Desempleo, 2(4), 1977. Muelle-Lopez,L. "La Eficiencia Interna del Sistema de Educacion Basica: Estado del Arte en America Latina' [The Internal Efficiencyof The Basic Education System: State of the Art in Latin America]. La Educacion, 28: 162-93, 1984. OAS (Organization of American States). Los PerfilesEducativos en Amerca Latina. Washington, D.C., 1984 ------."Status of Women in the Americas at the End of the Decade of Women (1976-1985)."Series Studies 14. Washington, D.C., 1985.

Papanek, H. "Class and Gender in Education Employment Linkages." ComparativeEducation Review, 29(3): 317-46, 1985. Peek, P. 'The Education and Employment of Children: A Comparative Study of San Salvador and Khartoum." Population and Employment, Working Paper #33, ILO, 1976. Piotti, D. "Mujer Joven y Educacion en el Uruguay" [Young Women and Education in Uruguay]. A paper presented at the Seminario Taller de Politicas sobre la Mujer Joven en Latinoamerica, Montevideo, 1988. Plank, D. "The Expansion of Education: A Brazilian Case Study." ComparativeEducation Review, 31(3): 361-75, 1987. Psacharopoulos,G. "The Economics of Higher Education in DevelopingCountries." Comparative Education Review, 26(2): 139-59,1982. ------. "Returns to Education: A Further International Update and Implications."Joumal of Human Resources, 20(4): 583-604,1985. Latin America 175

"----.Public versus Private Schools in DevelopingCountries Evidencefrom Colombiaand Tanzania." Educationand Training,Report, 33. Washington,D.C.: World Bank,1986. Psacharopoulos,G. and A. Zabakza."The Destination and Early Career Performanceof SecondarySchool Graduatesin Colombia."Washington, D.C.: World Bank,Staff Working Paper, 653, 1984. Psacharopoulos,G. and F. Steier.'Education and the LaborMarket in Venezuela,1975-1984." Economics of Education Rewew, 7(3): 321-532,1988. Psacharopoulos,G. and A.M. Arriagada. "The Determinantsof Early Age Human Capital Formation: Evidencefrom Brazil."Economic Development and Culudl Change,37(4): 683-708, 1989. Ram, R. 'Sex Differencesin the Labor MarketOutcomes of Education."Conparative Education Review, 24(2):S53-S77, 1980. Rama, G. "Educaciony Sociedaden AmericaLatina." La Educacion,101(2): 45-66, 1987. Rosemberg,F. "Education:Democratization and Inequality."BraziL Carlos Chagas Foundation, 1986. Sanguinetti,J. *AcademicAchievement School Qualityand FamilyBackground. Study in SevenLatin AmericanCountries." Paper presented at the Comparativeand InternationalEducation Society conference,Atlanta, Georgia,1983. Schiefelbein,E. "La InvestigacionSobre Calidadde la Ensenanzaen AmericaLatina." La Educacion28: 88-116,1984. "EducationCosts and FinancingPolicies in LatinAmerica." Education and TrainingReport, 60. Washington,D.C.: World Bank, 1987. Schiefelbein,E. and J. Farrell. "Selectivityand Survivalin the Schoolsof Chile."ComUarative Education Reki 22: 326-41,1978. -. -----"Women, Schooling, and Work in Chile: Evidencefrom a LongitudinalStudy." Comparatve EducationReview, 24(2): S224-63, 1980. "Educationand OccupationalAttainment in Chile:The Effectsof EducationalQuality, Attainment, and Achievement."Amencan Joumal of Educaton, 92(2):125-62, 1984. Schiefelbein,E., and others. "FmancialImplications of Changesin Basic EducationPolicies." In The Financingof Educationin Latn Ame,ica. Washington,D.C.: MDB,1978. Schiefelbein,E., J. Farrell and M. Sepulveda-Stuardo."The Influence of SchoolResources in Chile."Staff WorkingPaper, 530, Washington,DC: WorldBank, 1983. Schiefelbein,E., F. Sanchez,and G. Galvan."Caracteristicas de la Educaciony el Analfabetismoen Siete Paisesde AmericaLatina." Revista LaI_oame4cana de Estudos Educahivos,13(1): 87-102, 1983. Simmons,J. and L Alexander.'The Determinantsof SchoolAchievement in DevelopingCountries: A Reviewof the Research."Economic Development and Cltural Cange, 26:341-57, 1978. Smock,A.C. Women'sEducaton in DevelopingCoaunties. New York: Praeger Publishers,1981. Stein,W. "Mothersand Sonsin the Andes:Developmental Implications." Paper Presentedto the Annual Meetingof the AmericanAnthropological Association, 1972. 176 Latin America

Stromquist, N. "Empowering Women Through Knowledge: Policies and Practices in International Cooperation in Basic Education." Draft. School of Education. Stanford University.A report prepared for UNICEF, 1986. "School-Related---. Determinants of Female Primary School Participation and Achievement in DevelopingCountries: An Annotated Bibliography."Education and Training Report, 83. Washington, D.C.: World Bank, 1987. Tedesco, J.C. "Paradigmsof SocioeducationalResearch in Latin America."Comparative Education Review, 31(4): 509-532,1987. Tienda, M. "EconomicActivity of Children in Peru: Laboi Force Behavior in Rural and Urban Contexts." Rural Sociology,44(2): 370-91, 1979. UNESCG. "Study on the Equality of Access of Girls and Women to Education in the Context of Rural Development."Paris: UNESCO, 1973. Female------. Palicipation in HigherEducation, Paris: Division of Statistics on Education, 1975.

"Comparative------. Analysis of Male and Female Enrollment and Illiteracy."Paris: UNESCO, 1980. El .------.Acceso de la Mujer Venezolana a la Ensenanza y la Forniacion Cientifica y a las Carreras Correspondientes." 1981. ------. "Evolutionof Wastage in Primary Education." Paris: UNESCO, 1984. "Female------. Participation in Higher Education." Paris: UNESCO, 1985. Statistical------.Yearbook Paris: UNESCO, Various years. "Trends------. and Projections of Enrolment by Level of Education and by Age 1960-2025(as assessed in 1989). Paris: UNESCO, Office of Statistics, 1989. Velasquez, M. "Does ForeignEducation Benefit Rural Women? The Case of Mexico."Mexico City: Colegio de Mexico, 1987. White, K, M. Otero, M. Lycette, and M. Buvinic.Integrating Women into DevelopmentPrograms: A Guide for Implementationfor Latin America and the Caribbean.Washington, D.C.: ICRW, 1986. WVier, D. 'The Distribution of Educational Resources in Paraguay: Implication for Equality of Opportunity."Comparative Education Review, 24(1): 73-86, 1980. Wolfe, B. and J. Behrman. "Who Is Schooled in DevelopingCountries? The Roles of Income, Parental Schooling, Sex, Residence and Family Size."Economics of EducationReview, 3 (3): 231-45,1984. Woodhall, M. "Student Loans as a Means of Financing Higher Education." Washington, DC: World Bank, Staff Working Paper, 599, 1983. Latin America 177 App,ndix Table 5.1 Educational Attainment in Latin America and the Caribbean (for all regions)

Total Prinwm Entred Post Populadon % No Incompkew Complete Secondary Secondary Secondary Count Sex/Age (millons) Schooling School S-l School S-2 School

Antigua and Barbuda Total All 0.06 15.0 79.2 -) 45 -) 1.3 (1970) FemaleAll 0.03 14.4 80.6 -) 4.1 -) 0.9

Argentina Total 25 + 14.91 7.1 33.4 33.0 20.4 -) 6.1 (1980) Female 25+ 7.71 6.7 32.1 35.2 20.1 -) 5.8

Barbados Total 25 + 0.12 0.8 635 -) 32.3 -) 3.3 (1980) Female 25 + 0.07 0.9 65.2 -) 32.0 -) 1.9

Belize Total 25+ O.O5J/ 10.7 75.3 ) 11.7 -) 2.3 (1980) Female25+ 0.02 10.5 76.6 -) 11.7 -) 1.2

Bermuda Total 25 + 0.031/ 2.0 43.8 -) 46.8 -) 7.4 (1970) Female 25+ 0.01 1.3 415 -) 51.0 -) 6.2

Bolivia Total 25 + 1.76 48.6 28S -) 10.8 7.1 5.0 (1976) Female25 + 0.92 62.2 20.7 -) 8.2 5.6 3.3

Brazil Total 25 + 48.31 32.9 50.4 4.9 6.9 -) 5.0 (1980) Female 25+ 24.58 35.2 48.8 4.6 7.2 -) 4.1

Chile Total 25 + 5.20 9A 56.5 -) 26.9 -) 7.2 (1982) Female 25 + 2.72 10.0 56.9 -) 27.1 -) 5.9

Colombia All 20+ 8.48 22.4 55.9 -) 18. -) 3.3 (1973) Female20+ 4.48 23.7 56.0 ) 18.5 -) 1.8 Costa Rica Total 25 + 0.66 16.1 49.1 17.8 6.3 4.9 5.8 (1973) Female25+ 0.33 16.0 49.8 17.7 65 4.5 5.4 Cuba Total 25-34 1.41 1.9 15.8 24.7 49.9 -) 7.8 (1981) Female25-34 0.71 1.8 19.0 26.5 46.4 -) 6.3

Dominica Total 25+ o.03aJ 6.6 80.5 ) 11.1 -) 1.7 (1981) Female25+ 0.01 6.8 81.6 ) 10.6 -) 1.0 Dominican Republic T ^. 25+ I.1SA 40.1 41.6 4.3 9.6 25 1.9 (1970) Female25+ 0.56 42.8 40.9 3.9 8.7 2A 1.3 Ecuador Total 25+ 2.89 25.4 17.0 34.1 8.1 7.9 7.6 (1982) Female25+ 1.46 29.6 16.8 31.1 8.3 8.7 5.6

El Salvador Total 10+ 3.13l/ 30.2 60.7 - 6.9 -) 2.3 (1980) Female10+ 1.64 33.1 58.3 -) 6.6 -) 1.9 .... continued 178 Latin America AppendixTable 5.1 Educational Attainmentia lt:n America and the GCaibbean (for all regions) ..... continued

Total pmv Entered Post Populaton % No Incomple Complet Secondary Secondaoy Secondery County Sux/Age 'milons) Schooling School S-l Schoo S-2 School

Guatemala Total 25-34 0.62 91.6 -) ) 6.7 -) 1.7 (1973) Female 25-34 0.2S 92.5 ) ) 6.7 -) 0.9 Guyana Total 25+ 0.27 8.1 72.9 -) 173 -) 1.8 (1980) Female 25 + 0.14 10.6 73.0 -) 15 -) 0.9

.aiti Total 25-34 0.68 68.1 203 -) 10.8 -) 0.8 1982) Female 25-34 0.37 73.1 17.2 -) 9.1 -) 0.6

londuras Total 25-34 17.1 60.6 -) 6.6 103 49 1983) Female 25-34 ..... 16.7 61.1 -) 6.6 11.6 4.0

Jamaica Total 25+ 0.703/ 3.2 79.8 -) 15.0 -) 2.0 (1982) Female ?S+ 0.37 3.0 79.4 -) 15.8 -) 1.8 Mexico Total 25+ 24. 1 34.2 31.4 17.2 11.8 -) 53 (1980) Female 25+ 12.' b 37.1 30.7 18.0 115 -) 2.7 Panama Total 25+ 0.73 18.3 27.1 23.2 11.7 115 8.3 (1980) Female 25+ 0.36 19.1 26.3 23.1 11.6 122 7.8

Paraguay Total 25+ 1.14 14.2 51.0 15.34 16.0 ) 3.4 (1982) Female 25+ 057 17.9 49.5 1S.63 14.8 -) 2.5 ru Total 25+ 6.53 24.0 273 17.2 10.7 10.7 10.1 981) Female 25+ 3.31 34.2 244 1M5 8.6 9.7 7.7 Puerto Rico Total 25+ 158 8.0 17.8 11.4 16.4 27.9 18.4 (1980) Female 25+ 0.84 9.1 17.8 11.9 1S.7 27.2 183

St. Kitts and Nevis Total ;0 # 0.02 1.1 29.6 -) 67.2 -) 2.1 (1980) Female 25+ 0.01 1.0 30.3 -- ) 675 -) 1.2

St. Lucia Total 25+ 0.04 175 745 -) 6.8 -) 13 (1980) Female 25+ 0.02 16.8 75.6 -) 6.8 -) 0.7 St. Vincent and Total 25+ 0.03 2.4 88.0 -) 82 -) 1.4 The Grenadines (1989) Female 25+ 0.02 25 88.6 -) 8.0 -) 0.9 Trinidad and Total 25+ 0.41 13 29.4 42.6 19.7 4.0 2.9 Tobago (1980) Female 25+ 0.20 13 29.9 42A 20.4 4.1 1.9

ruguay Total 25+ 1.59 62A -) -) 28.6 -)9.0

Venezuela Total 25+ 554 23.5 47.2 -) 22.3 -) 7.0 (1981) Female 25 + 2.80 26A 46.2 -) 21.9 -) SS

/ Detail may not add to 100 because of rounding urces: UNESCO, Stadstial Yearbook Paris: UNESCO, 1990. SouthAsia 179 Chapter 6. South Asia

Shabrukh R. Khan

South Asia is the region, alongwith Sub-SaharanAfrica, where girls' educationlags behind boys' most dramatically.Except for Sri Lanka,all South Asian countries have sharply lower primary school enrollment rates for girls than for boys,ranging from a 15 percentagepoint differencein Bhutan to more than 50 percentagepoints in Nepal. Overall,primary school enrollments in thesecountries expanded significantly from 51 percentto 78 percent between1960 and 1987. But the growth has been less spectacularthan in other regions,and SouthAsian expenditureson educationare lowcompared with other parts of the globe. At post-primaryeducation levels, South Asia has the largestgender gap of anyThird Worldregion. SouthAsian countrieshave one of the world'srichest mixes of religiousand culturalinfluences. India is overwhelminglyHindu (about 80 percent) vith Moslem,Buddhist and Christianminorities. Pakistanand Bangladeshare predominantlyMoslem; Bhutan, Tibet, and Burma Buddhist; Nepal Hindu; and Sri Lanka a mixof Buddhismand Hinduism.The "jewelin the crown"of the Britishempire untilindependence, the count ies' educationsystems were patterned after Britain's,but in the past twodecades have taken on more of a mixof Americanand Britishschool characteristics, and in twc Communist-ruledIndian provinces are shapedafter Sovietschools. All the countrieshave tried in recentdecades to introducereforms to their Britishsystems (set up by the colonialiststo "train a nationof bureaucraticfunctionaries") to make them more attuned to indigenouscultures. Grindingpoverty is seen as the biggestbarrier to educationin SouthAsia, makingthe direct costs of schoolingand the opportunitycosts of foregonechild labor too expensivefor manyfamilies. SouthAsian countriesare amongthe poorestin the world,suffer someof the highestinfant mortalityrates, and have relativelylow urbanization levels. In no other ThirdWorld area is femalelife expectancyequal to or lower than that of males, a measure of the harsh conditionsthat womenof the region face. Amongthe five countriesexamined in this chapter--Bangladesh,India, Nepal Pakistan,and Sri Lanka--percapita GNPs range from$160 to $400.Education levels are drasticallylower in rural areas than in cities,in part because of povertyand becauseof poorer accessto schools. Culturalfactors such as the customsof early marriageand dowries,and concernfor younggirls' physical and moral welfare,also limit the educationof girls and women. Islam and Hinduismboth exert some restrictiveinfluences on girls' educationin the sense of callingfor wide-spreadsegregation of the sexes, veiling,and seclusionof women,although these practicesvary considerably by region. The Hindu caste systSm,despite India's constitutionaldemocracy, indirectly constrains educational opportunities of lower caste children. Constitutionally,all Indiansare equal. To increaseequity of accessto educationalservices and the job market,the governmenthas institutedquotas in educationand publicservice jobs. But the governmentreforms have not been as effectiveas intended. Teachersmay unconsciously treat low-caste childrendifferently, or havedifferent expectations for them. Culturalpractices, however, can be alteredby economicconditions. In the rural southof India,for example,families educate daughters to increasetheir chancesof marryingwhite-collar husbands who mighthelp the familyin case of a famine. In Nepal,on the other hand,poor familieseducate one son throughpart or all of secondaryschool to equip him for a white- collarjob.

Thischapter reviews the state of women'seducation in SouthAsia and identifieseducation and social policy changesdesigned to increasefemale schooling. Because the determinantsof primaryschooling are different from those at higherlevels, especially the teriary level,terdiary education is treated separately. First,the socioeconomicprofile of thosewho obtainhigher education in SouthAsia maybe differentfrom that for lowerlevels because access to highereducation is more limited.Second, admissions restrictions at the postsecondarylevel imply a supplyconstraint. Third, highereducation is probablymore directlyrelated to the 180 South Asia

labor mi set than is primary schooi and is also more important with respect to budgetary requirements given its high private and social costs.

The studies reviewed here vary greatly. They include general studies based on secondary materials, village studies, analyses of purposivelyselected samples, and studies based on well-designedprobability samples. Thus, some information is more representative of the whole country than is other information. Appendix A provides more details about the kinds of data these studies used.

In general, even when using data generated from sample surveys, few analyses go beyond one-way crosstabulations. Like the nonrepresentative studies, the sample survey studies essentiallygenerate hypotheses instead of specific information useful for resource allocation decisions.

Economic and Educational Setting Economic Conditions

The amount of resources a countr; corfamily can devote to female education strongly depends on the country's general economic and social development. Based on World Bank classification,all five of the countries reviewedhere rank in the low-incomegroup. Their recent economic growth rates have been fairly robust. As table 6.1 illustrates, their gross domestic products have grown faster in recent years than that of the higher-incomecountry groups. But except for Pakistan, the GDP growth has been slower than the average for low-incomecountries.

Urbanization, which economists view as part of the structural change that accompanies economic development, varies widely among the five countries. Pakistan is the most urbanized with 31 percenmof its population living in cities compared with 9 percent in Nepal and 13 percent in Bangladesh. All five, like the reference group countries, lag far behind higher income countries in urbanization. Life expectancyand infant mortality are significantlylinked to literacy. Sri Lanka clearly outranks the other four countries in literacy, with 80-90 percent of the population able to read as of 1981, and its infant mortality and life expectancy rates are more favorable even than those of the average middle-income country. For the remaining four countries, literacy rates are less than 20 percent in rural areas and, except for India, less than 40 percent in cities; infant mortality is correspondinglymuch higher and life expectancy sharply lower than in Sri Lanka.

Nutrition is another variable that interacts with education and literacy. Higher literacy provides knowledge about nutritional intake and higher average nutritional intake may Lncouragethe acquisition of schooling. Pakistan's per capita daily calorie supply (an indicator of nutrition) is close to that for Sri Lanka. In fact, the daily caloric intake for the other three countries are not far below the 2,500 requirement specified for a 150-poundadult male engaged in moderate activity. Activitylevels in these countries, however, are more likely to be strenuous than moderate, and hence the needed caloric intake probably far exceeds that for people in the middle- and high-income country groups.

' An implicit assumption here is that the major issues are sufficientlysimilar across countries. Where important differences exist, they will also be mentioned.

2 Material on Sri Lanka is scarce. Three separate data-based searches produced few current published or unpublished works for Sri lanka. This is unfortunate because Sri Lanka is the dear front runner in edncational progress in South Asia. A review of studies attempting to explain why this is the case would be eful, although the explanation is implicit in the recounting of the deterrents to female education in oth. wuthAsian countries in different cultural settings. SouthAsia 181 Worth notingtoo is that whereasthe life expectancyof femalesexceeds that of males in all of the country incomegroups, in SouthAsian countries (including Bhutan but excludingSri Lanka),female life expectancy is the same as or even lowerthan male life expectancy.In no other countriesis this true, and it hintsat the relativelyharsher conditions for SouthAsian females. Populationgrowth, external indebtedness,and defenseexpenditures also influencea country'sability to expandfemale education. Exceptfor SriLanka, these countrieshave experiencedpopulation growth rates that matchor exceedthat of the referencegroup. OnlySri Lanka and Indiahave succeeded in slowingtheir populationgrowth; in fact, Nepal'saverage annual population growth rate for the 1980-87period exceeded that for the 1965-80period. Nepal'sexternal borrowing apparently has been relativelyrestrained--its debt-service burden, at about 10 percentof its exports,is verylow comparedto that of the other four countriesand of the other country groups. The drive to develop rapidly despite limited resources and heavy populationpressure has contributedto heavyand increasingforeign indebtelness for mostdeveloping countries, and the other four countriesin this group are no exception. 182 South Asia Table 6.1 Basic SocioeconomicIdicators

SouthAsia Low- LOW Indicaw MAgh Mide fddle Group Bangladesh India Nepal Pakinan Sn Lanka

GNP per capita (1987$) 14,430 2,710 1,200 290 160 300 160 350 400 Avg. annual GDP grwtb rate 1965480 0.8 6.7 5.7 SA 2A 3.7 1.9 5.1 4.0 1980-87 2.8 3.4 2.1 6.1 3.8 4.6 4.7 6.6 4.6 Uzban population (%of total) 77 66 51 30 13 27 9 31 21 Life expectancyat birth (ym) Females 79 69 66 62 So 58 S0 54 73 Males 73 64 61 60 51 58 52 55 68 Infant mortality (per 1,000 lve births) 10 S0 61 76 119 99 128 109 33 Daily calorie supply per capita (1986) 3,375 2,9110 2,777 2,384 1,927 2,238 2,052 2,31S 2,401 Avg. annual population growtb rate (%) 1965-80 0.9 2.1 25 2.3 2.8 2.3 2.4 3.1 1.8 1980-87 0.7 1.9 2.3 2.0 2.8 2.1 2.7 3.1 15 Debt service ratio (% of exports) 1970 1 10.6 12.6 na 0.0 222 32 235 10.9 1987 58/ 25.8 21.7 1S.7 24.2 189 9.7 25.9 19.2 Defense sbare of central government expenditure (96v) 1972 21.8 11.9 115 na 5.7 26.1 7.2 39.7 3.1 1987 14.9 na na na 10.0 215 12.1 295 9.6 Education sbare of centrd govemment expenditure (%) 1972 Oa 7.7 17.7 na 14.8 2.3 7.2 1.2 13.0 1987 4.6 na 13.1 na 10.6 2.7 6.2 2.6 7.8 na m not available Note: Al data are for 1987 except a noted. A/ For non-OECD countries only.

Sourwe: World Bank, WorldD.etopment Rpon 1989

In additionto large externalcommitments, the substantialproportion of their nationalbudgets that these countries spend on defense appears to be strainingtheir capacity to undertake social development. Althoughdefense expenditures are not necessarilyinversely related to educationspending, yet, as illustrated in table 6.1, such a negativerelationship emerges for each country. Despitethe fairlyhealthy performance of economic,and somesocial, indicators for Pakistanand India, and their lowliteracy rates, these countriesspend relatively little on education.In 1987,they dedicated less than 3 percentof the centralgovernment's budget t- -ducation; as a fractionof GNP, educationexpenditures representedonly 0.6 and 0.5 percent,respectively. Althougb Bangladesh spent over 10percent of its central budgeton education,this againrepresented only 13 percentof GNP. Comparedto low-incomedeveloping SouthAsia 183 countrieswhich on &.erage allocate13 percentof their governmentbudgets, or 33 percentof their GNP, to education,these are indeed very low levels of spendingfor the sector. Sri Lanka"sstellar social developmentappears to have been hamperedby politicalturmoil in 19&7;this partly explainsits rising burdenof foreignindebtedness and defenseexpenditures, and its shrinkingsupport of education.

Education Ivels SouthAsian countries'progress in expandingfemale primary and secondaryschool enrollments in the past three decadeshas been mixed(table 6.2).3 In Sri Lankawhere gross enrollment rates at the primarylevel were already high in 1960,growth rates have been minimalthroughout the period. Contrast this with Nepal'sexperience of two-digitgrowth rates. The resulthas been that the shareof girlsin total enrollment about tripledin Nepal from 13 to 32 percent (table63). In Pakistan,the femaleshare increasedfrom 21 to 33 percent. Bangladeshmoved up to a par withIndia by raisingthe femaleshare from 28 to 40 percent. India, continuingits push for universalprimary enrollment, increased the female portion from 33 to 40 percent. Girls in Sri I anka movedeven closerto paritywith boysin primaryenrollment, accounting for 48 percent of the total, up from 46 percent. Table 6.2 Average Annual Growth Rtatesof Female Enrollment

P km_a Seconda=, 1960-70 1970-80 1980-87 1960-70 1970-80 1980-87

Bangladesh 59 6.0 3.0 23.1 72 8.2 India 6.5 3.0 4.4 9.5 5.8 9* Nepal 13.9 19.7 155 10.0 11.0 7.5 Pakistan 9.1 5.2 5.7 13.9 9.5 8.4 Sri Lanka 0.3 2.4 -0.2 6.9 3.2 7.4O

OVMost recent data is availableonly for 1985and excludesteacher training k/ This refers onlyto generalsecondary education. Source: UNESCO(1989) and computedfrom data in UNESCO,Stadsdcal Yearbook (1990).

3 Twu sourcesof secondarystatistics have been used to demonstratethe nature of gender differentialsat the primaryand higherlevels. The firstsource is the latestUNESCO data (publishedin 1990).This source has been complementedby censusand surveydata fromthe countriesthemselves; these data werecompiled on behalfof the U. S. Departmentof Commerce,Bureau of the Census,aid distributedas data tapes by the Inter-universityConsortium for Politicaland SocialResearch (ICPSR). In additionto disaggregations by region and age, these tapes contain such informationas 'age at first mariage and 'labor force participation,both of whichcan influencefemale education. The drawbackin usingthese data for South Asia is that they are not current. They do, however,provide useful benchmarks and indicateimportant trends. Neitherthe originalsources nor the Consortiumbear any responsibilityfor the analysesor interpretations presentedin this text based on tables drawnfrom these sources. 184 South Asia Table 63 Gross EnrollmentRates at Primary and SecondaiyLevels

GrossEnrollment Ratio Percente Female of Total 1960 1987 1960 1987 Primary Bangladesh 56 59 28 40 Nepal 11 86h 13 32 Pakistan 25 40 21 33 SriLanka 101 107 46 48 Secondary Bangladesh 7 18 8 31 India 16 39PJ 23 34/ Nepal 6 30' 10 27 Pakistan 10 18 15 28 Sri Lanka 39 71 40 51

at 1986 / '1988

Source: UNESCO, Staistical Yearbook(1990), and UNESCO (1989).

Except in Sri Lanka, the gender differential at the secondary level also remains large despite very steep increases in female enrollment throughout the region. In Nepal the female share increased from 10 to 27 percent, in Pakistan from 15 to 28 percent, in Bangladeshfrom 8 percent to 31 percent, and in India from 23 to 34 percent between 1960 and 1987. If the annual growth rate in the 1980s continues, female enrollment wonld reach 64 percent in Nepal in 1987, and 41 percent in Bangladesh.

Enrollments result in the educational attainment of the population. In particular, compare the schooling levels of adults in the potential labor force (table 6.4). In Pakistan, nearly 90 percent of adult women aged 25-64 did not complete any school grade, compared to 66 percent of adult men. Bangladeshand India have only a slightlybetter record than this. Nepal and, most especialy, Sri Lanka have done better h respect to this indicator.

Note also that education levels vary greatly by region within each country, both for women and men. In Bangladesh,& erent of rural women have no schooling compared to 68 percent of urban women. This disparity measured in percentage points is larger between rural and urban men--62 percent versus 40 percent. But although a large fraction of rural men have no schooling, this fraction is still lower than for urban women. In the other countries (except India), however, urban women appear to be at least as educated as rural men. Examiningurban-rural educational levels thus reveals an aspect of the gender gap problem that should influence policy making.

Expanding educational opportunities have also improvedSouth Asia's very low literacy levels (figure 6.1). The female literacy rate in rural Nepal rose from 2.7 percent in 1971 to 7.8 percent in 1981, and in rural Pakistan from 4.2 percent to 73 percent. The literacyrates of urban women in those countries were at least sever umes higher than rural rates in 1971, and at least four times higher in 1981. The literacy rates of rurai women in Bangladesh rose from 11.5 percent to 15.3 percent and in India from 13 to 17.6 percent. South Asia 185

Bangladesh's urban female literacy rate was more than twice the rural rate, and India's more than triple the rural rate. Table 6.4 Percent Distnrbution of Adults Aged 25-64 by Highest Education Level Attained About 1980

Females Males Country Total Urban Rural Total Urban Total

Banv3adesh No Schooling 83.2 68.2 85.5 57.5 39.7 61.7 Primary Level 12.0 15.6 11.5 21.1 20.1 21.3 Secondary Level 4.4 14.5 2.9 19.1 32.1 15.9 Postsecondary 0.3 1.8 0.1 2.4 8.0 1.1

In&i No Schooling 83.2 57.6 Primary Level 7.9 15.6 Secondary Level 7.6 22.4 Postsecondary 1.2 4.3

No Schooling 36.4 27.8 39.0 41.2 24.9 43.8 Primary Level 41.6 24.5 46.6 26.3 16.3 27.9 Secondary Level 17.1 32.3 12.7 25.0 34.4 23.5 Postsecondary 4.9 15.3 1.9 7.o 24.4 4.8 Pakistan No Schooling 89.1 71.2 96.0 65.8 44.5 75.3 Primary Level 5.2 11.5 2.8 12.8 14.9 11.8 Secondary Level 4.8 143 1.2 18.1 31.9 11.9 Postsecondary 09 2.9 0.1 3.3 8.6 1.0 Sri Lanka No Schooling 20.1 10.4 22.8 8.3 5.2 9.5 Primary Level 45.6 39.4 47.3 51.2 39.9 55.6 Secondary Level 33.5 48.3 293 39.0 51.6 34.0 Postsecondary 0.9 1.9 0.6 1.5 3.4 1.0

9/ No rural or urban breakdown available for this period. Source: United Nations, Women's Indicators and Statistics database (1988). 186 SouthAsia

Figur 6.1 Female Iiteracy Rates

Figure 5.1: Female Literacy Rates

Bangladeah, 1971 i 1981

Indli, 1971 I . I 1981 !

Nepal. 1971 1981 . . . *

Paklistan, 1971 i 1981 1

Sri Lanka, 1971 i . -_ 1981 _II _ 80 60 40 20 20 40 a0 80 Rural Urban

Womencontinue to be under-representedin highereducation in SouthAsia (table 6.5). In Bangladeshand Pakistan,women account for less than one-fifthof post-secondarystudents, in Nepal,about one-fifth,and in India, about 30 percent. Sri Lanka,once again the front runner, boasts a 41 percentshare of women enrolledin post-secondaryprograms. The patternsof specializationof femalestudents are similarin these countries(table 6.6). The share of femalestudents in educationscience and teachertraining, and in the humanitiestend to be amongthe largest. In SriLanka, women comprised the large majorityin education and the arts, in additiontc homeeconomics and businessadministration. In Nepal and Sri Lanka,women alsoappear to be wellrepresented in medicineand other health-relatedfields, with their sharereaching 44 percent. SouthAsia 187 Table 6.5 Womens Share in Higher Education Enrollment

LeveiG Total 5 6 7 Bangladesh (1988) 17.2 -- 17.6 12.2 India (1985) 29.9 24.5 30.7 305 Nepal (1980) 21.6 na na na Pakistan (1986) 183 - 17.1 25.5 Sri Lanka (1986) 40.8 373 44.1 53.0

-- = not applicable na = not available / Level5 = Diplomaor certificate. Level6 = Bachelorsdegree. Level 7 = Postgraduate degree. Source: UNESCO, StaWisticaYeabook (1990).

Table 6.6 Women's Share in Higher Education Enrollment by Country, Field of pecialization,and Educational Level (percent of total enrollment)

Field Bangladesh India Pakdsian Sn Lo*& Nepal (1988) (1987) (1986) (1986) (1980)

AIDFields 17.2 29.9 183 40.8 21.6 Educationscience & teacher training 9.7 47.0 43.5 643 30.1 Humanities,religion & theology 203 39.5 30.0 0.0 32.1 Fie & appliedarts 0.9 683 0.0 75.7 a Law 6.7 10.0 8.1 44.0 4.9 Social& behavioralscience 17.2 0.0 A/ 48.0 a' Commercial& businessadministration 13.9 21.8 14.9 55.5 11.9 Mass communications& documentation 0.0 33.7 0.0 0.0 0.0 Home economics(domestic science) 100.0 0.0 100.0 100.0 0.0 Servicetrades 0.0 0.0 0.0 0.0 0.0 Naturalscience 16.5 32.4 22.8 43.2 12. Mathematicsand computerscience 18.2 Q/ 1 / Medicine& health-relatedsciences 26.2 32.8 33.7 44.4 44.4 Engineering 3.9 8.7 2. 19.0 2.4 Architectureand town planning 23.2 a/ 35.8 0.0 Trade, craft & industrialprogrammes 0.0 a 0.0 2.4 0.0 Transportand communications 0.0 0.0 0.0 0.0 Agriculture,forestry & fishery 6.6 8.0 0.2 33.0 0.0 Other & not specified 0.0 24.6 16.4 37.6 0.0 Note: Where zero percentageare noted,size of enrollmentreported was nil. a' The figureimmediately above includes data for this field of study. Source: UNESCO,Statical Yearbook1987. 188 SouthAsia Factors AffecftngEducation Quality and Reach and Attainment A Framework for AssessingEvidence oh schooling Deisions Many factorsaccount for the large gendergap in educationin SouthAsian countries. Althoughmost of the studiesreviewed here do not explicitlyapply a conceptualmodel, the analyticalmodel presented in Chapter 1 capturesthe implicitbehavioral framework underlying these studies.4 The postulateunderlying the frameworkis that households(or families)act as rationalwelfare-maximizing agents, the limiton family welfarebeing set by familyresources. In the contextof developingSouth Asian economies,"iewing the familyas poolingthe resourcesof its membersand sharingconsumption and investmentdecisions seems reasonable.For expositiontlsimplicity, the factorsthat affecteducational choices can be groupedinto two broad sets: familyand communityfactors, and schoolfactors. Educationis both a consumptionand an investmentdecision that the householdmakes. This decisionis influencedby prices,wages, unearned income, child-specific and family characteristics, school characteristics, and variouscommurnity factors, including social infrastructure and sociocultural norms. Thesefactors affect the cost of schoolingwhich inclu' es both the opportunitycost of sendingchildren to schoolrather than havingthem work for wagesor at home, and the direct cost of schoolingwhich refers to fees, books, transportation,and any other school-relatedexpenditures. In addition,culture may be of particular importancein determiningthe acquisitionof schooling.If uneducatedgirls are perceivedculturally as better bridesor if girls' future earningswill go to the familiesthey marryinto, educatingthem mightadd littleto familywelfare. The literature,however, provides evidence that attitudesand preferencesare themselves likelyto differ withthe incomeand educationlevel of the family. To the extent that the costs differ for girlsand boys,there will be gender differencesin education. In addition,costs and incomethat do not differfor boysand girlsin a familymay have different effects on the education,nevertheless, because of tastes and attitudesthat governbehavior. In order to examinewhich factorsaffect males and femalesdifferently, it is necessaryto estimateseparate demand functions for them. For policypurposes, estimating separate demand equationsfor femaleswould be imperativein order to predicthow changesin the availabilityof schoolsor textbooks,or increasesin familyincome, might alter femaleenrollments or achievement.

Primay and SecondaryEducation Familyand Communiy Factors Most studiesmention poverty as the majorreason families fail to enrolltheir childrenin, or withdrawthem premature!yfrom, primary education. 6 Thissuggests that familiesfind both the directand the indirect,or opportunity,costs of schoolingdifficult to bear. Related to povertyis the demandfor femalechild labor to take care of siblingsand to do householdand farm work. Authorswriting on this issue emphasizethat

4 The semilal workson this topic are Beckerand Lewis(1974) and de Tray (1974). Birdsall(1985), pp. 25-35;Rosenzweig and Evenson(1977); and Rosenzweig(1980) have contributeduseful expositions and estimatiorsof such a modeL Of particularrelevance is King and others (1986),in whichthe model is extendedto explainand subsequentlytest the intergenerationaltransmission of gender differentials. s This has been establishedby Rosenzweigand Evenson(1977, p. 1076),and by Rosenzweig(1980, p. 18), usingtwo separate data sets. 6 For example,this is mentionedto be the case for Bangladeshby Qasem (1983),p. 21 and Islam (1982, p. 34);for India by Tara (1981,p.178); for Pakistanby Anderson(1988, p.1) and Shahand Eastmond(1977, p.14);and for Nepal by UNESCO(1980, p.99). SouthAsia 189 the demand for girls' labor for domesticwork is muchgreater than that for boys' labor.7 Jamisonand Lockheed (1987, p. 283) cite studies showingthat the demand for girls' labor in Nepal exceeds that for boys' by about 50 percent. Papanek(1985, p. 334)points out thatfor Bangladeshthe age-specificfemale activity rates for the youngestage group(10-14) were the highestof all agegroups and for malesin that age group the lowest. Tara (1981,p. 74), notingthat takingcare of siblingsusurps much of a younggirl's time, recommendsthe developmentof someform of communityday care. In addition,labor-saving technologies, in principle,can make timefor schooling.Seetharamu and Ushadevi(1985, p.101) suggest that the provision of a safe and stablewater supplyshould be a priority,because bringing water can occupyup to four hours of a girl's day. These policyrecommendations typically are not viewedas directly in the domainof educationpolicy as are schoolfactors, which also influence schooling. Rosenzweig (1980, p. 18)showed that in India,women's and girls' (but not boys')labor is interchangcable,so a 10percent rise in femalewages reducedgirls' school attendance by about 5 percent. Participationin the pa½dlabor force,however, is often a poor measureof the labor demandfor girls. The expectationthat poorer rural familieswill require their sons'wives to workat homediscourages some householdsfrom enroling girlsin schoolbecause they wouldthen becomeless desirableas housewives. Educationis perceivedas corruptingthe traditionalattitudes of femalesand causingthem to be less willing to do physicallabor, according to Seetharamuand Ushadevi(1985, p. 61) and Desai (1987,P. 17) reporting for India,and Shrestha (1986,p. 31.) for Nepal, Smock(1981, p. 91) reportsresults of an attitudinalsurvey in Pakistanin whicheducation was perceived as makingfemales self-centered, defiant of parentalauthority, and uninterestedin householdaffairs. Thus,schooling imposes a socialcost in additionto the directand opportunitycosts.

Familiesmay lack interestin or be openlyhostile to formal educationof their daughtersfor other reasons. Singhal(1984, p. 367) citesa studyconducted by the NationalCenter for EducationResearch and Training in India whichfound that domesticwork, marriage,betrothal, and parental indifferenceaccount for 55 percentof total wastagein girls' educationat the middlelevel. Clason(1975-76, p. 182)reports that poor rural parents in Nepal viewfemale education as immoral; a similarattitude was reported in UNESCO (1975,p. 37). In culturallyconservative enviromnents, adolescent girls may be viewedas morallysuspect if theycontinue going to school.The youngage at first marriageand the importanceof preservinga girl's good reputation in such cultures lead to widesprad withdrawalof females from school at puberty, particularlyif they are attendingcoeducational schools. 8 Early marriageis perceivedto be such a barrier to femaleschooling that one writerin a UNESCOstudy (1980, p. 12) suggestedlegislation to forbidearly marriage.

Table6.7 reportsminimum legal and actual agesof marriagein four of the fivecountries during the mid- 1970s. The legalminimum age of marriagefor femaleswas either fifteenor sixteen. In Bangladesh,75 percent of rural ever-marriedfemales had married by the age of seventeen. For Pakistanand India, 75 percentwere married by ages 22 and 19,respectively. In contrast,in Sri Lanka, only25 percentof rural ever-marriedfemales married before the age of 21 and 50percent were olderthan 23. In all cases,women in urban areas tend to marry later. Also in all cases,the legalage of male marriageis higherby at least two yearsthan for females,and the actual age of marriagefor both rural and urban men is considerably older. Thus,marriage may not havedeterred male educatior as muchas it discouragedfemale education.

7 For additionalevidence on this issuesee Chamie(1983, p. 32). 8 See Caldweliand others (1985,p. 68) for India,Anderson (1988, p. 6) for Pakistan,and Qasem(1983, p. 6) and Papanek(1985, p. 334) for Bangladesh. 190 South Asia Table 6.7: MinimumLegal Age at Mariage and Age ActuallyMaried by Country, Sex, and Region

Tota! Uban Rural PerwentEver Manied Male Female Male Female Male Female

Bangladesh (1974) Minimum legal age 18 15 18 15 18 15 25% 21 14 23 15 21 14 50% 24 15 26 17 23 15 75% 27 17 28 19 27 17 Mean singulate age at marriage5J (1981) 23.9 16.7 India (1971) Minimum legal agel1 18 15 18 15 18 15 25% 19 14 21 16 18 14 50% 22 17 25 19 22 16 75% 26 19 28 21 25 19 Mean singulate age at marnage/ (1981) 23.4 18.7

Pakistan (1973)bI Minimum legal age 18 16 18 16 18 16 25% 22 17 22 18 21 17 50% 25 19 26 20 25 19 75% 29 22 30 23 29 22 Mean singlate age at marriage' (1981) 215 17.9 Sri Lanka (1971) Minimum legal ageLJ 18 16 18 16 18 16 25% 25 20 26 20 24 20 50% 28 23 29 24 1 23 75% 32 27 34 29 32 27 Mean singulate age at ma age§) (1981) 27.9 24.4 flMinimum legal age tcvsed to 21 in 1976. /sample xcludes tribal areas and the Malakand Divisionin the North-West Frontier Province. S/ Legal marital ages enacted in 1978. 4/Source of data is UN Demographic Yearbook 1982. Sources. Bangladesh-BangladeshBureau of Statistics,Population Census of Bangladesh,1974, National Volume,Repo.t and Tables, table 5, )acca, 1977; Population Information Program, series M, no. 4, table 15, November 1979 (age at marriage). India-(minimum legal age) World Health Organizatiun,World Heath, Aug.-Sept. 19r.1,p. 6; (age % ever married) India Registrar Gereral, Ccnsus of India, 1971,Social and Cultural Tables, series 1, part ff-c, table c-1l, derived at the U.S. Bureau of the Census by fittinga makeham model to data from the 1971census. Pakistan-Census Organization, Pakistan, Housing, Economic and Demographic Surey, 1973, vol. 11,part 1, table 2; Katherine Peipmeier and Elizabeth Hellyer, 'Minimum Age at Marriage: 20 Years of Legal Reform,' in Peopk, vol. 4, no. 3, 1977. Sri Lanka-Department of Censusand Statistics,Census of Population, 1971,vol. 2, part 1, table 8; Population Information Program, Populaton Rqport series M, no. 4, tIle 15, Johns Hopkins University,November 1979. Data made available by the Inter-university Consortium for Politicaland Social Research (U.S. Department of Commerce, Bureau of the Ceasus), (ICSPR no. 8155, WomJm un Developmens, IV, 1963. South Asia 191

Almost al} South Asian cultures are conservative,but the manner in which Islam and Hinduism have been adopted normaly leads to an even more restrictive environment for female education. According to Gunawardena (1987, p. 8), the Muslim commurty's progress in education in Sri Lanka has lagged behind that of the Sinhalese, Tamils, and Burgers. The iaportance of cultural conservatismin discouragingfemale education can be overstated, however. Chamie (1983, p. 2) challenges the conventional view that Islam contnbutes to low enrollment rates among girls by citingthe high rates in Libya and Bahrain. Rosenzweig's results (1980, table 4) imply that being Muslim was not a significantbarrier to female enrollments in India. Sarkar (1986,p.88) found mixed evidence and conduded that the hypothesisthat Muslims are opposed to educating both males and females does not hold. This evidence suggests that unidimensional views are inadequate in explaininggender differentialsin schooling.

Cultural norms, as reflected by parents' attitudes about educating daughters, can affect enrollment rates. In an empirical study addressing female education in Asia, King and others (1986, pp. 56-59) tried to ascertain how much of the gender gap in enrollment results from social norms and the attitudes of parents and how much from differences in individual characteristics. To do this, the authors computed a family- specific discrimination index. They found that for families with daughters, if daughters were treated similarlyto sons, their educational level would rise by 65 percent in middle-classurban Lahore (Pakistan), by 129 percent in lower class urban Lahore, and by 224 percent in rural Punjab. Mothers' attitudes also may influence their daughters' education. Smock (1981,p. 61), citinga village survey in Pakistan, reported that only 10 percent of the village women supported the notion of equality of opportunity for women.Shah (1986,pp 246-47)reported the results of a surveyin Pakistan whichfound that among households that owned no assets, 51 percent of the urban mothers in the sample and 58 percent of the rural mothers believed that religious education (equivalentto about one year of formal schooling)was enough for their daughters. Many studies have reported that rural familiesperceive the formal education curriculum as useless.9 In one villagestudy in Bangladesh,Khatun (1979,p. 267) asked about the parenz' perception of a useful education for girls; an overwhelmingmajority desired to see child care, cooking, and handicrafts in the curriculum. This is consistent with the role anticipated for females when they marry into another family. In fac;, such training may enable girls to be more eligible brides, and hence more likelyto marry into an economically and socially powerful family. At the same time, if the returns from a family's investment in educating daughters accrue to another household rather than to the investorhousehold, 10 the familyhas little incentive for investing in female education.

Family opposition to secondary education for girls is much greater than to primary education because the direct costs are higher and the girls are already of marriageable age. Unexpectedly,a study on Bangladesh found little difference in parental attitudes towards the continuation of daughters' or sons' education at the primary level (FREPD 1981,p. 109). However,91 percent of the household heads wanted their sons to go on to the lower secondary level, while only 61 percent wanted their daughters to do so. In an interesting study on Nepal, Ashby (1985, pp. 70-72) poinrs out that poor families develop a strategy of educating one son in the family up to an upper secondaryor higher level so that he may obtain a white- collar job. Her resLlts indicate that additional women and young girls in the pool of family labor significantlyincrease the amount of schooling male children get (p. 78). In contrast, Caldwell and others (1985,p. 33) report that in rural South India the case is sometimesthe opposite. Growingpressure on land, technologicalchange, and a trend toward hiring individual labor rather than whole families have produced excess labor in families. As a result, girls are allowed, and are even encouraged, to go to school as a famine-fightingstrategy. Girls with at least some level of education would be a better match for white-

9 See UNESCO (1980, P. 61).

10 Such a viewis reported by Qasem (1983,p. 21) for Bangladesh (study based on two purposivelyselected villages), Shah (1986, p. 253) for Pakistan, and Junge and Sharesta (1984, p. 70) for Nepal. 192 SouthAsia collarhusbands, who may help the familyduring a famineor who at least -'c ld not be a liability.This suggeststhat changingeconomic conditions can alter deep-seatedcultural practices. 11 Culturalpractices do not prevailacross all cross sectionsof societyuniformly. Parental attitudestoward education and marriage prospectsdiffer dramaticallyby class. Middle-classfamwes may viewsome educationas promotinga good marriagefor girls,because the womencan then managetheir households more efficiently.12 Becausemarriages are perceivedas an aliance of familiesrather than as a commitment of two individualsto each other, some educationfor femalesmay conferpositive welfare if it raises the probabilityof a strongsocial alliance. Thus, the expectationsfrom a gi "'smarriage and concomitanteffects on her educationcan be expectedto differby incomegroup. Lowermiddle-class families have reason to regard their daughters'education as costly. Womenin South Asia are usuallyexpected to marrymen withmore educationthan theyhave. In culturesin whichdowries are customary,securing a more highlyeducated husband for an educatedwoman would require a larger dowry--another"hidden" cost of educatingfemales. 13 Seetharamuand Ushadevi(1985, p. 61) recommend an antidewrycampaign in the mediaand a policylinking a dowrysubsidy to the educationalattainment of girls. A family'sincome level and the educationlevel of the parents or other adult familymembers appear to be related positivelyto the educationof the girlsin the family.A higherincome level enables the familyto bear both the direct and indirectcosts of education.Rosenzweig (1980, p.1 9), usingmultivariate analvsis and the appropriateseparate equations by gender, foundland size and unearned incometo be positively and significantlyassociated with rural femaleenrollments in India. Ahmedand Hasan (1984,pp. 2-3),using data from a probabilitysample surveyin Bangladesh,presented simple two-way crosstabulations to show that girls' educationvaries positively with their family'sincome and land ownership. The parents' educationlevel can also lead to highereducational attainment among the daughtersbecause more educatedparents mayhave a more enlightenedattitude towards female education or providea more stimulatingenvironment for such education.14 For Bangladesh,Islam (1982,p. 35) reported a high correlationbetween girls' enrollmentand the proportionof adult householdmembers who are educated. Thisevidence is corroboratedby Ahmedand Hasan (1984,pp. 2-3)who reported that 91 percentof children from the most educatedfamilies in their sample survey,those with eight years of educationor more (4 percentof the familiessampled), were enrolledin school,while only 12 percentof boysand 7 percent of girls from illiteratefamilies were in school. For Pakistan,Shah (1986,pp. 246-4)cites a studyindicating that about two-thirdsof illiteraterural womenwanted only religious education for their daughters,while approximatelythe same proportionof rural womenwith up to sixyears of educationdesired up to higher secondaryeducation for their daughtersand 17 percentwanted them to obtain a collegeeducation.

1 One can also argue that some culturalpractices evolve as a result of economicimperatives. 12 See for exampleJunge and Sharesta(1984, p. 70) for Nepal,and Ahwad and others (1978,p.7) for Bangladesh.Also, a changein norms in the upper incomegroups mayeventually filter down,possibly in conjunctionwith economic changes. A countrystudy of Indc. in UNESCO(1980, p. 67) reports that even poor villagersview female education as a meansof enhancingsecurity and marriageprospects. This is not the opinionexpressed by most authors,however. 13 See Shresta(1986, p31) for Nepal;Seetharamu and Ushadevi(1985, p. 61.),and Desai(1987, p. 10)for India, and Lavador(1986, p. 92) for Bangladesh.The mostfrequent referencesto this phenomenonwere for India. 14 For example,a studyof secondaryfemale students in Egyptfound that the daughtersof more educated mothersreceived greater encouragementin their studentsthan thoseof less educatedmothers (Bach and others 1985). South Asia 193

Using a large 1979 data set for Pakistan, Irfan (1985) explored the intergenerational effect of education. r rosstabulationsdemonstrated that within all income groups, the relationship between the education of the head of the household and the enrollment rate for girls (aged 10-14) in the family was positive (p. 33).2 In a multivariateanalysis, Irfan used enrollment ratios as the dependent variable in separate regressions by gender to explore the relative effects of demand and supply factors (p. 36). On the supply side, the existence of a middle school or a high school in the community proved to be a significant influence on enrollment in nonfarm households,but not in farm households. On the demand side for both farm and nonfarm households,income of parents could be a significantbarrier. For farm households,land ownership was an important factor, whereas for nonfarm households,the educational level of the household head was significant. For both groups of households, the literacy rate in the village during the last census period (1972)also proved to be an important determinant. That education generates education is evident not only from the significanceof the variable representing the education level of the head of the household but also from the independent influence of the general level of literacy of the village. This finding is also confrmed by elasticities computed at means by King and others (1986, p. 52) from equations estimating the determinants of education. In a cross section of families in Pakistan, the father's education had a significantimpact and was among the most powerful influences on the education of both sons and daughters. Although the elasticities are approximatelyequal for the sexes in middle-class urban families in Lahore (.30 and .29 for sons and daughters, respectively),they are much higher for daughters than for sons in lower class urban familiesin Lahore (.32 and .18, respectively)and in rural Punjab families (.32 and .12, respectively).16 Thus, an educated father may play a more important role in the education of daughters than of sons in lower class urban and rural families. The mother's schooling was not as influential as the father's; the elasticity for mother's education was significantonly for females belonging to middle-classurban families in Lahore (.14). Although the conservatismof a lower class urban or a rural family may be gauged by male attitudes (or male educational level as a proxy), the social attitudes of middle- and upper-class familiesmay well be determined by the educational level of the mother. Thus, the significanceof the elasticity for females belonging to middle-classurban families is interesting.

School Fators

School location, facilitiesfor female students and teachers, curriculum,and examinationpolicies are among the various school-related factors that can contribute to gender differentialsin enrollments. These factors can influence parents' decisions to educate their female children. The importance of school location may vary within and among countries. In the analytical framework guiding this literature review, location, or the distance to the nearest school, is often used as a measure of school supply, and thus of the cost of attendance, in studies estimating school demand functions. The hypothesisis that, other factors held constant, enrollments should have an inverse relationship with distance; that is, the shorter the distance to school, the greater the likelihoodthat females will attend. Many authors have reported distance to school to be an important barrier to female education.17 Even a short distance may seem long to some parents, however. Islam (1982,p.36), on the basis of evidence for Bangladesh,argued that increasing the number of schoolswill not necessarilyfoster greater enrollments. She cited a survey that interviewed 208 female dropouts, 84 percent of whom lived within a mile of the school. Similarly,Sattar (1984,p.13) argued that access is not the issue for India, because 90 percent of the children have access to a primary school or to a primary section in a secondary school within a kilometer

15 The page references given in this paragraph conform to the revised version of Irfan's report and hence may not tally exactlywith the citation in the bibliography.

16 All the elasticity coefficientsare significantlydifferent from zero at the 10 percent level.

17 See Caldwell and others (1985, p.41) for India, Shah (1986,p. 255) for Pakistan, and UNESCO (1980, p. 99) for Nepal. 194 South Asia of their homes. Ahmed and Hasan (1984,p.26) reported that enrollmentis negativelyassociated with distancein Bangladeshbecause parents may be unwillingto allowgirls even to crossa majorroad or a river on the way to schooL For Nepal, with its ruggedterrain, Clason (1975-76,p.182) reported that the remotenessof schoolscan be a major cause rf lowfemale enrollment. Jamisonand Lockheed(1987, pp. 298,301) did not find distanceto be an importantdeterminant of schoolparticipation. They admitted, however,that the extentof schoolavailability for the samplethey were usingwas atypicalfor Nepalas a whole;also, they did not estimateseparate equations by gender,thus statisticallyconstraining the effectto be identicalfor malesand females.Distance can be a greaterdeterrent to secondaryeducation of girlsthan to primaryeducation, because often one secondaryschool must serve several villages UNESCO (1975, p.36). As publiceducation facilities expand in SouthAsia, the variationin schooldistance will continue to decline and the supplyof schoolplaces will no longerbe as constraininga factor. This suggeststhat other school factorsbesides supply would loom larger as constrairtsto enrollment.There is need to considerthe impact of those other factors. A lack of some basic facilitiesmay affectgirls schoolattendance more than boys. Qasem (1983,p. 38) pointedout that 71 percentof rural schoolsand 53percent of urbanschools in Bangladeshhad no latrines, a problemthat could discouragefemale attendance. In fact, Ahmed and Hasan (1984,p. 58) found that familieshave withdrawn girls fromschools lacking latrines. In Pakistan,many parents feel uncomfortahle about enrollinggirls in institutionswithout solid and highboundary walls, which provide privacy (Anderson 1988,p. 6). The culturalconcern for the privacyof girls engendersalso the need for sex-segregatedschools. Other studiesfrom Bangladesh and elsewhereindicate that parentsare concernedabout a lackof separatescheols for girls.18 Parentsdesire segregationeven at the primarylevel, so the lack of segregatedfacilities at the secondarylevel may be an even more seriousbarrier to continuedfemale education. Testing fo.' the effect of separate schoolingfacilities, boundary walls, and latrines on the reductionof female enro';'nentsat pubertywould be of great value for policy. However,as Andersonhas pointedout, coeducatioiL a d fac reality in Pakistan at the primary level and the expenseof segregatedschools with scparate administrationsis unwarranted.19 Islam(1982, p. 67) indicatedthat after the nationalizationof schoolsin Bangladesh,the educationaladministration strongly discouraged segregated schools, and the statisticsimply a downwardtrend in the numberof such institutions. The samecultural forces that createthe needfor single-sexeducational facilities also result in broadsuppoit for employingfemale teachersto teach girls.20 The Governmentof Nepal in conjunctionwith several internationaldonors has undertaken,since the early19 70s,an extensiveproject to providewomen with equal accessto education. The key strategyof this projecthas been to recruit and train femaleteachers from variousregions, including remote areas, wherethey couldthen serveas teachers. A midwayevaluation report of the program(UNICEF 1978, pp. 25-33)showed it he,, somesuccess in encouragingand retaining femaleenrollments. Pakistanhas also experimentedwith traininglocal femaleteachers. Marker and Gah (1985)describe a "mohalla^(home) school project in Baldia,a large squattersettlement in Pakistan. In additionto training teacherswith a high schooleducation chosen from amongwomen within the community,the projecthas emphasizedcostcutting by holdingclasses in homes and by not requiringuniforms or shoes. At lea-. initially,the project has experiencedgreat success. By the time the report was written, there were 64 schoolsand an enrollmentof 1,600.The successof the projectwas aaributedto the hard workof the two-

18 See for exampleFR EPD (1981,p.89) and UNESCO(1980, p. 44) for Bangladesh,the latter sourcefor India (p. 61), and Shah (1986,p. 255) for Pakistan. 19 See Nayar(1985) on this issue for India. 20 This issuehas probablyachieved the greatestconsensus in the literature. See for exampleUNESCO (1980,p. 15). South Asia 195 member project team in involvingand gaining the support and participation of the local communities (p. 9).21

The wisdom of recruiting female teachers locally is not universallyaccepted, however. A monitoring study by Qadir (1986, np. 18-19)in Bangladesh found that, although local women are able to communicatewith and gain acceptance by villagers, local teachers (induding women) are chronicallyabsent because of their household chores. In addition, they were primarily concerned with the income from private tutoring and were reported to practice favoritismtoward these students being tutored.22 Also, well-connectedteachers continuallypestered the local administration for a transfer to town schools. Perhaps for these reasons, villagers said they opposed local area teachers (pp. 36-38). Drawing and retaining female teachers from outside the village poses another set of problems, however,since they would be required to relocate, to gain local acceptance and, in particular, to find suitable accommodations.23 Another school issue is the rigid examinotionpolicy, which may affect girls more adversely than boys.24 Because girls are under more pressure to carry out household and farm work, they are absent more often than boys. Their subsequent failure in examinations causes the family to perceive that the educational investmenthas soured and to withdrawthem.25 Repeating a grade, along with late entrance and withdrawal at puberty, results in low schoolingattainr. ent.26 Sattar (1984, p.17) has suggested compulsoryenrollment at the prescribed age to ensure more years of schooling. Possibly,forced withdrawal of girls from school by their parents may become more difficultthe greater the number of years of schoolingthey have attained; certainly, this hypothesis is worth testing. In its camqaign to increase girls' school enrollments, India has experimented with a host of school-related incentives.7 In the analytical framework being used here, these are aimed at reducing the direct cost of education. These policies include providing attendance scholarships for girls, free textbooks, stationery, uniforms, and midday meals. Seetharamu and Ushadevi (1985,p. 68) suggestedthat the Indian public has limited awareness of these incentives. Bangladesh has experimented with free uniforms and is currently

21 UNESCO (1980, pp. 112-116)also describes the mohalla program. Anderson (1988, p.7) is currently investigatingthis program and another one involvingmosque schools, partly to see whether they provide access to groups currently excluded or under-represented.

22 This is also a problem in Pakistan. Policymakersmay consider legalizingtutoring but restricting teachers to providing tutoring to students from other than their own schools.

23 Marker and Gah (1985, p.6) make.this point for Pakistan. In fact, not only is adequate accommodation not provided but female teachers are denied a rent allowance. The Indian government has tried special allowancesto attract rural female teachers and provided accommodationfor them (UNESCO 1980 r 58, 65). No account was available about the success of this scheme.

24 See Chamie (1983).

2s Sattar (1984,p.15) makes this point for India. This does not indicate that girls are poorer performers; in fact Khan and others (1986, p. 19) provide overwhelmingevidence from secondary data that girls in Pakistan have consistentlybeen outperforming boys at the secondary leveL

26 See Ahmad (1987, p.28).

27 These policies started with the Second Five Year Plan and continued until the Fifth Five Year Plan. The current policy is universalelementary education. It is argued that this policywill necessarily focus more on females since their enrollments are lower. An evaluation study of the special program for girls' education, undertaken by the Programme Evaluation Organizationof the PlanningCommission, concluded that where the special programs were weil planned and implemented, they did have considerable impact on female enrollments (UNESCO, 1980, p. 67). 196 South Asia

disbursing scholarships to girls attending secondary schioolsto defray par. of their direct costs.2 Qadir (1986, p.20) reported that both rural and urban households in Bangladesh felt a midday meal also would help increase enrollments. Bangladesh's Shawniwarmovement, although not focused exclusivelyon increasing female education, had that goal and the expansion of other forms of female activityamong its major objectives. Sattar (1981) has evaluated this program and views it as very successful. Shawniwar emphasized the importance of local participation to create demand for various governmentalsocial services such as education (p. 1). The girls' share in total enrollment in the Shawniwarvillages was 44 percent at the primary level, a significantlyhigher rate than the national average of 38 percent (p. 4). Also, village opinion leaders reported that the average enrollment of primary and secondary female students in the Shawniwarviulagcs increased by more than 50 percent after the movement began (p. 29). Local participation, apparently the key to the Shawniwar movement's success, was also emphasized in anotb,r study on Bangladesh. In a survey to evaluate Bangladesh's drive towards universal primary enrohment, Qadir (1986, p.18) noted that villagerswere suspiciousof corruption in the way contracts were issued for building schools and other social infrastructures. The villagets were willing,and believed they 2 were able, to participate in building such infras.ructures themselves.9 Two experimental projects have highlightedthe special importance of flexibleschool hours for encouraging female enroline t. If hours for schoolingdo not conflictwith the time giris are needed for domestic chores, the opportunity cost of female schoolingfor the family is reduced or eliminated. Naik (1982)reported that in a village school project near Pune in Maharashtra State in India, the key feature was holding classes between 7 and 9 o'clock in the evening, after householdchores and dinner. Parents supported the project and the communityprovided rent-free accommodations.Teachers were local primary school graduateswith some secondary education, intensivelytrained for one week after everysix-week period. Learning wasbased on small-groupteaching, with peer teachingheavily encouraged. Within one year, 75 percent of the children in the school could read fairly well, write a little, and do the expected arithmetic. The dropout rate was much lower than the average for formal schooling. Although marriage or betrothal still proved to be one of the main reasons for dropping out, 8 percent of girls who left the program did so to enter a regular school (p. 169). Junge and Sharestha (1984) have reported on a similar project aimed at low-castegirls in Nepal. School took place from 7:30 to 9:30 in the morning--beforehousehold chores began. Also like the Indian project, this one recruited and trained local teachers, and students started reading by using cards to learn whole words, chosen from their own life experience, rather than by memorizing the alphabet; enrollees were expected to read at the third or fourth class level in one year. Some girls did drop out from the project, but generally to pursue short-term cash opportunities, and hence the overall retention rate was excellent. There was concern, however, that uppercaste brahmins could oppose the subsequent enrollment of these girls in the local public school.

Curriculum content is another issue that may affect female schooling. Kalia (1980,p. s209) pointed out that as early as 1965 the Government of India, having nationalized preparation and approval of textbooks in 1955, proposed to create a curriculum conduciveto sex-role equality.Kaha's content analysis of forty-one Indian textbooks (in Hindi and English) in four states and Delhi showed males to be the exclusiveleading actors in 75 percent of the lessons, while females took precedence in only 7 percent of the lessons (p. s212). She found only seven bibliographic sketches of women but forty-sevensuch sketches of men (p. s213). She concluded that women were still being prepared for a role requiring only servitude and support (p. s223).

2 See Qadir (1986,p. 83) on the distributionof free uniforms, a policy reported to have been discontinued. Several organizations are involved in the scholarship scheme, including the Bangladesh Association for Community Education. USAID has dispersed mere than 20,000fellowships.

29 See also UNESCO (1980, p.18) on the provision of low-cost schooling built with community participation. SouthAsia 197

Singhal(1984, p. 367 alsocited evidencefrom India confirmingthe existenceof extensivesa ..tereotyping in Hindi textbooks. In contrast,Gunawardena (1987, pp.17-18) cited a textbooksurvey showing that in Sinhalaanu Tamiltextbooks, nonstereotyped sex rolesoutnumbered stereotyped roles. The oppositewas stronglytrue for Englishlanguage textbooks. The governmentsof SouthAsia cannot reasonablycontinue to talk about eliminatinggender differentialsin schoolingin their policydocuments while state-owned textbooksrestrict female motivaton to obtainan education. Perhapsdue to the increasein primaryschools in SouthAsia and the qualityof availabledata, viewsdiffer about the impact of schooldistance on femaleenrollment. Better informationneeds to be collectedon supplyconstraints to femaleeducation other than schoolavailability and distance. Data on the influence of specialfacilities such as latrinesand boundarywalls could help ferret out otner importantsupply-related barriers to girls' schoolattendance. Policymakers would benefit from knowinghow muchthese and other genderrelevant policies, such as sex-segregatedschools and femaleteachers, contribute to enrollmentand retentionof female students. In the samevein, it wouldbe importantto gaugethe marginalimpact of effortsto reduce the directcost of schoolingto poor families. Parents may welcomesuch incentives,but systematicevidence about the successof such policiesin increasingfemale enrollments is lacking.Similarly, the impactof flexibleschool hours needs to be investigatedsnore systematically.

Tertiary Edacation Only a very snA&ilproportion of womenin SouthAsian countries(except Sri Lanka) have any tertiary education. Amplee.idence stronglysuggests parental preference for continuationof a son'seducation as opposedto a daughter's.31 Khatun'sstudy (1979, p. 266)of a Bangladeshvillage reported that 75 percent of the parents desiredhigher educationfor boysbut only25 percentdid so for girls. Suchattitudes have also been documentedfor rural Sri Lanka;Smock (1981, p.91) reported findingsof a 1973study in which 60 percentof the respondentswhose children were currentlyin schoolwanted their sons to get a college educationwhile none had the same aspirationsfor their daughters. Rice and Wilber (1979,p. 58) found in or-eof the areas they surveyedin Sri Lanka that 73.1percent of the womenwanted their sons to have a universityeducation, while 76.6 percent wanted their daughtersto be "knowledgeable"but not university graduates.

Motivationsfor HigherEducation These studies elicitedthe opinionsof lowerincome groups, who would find the direct cost of higher educationeven more difficultto bear than the costsof primaryor secondaryeducation. Attitudes toward highereducation differ for the urbanmiddle and upperclasses. Jahan (1979,p5) notedthat the recognition that the post-Partitionyounger generation of Muslimmales preferred educatedwives provided the initial impetusfor the highereducation of womenin Bangladesh.As the cost of livingrises, middle-classwomen themselvesrecognize that theymay be expectedto be income-earnersfor the familyinto whichthey marry. Thus, like other levelsof schooling,higher education may have a marriage-relatedinvestment motivation,

30 See Smock(1981, p. 61) for evidenceof sex stereotypingin the curriculumin Pakistan. An interesting exampleis the referenceto FatimaJinnah only as the sisterof the founderoi the nationbut neveras a very prominentpolitician in her own right. 31 For additionalevidence on this issue,see for exampleIslam (1982,p. 37) for Bangladesh,who points out that familiesare evenwilling to mortgagea pieceof landfor the son's educationif need be. For Nepal, Junge and Sharestha(1984, p. 847) note a similarpieference for educatingsons. 198 SouthAsia althoughthe natureof this motivationmay be changingfrom simply providing good mothersto also training womento be a sourceof incomefor their in-laws. In a carefullyconducted survey of severalpopulations including young matriculates, Miyan (1979) found that 75 percentof the womenstated that their main motivationfor pursuinghigher education was to meet the increasedcost of lvingv Chaudhury(1977, p. 163),in a surveyof workingand nonworkingwomen found that almost80 percentof womcnwho were working did so for economicreasons.33 Husbands of working womenwere significantlyless educatedand had lower incomesthan husbandsof nonworkingwomen (p.162).Also, significantly,fewer working women had educatedfathers and more had a rural background (p.157-158).Chakrabarti (1977, p.179) pointedto the ever-increasingmatrimonitl columns in Indiaseeking workingwomen as brides. The growingenrollment of marriedwomen (22 percentof femalestudents) in Dacca University,reported by Islam (1982,p. 82), may also indicateeconomic pressure on single-income families. In Pakistan,an InternationalInstitute for EducationalPlanning (UNESCO)survey by Husainand others (1987)of severalpopulations, including graduate employees and students,showed that womenwere largely from a differentsocioeconomic background than men (p. 140). Crosstabulationsshowed that, among e.aployees,most of the womenwere from the middleclass, and a larger percentageof womenthan men belongedto the highersocioeconomic groups: forty-threepercent of the womenwere from upper-income families,while only 7 percentwere from lower-income families, compared to shares of 34 and 14 percent, respectivelyfor males. Amongfemale graduate students, 32 percenthad parentswith higher education degrees as opposedto 17 percentfor femaleeducated employees. Given the age differentialsin the samplesfor the employeesand students,this findingwas interpreted to suggestthat more womenfrom bektereducated backgrounds were enteringhigher education. Similarfindings from a studycited by Islam(1982, p. 82) indicatethat the bulk of DaccaUniversity female students come fromthe better educatedand wealthierfamilies. In Pakistan, more middle-and upper-classwomen may be attendingcollege because a collegedegree is nownecessary for a good maritalmatch. Singhal (1984, p. 367) also expressedthis view for India.

Bwieristo Hgher Eduation

Cukual Obstac In general,economic motivation could in timeerode the culturalrestraint on highereducation, but for now, culturalfactors are impediments.Desai (1987,p. 9) believesthat fallingfemale enrollmentsi in medicine result from the subsequentrequirement for the higher dowriesnecessary to marry a highlyeducated daughter to a highly educatedman. Anxietiesand reservationsabout female participatinain higher educationexisc in Bangladesh,where more womenare enteringthe labor market. Miyati 1979,p. 53) mentionedparental concern about the "freemixin of sexesin coeducationalschools, although parents se-o;n less concernedabout marriedwomen working with men. Parents,and eventhe womenthemselves, h. yr supportedthe idea of establishingseparate women's educational institutions (polytechnics in this case ' the parentssurveyed, 74 percentsaid theywould seriously consider sending their dazightersto a wom I's polytechnicschool should one become available. Apart from "moral"considerations, a female-e(nl; institutionwould circumvent the stiffmale competitionwomen currently face in tryingto secureadmission into coeducationalschools.

32 Papanek(1985, p. 3J' -Iso made this point. 33 Chaudhury(1977) p.-.med nonparametrictests but not at disaggregatedlevels of education. The educationallevels of the womensurveyed ranged from pre-high school graduation to bachelorsdegree and above. SouthAsia 199 Another obstacleto femaleparticipation in higher educationmay be the lack of hostel accommodation. Sincehigher education institutions tend to be concentratedin the major city or cities,long-distance travel or residingaway from hore is often necessary.Miyan (1979, p. 53) and Islam (1982,p. 81) observedthat the absenceof secure accommodationsdeterred young women from outside the metropolitanarea from pursuinghigher education.Rice and Wilber(1979, p. 52) reportedthat the lack of such accommodations restricteCthe enrollmentof womenir Sri Lanka'sKundasale School of Agriculturete 20 percentof total enrollments.

labor Market D is"a The direct cost to familiesof educatingdaughters may effectivelybe greater than educatingsons because sons are better able to contributeto their own support. Islam (1982,p.81) conjecturedthat parents are reluctantto let an unmarrieddaughter support herself whilesons routinely do so. Even if parentswere not the olstacle, labor market practicesmay prevent female students from becomingself-supporting. ;Womenmay simplynot be hired or, even if hired, may receivelower wages for the same work as men. Thus, labor marketdiscrimination can discouragewomen's education in two ways: (1) It mayprevent them from becomingself-supporting as students;(2) By reducingtheir potential future income,it also may dampenwomen's aspirations for careersand hencetheir motivationfor highereducation. rForall levels of education,but particularlyat the higher levelwhere the direct and indirect costs of educationare so much greater, the behavioralmodel of householdchoice can be complementedby the human capitalmodel in understandinggender differentials.The humancapital model viewsincremental expenditureon bighereducation as an investmentin the individualfor whicha return accrueslater in the form of incrementalearnings. The theory is that each incrementof educationleads to an incremental increasein productivity,which is rewardedin the job marketby greater earnings. If males earn more on averagethan femalesfor the same levelof education,the human capital model would predict a lower investmentin (or lowerdemand for) educationfor femalesthan for males. Kingand others (1986,pp. 33- 39) fruitfullycombined the householdchoice model with this insightfrom the humancapital model. Usingthe humancapital model Tilak(1980, p. 62) foundfor Indiathat mean earningsfor womenwere up to six times lower than those for men for the same level of productivity. Using the same model, Sambamoorthi(1984, p. 7) found discriminationto account for 21 percent of the male-femalewage differential.Raj (1982,p. 254) used Indian censusdata to showthat womenwere in lowerposts relative to men despite equal or better qualifications,that they tended to be concentratedin a narrow range of occupations,and that salarydifferentials widened with age. Raj also reportedsubtle sex discriminationin sel tion for conferencesand seminar committees;he added that womentended to receivelukewarm recommendations(p. 255). Rice and Wilber(1979, p. 57) pointedoi't that in the late 1970s,the repeal of Sri Lankanlegislation enabling employers to paywomen less than men failedto stem this practice. Singhal(1984, p. 363) cited a 1977-78Indian survey showing that the unemploymentrate amongwomen withhigher education was muchgreater than for males. Femaleunemployment rates were 44.8 and 35.9 percentin rural and urbanareas, respectively,while the correspondingunemployment rates for men were 19.7and 9.3 percent,respectively. Gunawardena (1987, p.19), using secondary data for Sri Lanka,reported an unemploymentrate among higher educatedwomen of 23.8 percent comparedto 6.5 percent for comparablemales. Also,in 1985,23.6percent of universityfaculty was female, even though the 1984female share in enrollmentwas 42.2 percent. In Pakistan,however, the 1980population census (Government of Pakistan,1984, pp. 132-35)showed an unemploymentrate of 254 percent for highlyeducated females comparedwithHP LaserJet SerieslIHPLASEl.PRSof men, female representationin administrativeand relatedcapacities was practicallynonexistent. The PakistanLabor ForceSurvey for 1986-87(1987, p. 141) found that womenconstituted about 17 percentof the professional,technical and relatedworker category but only3 percentof the admitrative and managerialworker category. About 5 percent of all women workingwereprofession,'/technical workers and fewer than one percentwere in the administrator/manager * eategory.t 200 South Asia

Not all the findings or opiniionsaboL educated women's labor force participation are negative. In an opinion surveyof middle-classeducated and fecund women conducted in Bangladesh by C:haudhury(1977, pp. 180-181),only 22 percent of the working women perceived sex discrimination on the job market with regards to spiary and promotion, even though 67 percent cited low salary as th_.ir reason for job dissatisfaction. Forty-eight percent of the women thought there was no discrimination in hiring; although Chaudhury suggested that this low percentage indicated that women were not in positions for which men would compete.

Miyan (1979, p. 1) noted that although the Bangladeshgovernment has imposed mandatory quotas for the employmentof women in the public sector. these quotas were not being fulfilledprimarily because qualified candidates could not be found. However, n Sri Lanka, Rice and Wilber (1979. p. 47) found that quotas were limiting the number of educated women hired. Once the quotas were met, instead of being an instrument of affirmative action, they set a ceiling on the number of women hired.

Some evidencesuggests that the extent of discriminationdeclines as women progress to higher professional positions. Tilak (1980, pp. 62-63) in fact found some negative discrimination coefficients at higher professional levels (for example,doctors) in India. Women were earning more than men, and their higher earnings could not be ettributable to greater productivity. As in other conservativ34cultures,this could be explained by the fact that most women are allowed to see only female physicians. A crosstabulation of mean earnings of highly educated employees by gender and occupation in a UNESCO survey study in Pakistan edited by Husain and others (1987, p.210) demonstrated that women wtre earning significantdy more than men in the scientificand medical occupations,significantly less in engineering and technology, and the same in teaching.

The positive evidence, while encouraging,is mild compared to the bulk of evidence suggestingstrong labor market discrimination,which in turn can deter women from pursuing higher educatic i. Also, no evidence has yet surfaced that men are willingto accept any significantchange in their domesUicroles. Even if they encourage women to enter the labor market, they continue to expect the women to shoulder the bulk of the domeFticwork load (Raj 1982,p. 256-57for India). The pressure to continue to bear the primary share of domesticresponsibility constrains careers of professionalwomen because they prefer convenientlocations, flexiblework hours, and vacations to higher positions (Desai 1987,p. 11).

Conclusions

This literature review has identified several factors that have discouragedfemale school and postsecondary enrollments in South Asia. Although this reviewcannot be considered a policy document that identifies the best way to allocate scarce resources for education at the national and local levels, it can be viewed as a starting point for rigorous research on factors that appear to be most detrimental to female enrollments and for which the greatest policy leverage exists.

Useful policy research must estimate the likelyenrollment increase from some Taka or Rupee input into specificprojects, such as providinglunches, free textbooks, or uniforms, or into broader programs such as media campaigns or providing child care. Such studies require extensive data bases. Much of the data needed would include information on various exogenousprice variables and on detailed family,school, and community characterist;cs. Because cultural norms are important in female enrollments, researchers may need to develop indexes (an index of conservatism,for example) to capture the effect of this variable. This approach would permit more accurate approximationof other effects in multivariate comparative studies, across countries or across different ethnic groups within one country or region.35

34 See Chapter 4 in this volume.

35 See King and others (1986, p. 46) for an empirical technique that avoids explicit use of such an index. SouthAsia 201 Raisingenrollments is not simplya matter of expendingresources on schools. Familyand community factors also affect female schooling. To some extent, effectivemedia campaignscan alter cultural constraintson femaleschoo!ing. Also, economicchanges can affect deep-rootedcultural traditions. In Bangladesh,for example,the nwd for a secondincome among the middleclass is makingfemale higher educationand labor force participationmore acceptable.In SouthIndia, because of changingagricultural labor marketconditions, one evolvingfamine-fighting strategy is to increasegirls' education so that theycan find a more suitablemate in the urban sector. In Pakistan,some employerscountered militant trade unionismby seekingout educatedwomen, who were cheaperto hire and more docilethan male union members. Althoughsegregati-n in the work placewas attempted,some contactbetween the sexeswas inevitabledespite culturalbarriers.36 Education itself appearsto have a "snowballingeffect" in that girls' educationalaccess and attainmentare positivelyrelated to the generalliteracy level of their villages,as well as the educationattainment of adult membersin their households,particularly the father. This literaturesurvey indicates that a reorganizationof resourcesor policychanges that require no resource input may sometimesgo just as far, if not farther, than la:ge and expensiveprograms. A massiveand expensivecampaign in Bangladeshto attain universalprimary enrollmenj was not rated a successdespite the large infusionof fundsfor materialsand infrastructuredevelopment. 7 The projects found successfulin raising femaleenrollment were all low in cost, and all entailed some measureof localparticipation. Although the idea of involvinglocal residents is not newto the development literature,it has not been tried extensively.In all the local-participationprojects in education,local communitywomen with at least a primary degree were trained as instructors. Althoughtraining is undoubtedlyindispensable, it is not necessarilyonly those with formal trainingwho have somethingto contributeto the local school. Among the major reasonscited for droppingout of schoolis the parents' perceptionthat the curriculumis irrelevantand that the childrenlack interest in it. Drawingon village womenwho possesshandicraft or domesticskills to demonstratethose skillsin schoolmay providesome side incometo thosewomen, improve community involvement in the schools,and pavethe wayfor girlsto gain accessto formalschooling. Some of the successfulprojects featured flexibility in schoolhours, which blurred the distinctionbetween formal and informalschooling. The Sri LankanMinistry of Educationin the late seventiesgave all rural schoolprincipals the authorityto determinea more flexibleschool holiday schedulein keepingwith the needsfor child labor in agriculture(Rice and Wilber1979, p. 15). None of these initiativesrequired large governmentoutlays. Instead,they relied on an awarenessof the real determinants,positive and negative,of parents'willingness to educatetheir daughters.

36 Weiss(1983, pp. 207-14)recounts the experiencein the pharmaceuticalindustry in the Punjabprovince in Pakistan. 37 See Qadir (1986). Appeai Table 6.1 Summay of Stdes Rc_wed

C_q"Y A*w(s) bd- -ao

Bangladesh Ahmed and Husan Sample nvey(1984) of four primaly villageschools Crotabulations Imp. i socifcconc. (1984) status on enrolments

ainudhury, RIH Sample sue (1974) of middass workng wome Non-pamtc 1-ermi Ants of female labor force partcon for (1977) (582) and non-workingwme (S48) stadsfa and di

1Rs , ) (1981) Stratified madomample indudig 2,480rural osehos Nn etrc Dete of dropouts and ItiStiCS caostabulations Miyan (1979) Survey (1979) of Vaious populations (parents, empbyers, Non-parametric Determmantsof roen pursung tehntcal educuon with institutbns) partly based or probability sampling statistics, partcular emphass oo parzital attiwes crosstabulations Qadir, SA (1986) Purposely seected sample of various populationsin eight Non -parametric Assessment of the success of drive to boost child villages statistics, enmollment crosstabulations Identifies parental and communityviews o n what would work better Sattar, E. (1981) Suney of 11 schools from 10 randomly selected villages Non-parametric Assessmentof a pnject referred to as Sbawnivarto see in 10 dispersed administrativeareas statistics, if educatiul performance was better than natonal crossiabulations average India Caldwell,and others Survey(1981482) of I large and 8 smaller hamlets in rural Crosstsbulations Parental atttudes to female scboolng and differential (1985) Karnataka causes of withdrawalby gender Sample induded 364couples and 1,294suIViving children Gould, L (1983) Sample survey (1974-76) of SSI Parsi households in Crosstabulations Sex discriminationin educationalattainment Gujvat among one includingurban and rural populations of the allegedly most %,rogressive communities in the subcontinent

osenzweig, Nt and 191 PopulationCensus Regression Analysisinduded determinants Evanson, R. (1977) of female childschooling Rosenzweig,M. 'Tird round sample (1975)of a three-year national survey Regession Analysis induded determnants of female child sdcooling (1980) of 4,000rural households Sambamoorthi,U. Probability samplesurvey of a small regionallabor market Regression Identified female labor market discrimination usiag (1984) eaming functions Seetharamu and ProbabDitysample survey in rural Karnataka Crosstabulations Identifying determinants of low fema!e educatianal Ushadevi (1985) attainment

.- continued Appendi Tabe 6.1

continued

CAV,y Auhfs) Dwha W Foas India Tara (1981) P abilty f villages in Tamhur district Crosstabuations Determinants o: high femak dropouts and low female enrollments Tiak (1980) Sample survey of 100 members of the W. Godavan' Regression Etimate district m Andra Prdesh earning ur to ident fema labor market discrimination Nepal Ashby (195) Systematicrandcm sampling drew 302 farm households Included regresson Causes from Kabur Pidancharkdistrict of low female educational attament and incuding effect of familv compositionanC size Jamison, D.T. and Samplesurvey data covering 795 household (15% sample Regression Determinants of adult attitudes Lockheed, NEL size) from 28 randomly selected villages in each and child schooling of six includinggender as independent variables (1987) punchayatsin two selected districts-Bara and Rautahat UNICEF (1978) Contacted beadmasters (38) male and female teachers Crosstabulations (31, 51), parents Evaluatingimpact ofa female teadhr prgrn on female (168) district education offes (9) and eroillments nuichayat members (26) Pakistan Hussain and othMrs Stratfied three-stag random sample of several Regression/ Includes (1987) populationsincluding employees genderdifferentials or varioushi.Aier educatonal and students crcsstabulat .-4 and empbyment issues [dn1, M. (1985) National population labor force and migration survey Regression/ Determants of rural child schooling (199) crosstabulations Sample included 1,208 girls from farm households and 133 from non-farm households King, E. and others Areas purposely selected and households selected via a Range of Indudes (1986) multi-stagedustered sample the intergwnerationaltransference of gender multivariate diffntials in schooling A sample size of 1,400included village, ulban poor, and analyses urban middle-classdistricts Shah, S. and Sunsample including416 villages Eastmoid, J.N. Crmsstabulations Causes of low educational attamment and dropouts (1977) includingthe different facts affecting females Sr I anka Rice, C and Wilber, Urban and rural household intermewsin three purposively Crosstabulations J.E. (1979) selected anas Includes impact of cdild care needs and other social facto.s on female education 204 SouthAsia References

Ahmad,P., and others.Women for Women:Women and Education. Dacca: Bangladesh Books International Limited,1978. Ahmed, LS., and M.Y. Hasan,A CaseStudy on Enrolment andAttendance of PrimarySchool-Aged Children by Socioeconomic Groupingfor the UPE/IDA Schools. Dhaka: The Foundation for Research on EducationalPlanning and Development,Dhaka University, 1984. Anderson,M.B. "On Girls' access to PrimaryEducation in Pakistan."The Bridges Foumm, January 3-8, 1988. Ashby,JA "Equityand DiscriminationAmong Children: Schooling Decisions in RuralNepal." Comparative Education Review, 29: 68-79, 1985. Asian Development BanIk Education Division, InfrastructureDepartmen4 Nepal: Education Sector Survey. Manila, 1986. Bach, R., S. Gadallah,HA.S. Khattab,and J. Gulick,"Mother's Influence on Daughters'Orientations TowardEducation: An EgyptianCase Study."Comparative Education Review, 29 (3), 1985. Becker,G.S., and H.G. Lewis,"Interaction Between Quantity and Qualityof Children."In Economicsof the Family,TheodoHP LaserJet Series IIHPLASEII.PRSrement of LivingStandards." LSMS Working Papers, no. 14. Washington,D.C.: The WorldBank, 1985. CaldwelLJ.C.; P. Caldwell,and P.H.Reddy, "Educational Transition in Rural SouthIndia." Population and DevelopmentReview, 11 (1985):29-51. Chakrabarti,AK. "The Causesof Women'sUnemployment in India."Economic Affairs, 22: 177-83,1977. Chamie, M. Nationa4 Institutional and Household Factors Affecting Young Girls' School Attendance in DevelopingCountres. The InternationalCenter for Research on Women and U.S. Agencyfor InternationalDevelopment--Office of HumanResources, Washington, 1983. Chaudhury,R.H. "MarriedWomen in Non-AgriculturalOccupations in a MotropolitanUrban Area of Bangladesh--SomeIssues and Problems." The BangladeshDevelopment Studies, 5: 153-200,1977. Clason,C. "Womenand Development:Three xperimental Projects."Literacy Discussion, vol. 6 no 4 (Winter):77-95 (1975-76). Desai,N. "EstablishingGender Justice through Education: A Case of Women'sEducation in India, 1951- 1987."Mimeographed. Research Centre for Women'sStudies, 1987. de Tray, Dennis N. "ChildQuality and the Demandfor Children."In Economicsof the Family,edited by TheodoreSchultz, pp. 91-116.Chicago, Illinois: University of ChicagoPress, 1974. Federal Bureau of Statistics,Statistics Division. Labour Force Survey, 1986-87. Karachi, Government of Pakistan. FREPD (Foundationfor Researchon EducationalPlanning and Development).The Study on the Situation of Childrenin Bangladesh.Dacca: FREPD, 1981. Governmentof Pakistan,1981 Census Report of Pakistan.Islamabad: Population Census Organization, 1984.

Gunawardena, C. "Women in Higher Education in Sri Lanka." HigherEducation Review, 19:8-23, 1987. South Asia 205

Husain, T., B.C. Sanyal, M.H. Abbasi, and S.R. Khan, HigherEducation and Employment Opportunitiesin Pakistan. International Institute for Educational Planning. Paris: UNESCO, 1987. Irfan, M. "Poverty,Class Structure and Household Demographic Behavior in Rural Pakistan."(revision of Poverty and household demographic behavior--Insightsfrom PLM survey). Pakistan Institute of Development Economics (PIDE), research report. Islamabad: PIDE, July 1985. Islam, S. Women's Education in Bangladesh:Needs and Issues. Dacca: Dacca University, Foundation for Research on Educational Planning and Development, 1982. Jahan, R., and H. Papanek, (eds.). Women and Development:Perspectives from Southeast Asia. Dacca: Bangladesh Institute of Law and International Affairs, 1979.

Jamison, D.T., and M.E. Lockheed, "Participationin Schooling: Determinants and Learning Outcomes in Nepal." Economic Developmentand CulturalChange, 279-306,1987. Junge, B., and S.M. Sharestha, "Another Barrier Broken: Teaching Village Girls to Read in Nepal." The Reading Teacher, (May): 846-52, 1984. Kalia, N.N. "Imagesof Men and Women in Indian Textbooks."Comparative Education Review, (June): 209- 23, 1980. Khan, S.R., R. Siddiqui,and F. Hussain, "Analysisof School Level Drop-Out Rates and Output in Pakistan." Pakistan Institute of Development Economics, research reports no. 149. Islamabad: PIDE, 1986. Khatun, S. "Women's Education in a Rural Community in Bangladesh."In Women and Development, R. Jahan and H. Papanek (eds.), 1979. King, E., J. Peterson, S. Adioetomo, L. Domingo, and S. Syed, Changein the Status of Women across Generationsin Asia. Santa Monica, California: The RAND Corporation, 1986. Lavador, S. "Women in Development: Issues and Strategies in Education." Mimeographed. New York: United Nations, International Labor Organisation, 1986. Marker, M.K, and S. Gah, "Women, Education and Development in Pakistan." Processed. 1985. Miyan, MA. Report on Job Market Survey for Women in Dacca. Dacca: Danida Mission, Royal Danish Embassy, 1979. Naik, C. "An Action Research Project on Universal Primary Education--the Plan and Process."In Women's Education in the Third World. Kelly and Elliot (eds.), 1982. Nayar, U. "Women in Educational Administrationin the Third World."In The ProfessionalPrepration and Developmentof Educational Administratorsin CommonwealthDeveloping Areas. Barbados: August 26-30, 1985. Papanek, H. "Class and Gender in Education-Employment Linkages." ComparativeEducation Review, (August): 317-46. Statistics Division,1985. Qadir, SA. School EnrolmentDrives and Enrolment of childrenin Class . Dhaka: 1986 BangladeshInstitute of Development Studies, Human Resource Division, 1986. Qasem, K.A. (ed.). Attitude of Parents TowardsSchooling of Children.Dhaka: Foundation for Research on Educational Planning aid Development, 1983. 206 South Asia

Raj, M.K "Women,Work and Science in India." In Women's Education in the Third World.G.P. Kellyand C.M. Elliot (eds.), 1982. Rice, C., and J.E. Wilber, ChildcareNeeds of Low Income Women in Rural and Urban SniLanka. League of Women Voters, Overseas Education Fund, in coliaboration with the Sri Lanka Federation of UniversityWomen, 1979.

Rosenzweig, i f.R. "Household and Non-household Activities of Youths: Issues of Modelling, Data and Estimation Strategies. Working paper no. 90, Population and Labour Policies Programme, World Employment Programme Research. New York: United Nations, International Labour Organization, 1980.

Rosenzweig, M.R., and R. Evenson, "Fertility,Schooling and the Economic Contribution of Children in Rural India: An Econometric Analysis."Econometica, 45: 1065-79,1977. Sambamoorthi, U. "Labor Market DiscriminationAgainst Women in India." Workingpaper no. 58, Women in Development. East Lansing: Michigan State University, 1984. Sarkar, B.N. "Enrolment and Primary Education Force in Rural India." Margin (April): 72-89, 1986. Sattar, E. Primary Education, Mass Literacy,Family Planning and Women's Programmein Ten Intensive ShawnirvarThanas of BangladeshVol. . Dacca: BangladeshAssociation for CommunityEducation, 1981. The------. Drop-out Problem in Primay Education. Bangkok: Regional Office for Education in Asia and the Pacific, UNESCO, 1984. Seetharamu, A.S., and M.D. Ushadevi, Education in Rural Areas. Bangalore: Institute for Social and Economic Change, 1985. Shah, N.M. PakistaniWomen: A Socioeconomicand DemographicProfile. Honolulu: East-West Population Institute, 1986.

Shah, S.U., and J.N. Eastmond, PrimaryEducation in Pakistan.Islamabad: Bureau of Educational Planning and Management, 1977. Shrestha, P. "Participation of Women in Higher Education ii. Nepal." The Economic Joumal of Nepa4 9: 30-34, 1986. Singhal, S. "The Development of Educated Women in India: Reflections of a Social Psychologist." ComparativeEducation, 20: 355-70, 1984. Smock,A.C. Women'sEducation in DevelopingCountries: Opportunities and Outcomes. New York: Praeger, 1981. Tara, S.N. Education in Rural Environments.New Delhi: Ashish PublishingHouse, 1981. Tilak, J.B.G. "Inequalityby Sex in Human Capital Formation, Labour Market Discrimination,and Returns to Education."Margin, 12: 57-80, 1980. UNESCO. Women, Educaton and Equality. Paris: The UNESCO Press, 1975. "Equality of Educational Opportunities for Girls and Women." Report of a meeting of a consultative panel for Asia and Oceania, October 1-8, 1979. Bangkok, 1980. -----. Statistcal Yearbook Parik: The UNESCO Press, 1987 and 1988 SouthAsia 207

UNICEF. Equal Aess of women to Education Propwmmes in Nepal. Center for EducationalResearch, Innovation and Development, 1978. Weiss,Anita M. 'Womenand FactoryWork in Punjab,Pakistan." Women and Workin the Third WoriS NagatM. El-Sanabary(ed.), pp. 207-14.Berkeley. Center for the Study,Education and Advancement of Women, Universityof Californiaat Berkeley,1983. East Asia 209 Chapter 7. East Asia _ 'AX4KOO~O~ Jandhyala B.G. Thlak

Developing East Asia, comprised of about a dozen small and large countries (Japan excluded) and a few tiny Pacific islands,enjoys the fastest economic growth rate and highest education level of any Third World region. The two achievements are linked. Their economic growth is affected by, and in turn affects, the rapid expansion of education and the large work force pool with postprimary schooling. The region has achievednear-universal primary school enrollment,and two-thirdsof the populationis literate, from the 65.5 percent of adults who cui read in China to 88 percent in Thailand. The countries have a relatively low education gender gap and women comprise a larger share of the labor force than in any other Third Wurld region. Economic development of the countries differs significantlydespite the overall high level, with per capita GNP ranging from $29(!in China to $7,410in Singapore. The region includes China, Hong Kong, Indonesia, the Republic of Korea (hereafter referred to as Korea), the Democratic Republic of Korea, the People's Democratic Republic of Laos, Malaysia, the Philippines,Singapore, Thailand, Viet Nam, Taiwan (Province of China), the People's Republic of Kampuchea (Cambodia), and Mongolia. Together these countries account for more than 30 percent of the world's total population and 11 percent of its area. This chapter focuses on oniy a few of these countries, partly because data on some of them, such as the Democratic Republic of Korea and Kampuchea, are unavailable. It pays particular attention to the newly industrializingeconomies (NIEs): the 'Four Little Tigers"of the East--Hong Kong, Korea, Singapore, and Taiwan--whichare the most developed areas in the region after Japan; two second-tier industrializing countries, Malaysia and Thailand; the developingcountries of the Philippines and Indonesia; and the poor countries of China and Viet Nam. These nations vary in geographical size from 1,000sq. km (as in Hong Kong and Singapore) to 9,561 sq. km (China); and in population from Singapore's 2.6 million to China's one billion plus in 1986. Although the countries are at various stages of development, they share some common characteristics. Historically,some were once very rich, industriallyadvanced, and materially prosperous. "The fame of their wealth earned for this region the appellation of th_.'gorgeous East' and inspired the quest that led to the discovery of the New World and created the preconditions for the Industrial Revolution in Europe" (Huq 1965, p. 5). Except for China and Thailand, the countries experienced colonial rule of different lengths by Japan, the United Kingdom,France, Spain, the Netherlands, or the United States, and became independent during the 1Q40s and 1950s,with Singapore attaining independence from the United Kingdom in 1965. Apart from their colonial legacy,the countries share a heritage of common culture and civilizationwhose traditions still exist. These include indigenouscultures and religionswhich sometimes bear similarities, as well as Buddhism, Confucianism, Islam, and Christianity, and to a lesser extent Taoism and Hinduism. Religion is important in most of the countries, and their societies are tradition-bound within a modern, technologically developing world. At the same time, each economy has its own "unique historical and cultural background, and its own social, political and economic institutions to inspire and guide its development goals" (Huq 1965, p. 5).

Economic and Educational Setting EownomicBackground In recent years, the overcrowded, resource-poor East Asian countries have succeeded so weil on the economic front that other less developed nations often look to them for lessons (Tang and Worley 1988). Many countries of the region, excludingthe People's Democratic Republic of Laos and China, are now well developed and fall into the World Bank classificationof middle-income economies (World Bank 1989). Table 7.1 Beaic Indiators for East Asia

Pomdation UrbanPop. GNPICaDita Growthin Income Share Total Growth as % of Area (000 USS Growth Employnent Poveo Gini of Bottom County (mil.) 1987 1980-87(%o) Total Pop. Sq. Kin) 1987 1965-87(Oo)1960-84 (%o) Rural Urban Coefficent 40% Pop.

China 68.5 12 38 9,561 290 52 na na na na 18.4

Viet Nam 65.0 2.6 21 330 na na na na na na na

Indonesia 171.4 2.1 27 1,919 450 45 2.6 44 26 0.492 14.4

Phiippines 58.4 2.5 41 300 590 1.7 3.7 41 32 0.452 14.1 Thailand 53.6 2.0 21 514 850 3.9 3.7 34 15 0.473 15.2 Malaysia 16.5 2.7 40 330 1,810 4.1 3.4 38 13 0.493 10.2

Korea, Rep. 42.1 1.4 69 98 2,690 6A 1.1 11 18 0357 16.9 Taiwan/ 18.9 2.7 66 14 3,048 6.6 3.6 na 5 0317 143

Singapore 2.6 1.1 100 1 7,940 7.2 3.2 na na 0.474 na

Hong Kong 5.6 1.6 93 1 8,/ 6.2 35 na na 0.481 A6.2

na = not available

Note: AU growth rates in this and other tables are per annum unless otherwise stated.

a" The Taiw;n figures refer to the most recent period for which data are available. I/ GNP data refer to GDP.

Soure: World Bank (1989); poverty-Tilak (1990); Gini coefficient-Rao (1988, p. 28); the data on Taiwan in this and other tables are from various sources such as Kurian (1982),Tang and Worley (1988),and national statisticalbases. East Asia 211

Table 7.1 shows the impressivegrowth of the East Asian economiesduring the last two decades. Compared to an average increase of 2.6 percent per annum in per capita GNP during 1965-87for the middle-income group as a whole,1 most countries in this region progressed at a rate of more than 4 percent per annum, the highest being 7.6 percent in Singapore, followedby 6.7 percent in Korea, and 6.6 percent in Taiwan. Even China, a low-income country, made rapid progress--5.1percent per annum. The countries also differ in their degree of urbanization. Singapore and Hong Kong are almost totally urbanized, but more than 75 percent of the population in Thailand, Laos, Viet Nam, and China live in rural ar,;as. In Malaysia, Indonesia, the Philippines,and Thailand, some 30 to 40 percent of the rural population live below the poverty line.2 Income inequality is generally high in most of these countries: the Gini coefficientof household income varies between 0.317 for Taiwan in 1985 and 0.493 for Malaysia.

Stahts of Women The socioeconomicdevelopment of these countries has improved the status of women markedly. On the whole, gender inequalities are not as marked as in some neighboringAsian countries. However,traditions, such as the highly patriarchal societies in China and Taiwan, the matrilineal society in Thailand, and the fading Islamic purdh culture in Indonesia, result in significantdifferences in the status of women across East Asia.3 Table 7.2 presents selected indicators on the status of women in the region in 1965 and about 1985. Life expectancy at birth, an important indicator of social development, was much higher among women than men, and rising in every economyin the region. In Hong Kong, Viet Nam, Korea, Taiwan, and Singapore, the life expectancyof women at birth exceeded that of men by five years or more. With total employment growing more than 3 percent per annum (except in Indonesia and Korea during 1960-84),increasingly more women participated in the work force. In 1985,women constituted at the two extremes 31 percent of the total labor force in Indonesia and 47 percent in Viet Nam. These levels were higher than the corresponding overall average figures for lower and upper middle-income countries. However,women still constituted a smaller share of the labor force than men, as in the rest of the world. Labor force participation rates present a slightly different picture. In developing East Asia the rate of participation of women in the labor force vHP LaserJet Series IIHPLASEII.PRS lower among women than men. Discrimination againstwomen was quite marked, however,in a few countries for which data are available, a point discussed later.

1 Group averages, unless otherwise noted, are from World Bank (1989).

2 These are the only countries for which data on poverty are available.

3 See several papers in Ward (1963) for descriptions of the changing roles of women in South and Southeast Asia. 212 East Asia Table 7.2 Selected Indicators on WomenWsDevelopment

Women % Rate of Lffe EMectancv as % of ParticivatLonin Labor Force Female Mate Labor Forc 1985 1985/ Country 1965 1987 1965 1987 1965 1985" Female Male Female Male

China 59 71 55 68 41 43 44 60 52 64 Lao, PDR - 50 - 47 47 45 52 58 45 53 Viet Nam - 68 - 64 47 47 46 54 44 52 Indonesia 45 65 43 58 29 31 2Ž 55 24 53 Philippines 57 65 54 62 34 32 25 49 24 49 Thailand 58 66 53 6A 48 46 47 52 48 56 Malaysia 59 72 56 68 29 35 20 47 28 51 Korea, R,ep. 58 73 55 66 29 34 20 48 28 53 Taiwant' - 73 - 67 na na na na 42 84 Singapore 68 76 63 70 22 33 15 49 33 63 Hong Kong 71 79 64 73 32 34 25 51 38 64

na = not available.

D/ Most recent available data--about 1985. k/ Latest data available are for 1975. Source: World Bank (1988b and 1989).

Education LeveLE The speedy economic and technological growth of the East Asian countries stems partly from the rapid expansion of their educational systems and the resulting fast-growingreservoir of workers with postprimary education. In turn, rapid economic and technologicalgrowth has had a profound influence on education sector growth. In certain respects, education levels attained in countries like the Philippines are comparable to that of the developed world. Here again, however, factors behind educational expansion vary from country to country in the region. With per capita GNP of $590, the Philippines' advanced education level and achievements in female schooling are due not so much to economic growth as tc the country's long-standing direct education policies. Even before World War II, and moreso afterwards, the Philippines implemented a massive expansion of schools. Moreover, in this most Western of Asian countries, there are no major cultural constraints to educating girls. Despite impressivegains in education, East Asia also shows substantial variation in literacy. As shown in table 7.3, literacy levels for women are lower than for men. The problem is particularly severe in China, where nearly half the women are illiterate. Throughout the region, illiteracy is higher in rural areas. East Asia 213 Table 7.3 Adult literaqy Rates (in pffent)

Country Year Total Females Males

China 1982 65.5 51.1 79.2 Hong Kong 1971 77.3 64.1 90.1 Indonesia 1980 67.3 57.7 77.5 Korea, Rep. 1970 87.6 81.0 94.4 Lao, PDR 1985 83.9 75.8 92.0 Malaysia 1980 69.6 59.7 79.6 Philippines 1980 83.3 82.8 83.9 Singapore 1980 82.9 74.0 91.6 Taiwan 1975 82.0 75.6 92.2 Thailand 1930 88.0 84.0 923 Viet Nam 1979 84.0 78.3 90.5

Note: Percentages are for ages 15 and older, except those for Laos, which are for 15- to 45-year olds. Source: Based on UNESCO (1987).

Similarly, the level of development of education varies considerably across the region.4 The educational composition of the labor force (table 7.4) makes these differences clear. Larger percentages of the labor forces in Taiwan and Philippines have higher levels of education, while the smallest shares of workers in China and Indonesia have higher education. In mean years of schooling of the labor force, a summary statistic of educational development,Hong Kong and Taiwan are at the top and Indonesia and China are at the bottom among the East Asian countries (see Oshima 1988).

4 See Postlethwaite and ihomas (1980)and Thomas and Postlethwaite (1983) for a description of primary and secondary education in the region. 214 East Asia Table 7.4 Educaonal Compositionof the Labor Force

HighestSchool Level (% of Labor Force) Pims v. Secondarv Mean Years Countoy Year None Incom. Comp. Incom. Comp. Higher of School

China 1982 28.2 13.1 21.3 25.8 10.7 0.9 4.5 Hong Kong 1981 7.7 17.6 19.1 213 26.2 8.1 8.8 Indonesia 1978 31.6 23.1 35.7 5.3 3.8 0.5 3.9 1980 26.2 18.9 33.4 11.4 8.9 1.2 4.9 Korea, Rep. 1969 44.9 9.1 30.2 7.3 6.1 2.4 3.9 1980 14.7 1.1 33.2 18.5 23.4 9.1 8.0 Malaysia 1967 27.0 1.7 55.7 9.2 4.6 1.8 5.0 1980 17.9 17.1 23.4 22.9 16.1 2.6 6.5 Philippines 1980 7.8 21.3 27.4 15.1 12.7 15.7 7.0 Singapore 1974 40.4 4.9 21.9 16.0 8.3 8.5 5.3 1980 21.9 3.0 46.4 18.4 6.3 4.0 6.0 Taiwan 1980 9.4 4.5 30.2 18.9 24.3 12.7 8.6 1983 9.0 5.1 32.7 17.7 24.0 11.5 8.4 Thaitand 1960 37.4 55.6 1.' 3.5 2.0 0.4 3.3 1980 10.1 64.2 7.0 11.2 4.1 3.4 4.6

Source: Psacharopoulos and Arriagada (1986).

All the economies in the region have progressed rapidly in education. Based on gross enrollment ratios (that is, unadjusted for overagedand underaged children),primary education is largelyuniversal (table 7.5). The Philippines, Korea, Hong Kong, and Singapore achieved universalprimary education more than two decades ago, whereas many neighboring countries in South Asia are still struggling to achieve this target. The repetition rates at the primary and secondarylevels in developingEast Asia (except Indonesia) are very low for both girls and boys; in some countries, such as Korea, they are virtualy zero (table 7.6). These low rates of repetition, however, are the result of automatic promotion and nonretention policies in primary education (Levy1971; UNESCO-ROEAP 1984). In general, the repetition rates are lower when enrollment ratios are high; and when gender equity is attained in enrollment, gender differences in repetition rates are also less. East Asia 215 Table 7.5 Gross Enrollment Raios in Education (pwcentof rekvwatage 5rip)

SchoolLemel &Mms 5eod Nipser Country 1965 1987 1965 1987 1965 1987

China 89 132 24 43 0 2 Hong Kong 103 106 29 74 5 1iW Indonesia 72 118 12 48 1 7 Korea,Republic 101 101 25 88 6 36 Lao, PDR 40 110 2 27 0 2 Malaysia 90 102 28 57 2 7 Mongolia 98 1029/ 66 429/ 8 221/ Philippines 113 109 41 68 19 29 Singapore 105 112 45 68 10 12 Ta .ivan 96' look/ 33 9/ 86w5 19 Thailand 78 95 14 28 2 16 Viet Nam 1079/ 102 39£/ 42V 2k/ 2V

a/ 1960. / 1982. i/ 1975. g, 1984. / 1986. V 1985. Il 1980. Source: WorldBank (1988band 1989);UNESCO (1990).

Comparedwith other developingregions, East Asia has made major progressin secondaryand higher education (table 7.5). In 1987in Korea,88 percentof the relevantage group childrenwere enroled in secondaryschools. The rate for Mongoliawas 92 percent,Singapore 68 percent,Hong Kong74 percent, and the Philippines68 percent. Thailand,China, and Indonesialagged substantially behind, while Laos and Thailandhad the lowestrates of the regionat 27and 28percent, respectively. Higher education enrollment reached29 pe.cencin the Philippinesand 36 percentin Korea,comparable to the 39 percentaverage for the rich industrialmarket economies.Once again,however, significant intercountry differences exist. In China,only 12 percentof the relevantage group woreenrolled in the highereducation system in 1987;in Malaysiaand Indonesia,only 7 percent. 216 East Asia Table 7.6 Rates of Repetition in Education (in percent)

School Level Primai Seconda=v County Year Females All Females AU

China 1988 na 7 na na Hong Kong 1984 2 2 7 8 Indonesia 1987 na 10 na 1 Korea 1987 na na 0 0 Lao, PDR 1987 na 27 na 10 Mongolia 1982 na 1 na 1 Philippines 1987 2 2 na na Singapore 1984 1 1 4 5 Thailand 1980 7 8 2 3 Viet Nam 1978 na na na 5

na = not available. Note: Data are not available for Malaysia Source: UNESCO (1987 and 1990).

High levels of enrollment do not necessarilycorrespond to high levels of public education expenditures in East Asia (table 7.7). Even though they spend relativelylittle on education, some couatries, such as the Philippines and Hong Kong, have made remarkable progress in education in general and in female education in particular.5 Conversely, some countries, such as Malaysia,that have devoted relatively high proportions of GNP aLd budget to education, have not made comparable progress. Malaysiainvested as much as 7.8 percent of its GNP in education in 1986 but had a secondary school enrollment rate of only 57 percent. In 1987,Korea and Singapore invested 3.9 and 3.8 percent, respectively,and had relativelyhigh enrollment rates. The Philippines expanded its education system remarkably given a meager investment of just 2.0 percent of GNP. The implicationis that the level of public spending on education only partially explains the development of education.

s In the Philippines, a substantial number of students are enrolled in private universities and colleges, a fact that may mean lower proportions of the government budget are needed for education. East Asia 217 Table 7.7 PublicInvestment in Educationin East Asia

Country Year % of GNP % of Budget

China 1987 2.4 8.19/ Hong Kong 1984 2.8 18.7 Indonesia 1981 2.0 9.3 Korea, Rep. 1987 3.9 26.6 Lao, PDR 1986 na 6.6 Malaysia 1986 7.8 16.9 Philippines 1987 2.0 7.OWJ Singapore 1987 3.8 11.5 Taiwan 1979 4.1 5.6@/ Thailand 1987 3.6 17.9

na = not available. Note: Data are not available for Mongolia and Viet Nam.

i/ 1983. k/ 1984. 9/ 1982. Source: UNESCO (1987 and 1990).

Gender Differences in Education

Information on South Asian females' school participation and enrollment absolutely and relative to males, and on gender segregation in education, can be gleaned from available data on literacy rates, education levels, enrollment rates, enrollment size and growth, share of women in total enrollment by level and type of education, and grade repetition. More than half of East Asian women are literate, an important indicator of their significantprogress. In Thailand and the Philippines in 1980 and in Korea in 1970, four out of every five women were literate as shown earlier in table 7.3. Female literacy was lowest in China at 50 percent (1982), stil high compared with the 20-30 percent rates in neighboringSouth Asian countries. The distribution of males and females by level of education is an indicator of male-female equality in education. Table 7.8 shows this distribution for adult males and females separately in the early 1980s (except Malaysia,for which the reference year is 1970). In every country of the region for which data were available, the proportion of women with no formal schoolingwas larger than the corresponding proportion of men. At the extreme, in China in 1982, nearly two-thirds of the women did not have any formal schooling, compared to one-fourth of men. Even in the NIEs, except Singapore, the proportion of females with no schooling is twice or more that of males. Nevertheless progress in female education is evident. In the NIEs and the Philippines, women are making strong advances in secondary and post-secondary education. 218 East Asia Table 7.8 Adult (Ages 25+) Population by Educational Level

Total _Pecentage Distributional County Year Populatfon (mil.) io School PnmawySecondary Post-Secondawy

China 1982 All 466.9 44.5 32.7 21.6 1.0 Females 227.2 623 23.6 13.6 0.5 Males 239.7 27.6 41.3 29.2 1.5 Hong Kong 1981 All 2.6 22.5 39.8 32.5 7.1 Females 1.2 35.9 34.8 24.3 5.0 Males 1.4 10.3 44.4 40.0 9.0 Indonesia 1980 All 58.4 41.1 48.4 9.6 0.8 Fenmales 29.8 53.9 39.5 6.3 0.4 Males 28.7 27.8 57.6 13.0 1.2 Korea, Rep. 1980 All 16.5 19.7 34.5 36.9 °.9 Females 8.5 26.9 39.4 29.6 4.0 Males 8.0 12.0 29.3 44.7 143 Malaysiak' 1970 All 10.3 43.4 42.6 13.9 / Females 5.1 51.0 38.2 10.8 si Males 5.2 35.9 46.9 17.0 Philippines 1940 All 17.9 11.7 54.1 18.9 15.2 Females 9.0 13.3 55.1 16.6 15.1 Males 8.9 10.3 53.1 21.2 153 Singapore 1980 All 1.2 43.7 38.8 14.6 3.4 Females 0.6 543 31.2 12.5 2.0 Males 0.6 33.5 46.1 16.6 4.8 Taiwand/ 1980 All 12.2 15.7 35.5 38.8 10.0 Females 5.8 22.8 36.5 34.0 6.8 Males 6.4 9.3 34.7 43.2 12.7 Thailand 1980 Ali 17.5 20.5 69.7 6.8 2.9 Females 9.0 26.3 67.0 4.3 2.4 Males 8.5 14.4 72.6 9.5 3.4

P/ Primary includes incomplete and complete; secondary includes 'entered." k/ Includes population of all ages. Included in secondary. Includes population aged 15+.

Source: Taiwan.-Kaneko (1987); others--UNESCO (1987).

The flow variables that explain these current trends do no- show such large gender differences, indicating remarkable improvement in female education in recent decades. The current enrollment ratios given in table 7.9 suggest that by the 1980s most countries had significantlyreduced or eliminated the imbalances in male-female enrollment. Particularly at the primary level,the imbalances were mostly eliminated by the East Asia 219 1970s,and primary educationis nowtotally or nearlyuniversal for boysand girlsin almostall countriesof the region. Becausethe populationsof these countriesare dividedalmost evenly, 50-50, between men and women equalityin educationwould exist if womenconstituted half the total enroDments.As shownin table 7.10, in the mid-1980sin general female studentsdid comprisenearly half the total enrollmentsin primary schools,a proportionthat also held for secondaryschools in muchof the region. Koreamade significant progressin equalizingenrolments at the secondarylevel with the proportionfor femalesincreasing by 6 percentagepoints in about a decade. These figurescompare very favorably with the average1986 figures of 51 percentfor lowermiddle-income countries and 59percent for upper middle-incomecountries. The growthrates in table 7.11show the pace at whichfemale and maleenrolments have been risingin the differentcountries. The negativerates for primaryeducation in Hong Kong,Korea, and Singaporereflect declinesin their populationgrowth rates, and thusin the youngestschool-age cohorts. In the countrieswith positivegrowth rates, femaleprimary school enroDment rose faster than or equal to that of males. In general the enrollmentgrowth observed at the postprimarylevels was much faster for femalesthan males. An exceptionis the Philippines,where maleenrollment at the tcsing levelhad been lagg behind that of femalesprior to 1980,as was the case in Chinaand Viet Nam at the tertiarylevel. 7 Despitesignificant progress, gender differences in educationpersist, particularly in the distributionof males and females by field of study at the secondaryand higher levels. For example,girls tend to take traditionallyfemale-dominated courses such as nutrition,nursing, and teachertraining. In the Philippines, more than 90 percent of the studentsin each of these fieldswere women. Males tend to dominatein engineering,law, agriculture, and technology(Neher 1982, p. 160;and BorceDe1985).

6 Even though the net enroDlmentratios (gross enrollmentratios adjusted for over- and underaged children)in the early and mid-1980swere generallylower than the grossenrolment ratios and wereless than 100 percent,education at the first levelwas still more or lessuniverWs in developingEas Asia. 7 In 1980,female enrollmentin secondaryschools in the Philippineswas 1,559,000and maleenrollment was 733,000.In 1985,female and maleenrollments were about equaL 220 East Asia Table 7.9 Gross Enrollment Rates in Education by Gender jercentoIf rekvat popuaion)

Flimagv Secondana Tertiarv County Year Female Male Female Male Female Male

China 1975 114 130 38 54 0.4 0.7 1987 124 132 37 50 1.2 2.3 Hong Kong 1975 117 122 47 51 5.3 14.7 1987 105 106 76 71 9.3W/ 16.80 Indonesia 1975 78 94 15 25 na na 1987 115 121 42 53 4.22/ 8.9p'

Korea, Rep. 1975 107 107 48 64 5.9 14.6 1987 94 94 86 91 22.2 48.6 Lao, PDR 1980 86 102 14 21 0.3 0.6 1987 98 101 22 31 1.2 2.0 Malaysia 1975 89 92 39 48 1.6 4.0 1987 102 100 57 56 6.2 7.1 Mongolia 1975 104 111 84 77 8.6 8.2 1986 103 104 96 88 260 17.4 Philippines 1980 113 113 69 61 28.5 26.8 1987 110 105 68 67 28.5k/ 26.8W Singapore 1975 107 113 52 51 7.3 10.7 1987 111 118 69 67 10.2 1330/ Thailand 1975 80 87 23 28 2.7 4.0 1980 97 99 28 30 na na Viet Nam 1975 108 106 41 38 1.6 2.6 1985 99 105 40 43 I.obW 3.8WJ na = not available. Notes. (1) Gross enrollmentrates are definedas the ratio (expressed.in percent) of all studentsenrolled in say, the primarylevel to the populationin the relevantage group. These rates can exceed 100percent due to a largw.number of underagedor overagedstudents. (2) Data for Taiwanare from another sourceand are givenin Appendixtable 7.1 W 1984. / 1980. G/ 1983. Source: UNESCO(1987 and 1990). East Asia 221 Table 7.10 Share of Females in Total Enrollment in Education (percentages)

Country Year Primary Secondary Tertiary China 1975 45 39 33 1987 47 41 32 Hong Kong 1975 48 47 25a/ 1987 48 49 35 Indonesia 1975 45 38 3P/ 1987 48 44 32 Korea, Rep 1975 44 41 25 1987 47 47 30 Laos, PDR 1980 45 332/ 29 1987 49 41 37 Malaysia 1975 48 48a/ 38-/ 1987 46 50 48 Mongolia 1975 48 52 50 1986 49P/ 53a/ 60 Philippines 1980 51 68 53 / 1986 52 50 54 Singapore 1975 47 49 39e/ 1987 49 50bj 4291 Thailand 1975 48 44 40 1980 49 na na Viet Nam 1975 49 49S/ 39 1985 52 na 24 /

na=not available. 1984. El 1980. s/ 1976. A/ 1982. e 1983. Source: Based on UNESCO (1987 and 1990).

Table 7.11 Averge Annual Growth Rates in School Enrollments, 1970-1985

Primary Secondar . Tetia County Females Males Females Males Females Males China -1.6 1.6 4.4 4.7 26.5 28.7 Hong Kong -2.1 -2.1 6.3 3.9 9.0 7.2 Indonesia 5.1 4.4 11.0 8.4 12.4 9.7 Korea, Rep -1.0 -1.1 8.0 53 15.8 13.5 Lao PDR 6.6 4.3 17.9 12.8 23.6 17.0 Malaysia 2.0 2.1 6.4 4.1 17.0 11.2 Mongolia 2.1 2.1 5.7 5.1 14.1 10.1 Philippines 1.7 1.6 4.6 4.0 7.5 7.9 Singapore -1.8 -1.8 2.1 1.6 10.3 6.4 Thailand 2.0 1.8 7.0 5.4 18.8 16.8 Viet Nam 1.0 0.8 5.1 3.9 -1.9 -1.4 na = not available. Source: computed from UNESCO 1989. 222 East Asia

Data on the female percentage of total enrollment by type of education at the secondary and higher levels are presented in tables 7.12 through 7.14. In most of these countries, women comprised a majority of secondary students in teacher training courses, but only in Mongolia did vocational courses attract more women than men.

Table 7.12 Enrollment in SecondaryEducaion, by Curriculum

Percentae Enroled in Cuniculumal Female Share of Males Females Total Enrollments County Year Genl Teach'g Vocat'l Gen'l Teach'g Vocat7l Total Genl Teach'g Vocat'7

China 1975 98.6 0.0 0.0 983 1.1 1.5 39.2 39.3 0.0 OD 1985 92A 1.1 6.5 92.4 0.9 65 40.2 40.2 39.0 40.6 Hong Kong 1975 965 0.0 3.7 91.9 0.0 7.7 46.6 47.8 0.0 293 1984 95.2 0.0 4.8 905 0.0 9A 50.0 51.2 0.0 338 Indonesia 1975 79.7 4A 15.9 73.5 2.0 245 38.3 40.2 583 28.7 1984 89.7 5.0 7.0 86.8 2A 9.6 41A 42.2 59.6 342 Korea, Rep. 1975 8.8 0.0 325 1986 82.8 0.0 17.2 83.7 0.0 16.3 47.0 46.8 0.0 483 Laos, PDR 1976 81.1 17.6 1.5 88.7 8.8 2.0 32.6 30.7 49.1 26.1 1980 86.8 115 1.6 87.6 9.9 2.6 38.6 38.4 42.1 28D Malaysia 1980 99.0 0.0 1.0 97.7 0.0 2.2 47.6 47.9 0.0 29A 1985 99.1 0.0 0.9 97.7 0.0 2.3 49.1 495 0.0 28.6 Mongolia 1975 91.6 1A 7.0 94A 0.3 4.7 51.7 51.0 81.3 615 1982 90.0 1A 8.6 93.8 0.3 5.6 52.6 51.5 82.6 63D Singapore 1975 100.0 0.0 0.6 92.3 0.0 7.1 49.1 51.1 0.0 7D 1984 96.9 0.0 2.9 93.5 0.0 6.7 49.7 50.6 0.0 298 Thailand 1975 792 4A 16.4 80.7 3.5 15.7 43.8 43A 49.6 449 Viet Nam 1976 97A 1.1 1.5 96.9 0.5 2.7 48.8 49.0 68.5 343

a/ Detail may not add to 100 percent because of rounding Source: UNESCO (1987). East Asia 223 Table 7.13 Females as a Percent of Total Enrollment in Higher Education,by Subject (International Stawdad Cimficauon)

Hong Kong Indonesia Korea,Rep. Malaysia Philyppines Singapor VietNan 1984 1984 1986 1985 1985 1983 1980

Total 34.5 323 30.1 44.5 54.4 41.9 23.6 Education, Science, & Teacher Training 63.4 38.0 58.8 63.2 73.7 72.4 28.7 Hum., Religion & Theology 613 36.6 44.6 45.9 33.8 75.1 30.7 Fme & Applied Arts 46.4 34.2 65.6 42.1 52.5 0.0 14.5 Law 52.8 273 10.2 50.0 27.2 533 4.8 Social & Beh. Sciences 493 33.3 20.1 45.7 74.3 A/ 20.7 Comm. & Business Admin. 53.0 40.1 15.1 47.6 75.0 56.5 25.1 Mass Communication & Documentation 61.0 37.9 54.2 40.5 51.2 0.0 11.0 Home Economics 0.0 80.8 95.0 50.6 96.5 0.0 0.0 Service Trades 30.9 44.0 48.6 75.4 71.2 0.0 0.0 Nat;ral Sciences 21.6 38.9 25.7 41.0 66.5 64.1 20.3 Math. & Comp. Sciences 25.8 12.3 28.2 44.7 56.9 62.3 25.4 Med. Sc. & Health-Related 32.2 32.1 48.1 47.0 77.2 33.0 16.0

Engineering 2.5 16.4 3.8 14.0 14.5 16.7 16.1 Arch. & Town Planning 11.7 19.6 5.8 20.3 17.5 47.8 26.9 Trade, Craft & Ind'L Prog. 36.1 25.3 32.5 29.1 35.3 0.0 20.6

Transport & Communications 0.0 22.8 14.4 223 63 0.0 13.9 Ag, Fozestry & FLshery 0.0 24.9 12.2 20.2 43.5 0.0 21.9 Other 51.7 10.4 32.4 353 353 20.7 0.0 a/ Included in commerce and business administration. Notes: Where nero percentages are noted, the size of enrollment reported was nil.

Source: Based on UNESCO (1987). TA&e 714 Dztluion of Students mIger Education by Subject (Intnuatow Sandard Ciw#caton)

Hong Kong Indonesia Korea,Rep. Malaysia Philippines Singapore Vet Nam Subject 1984 1984 1986 1985 1985 1983 1980 Fem. Males Fem. Males Fen. Mates Fem. Males Fem. Males Fem. Males Fem. Males

Total 100 100 100 100 100 100 100 100 100 100 100 100 100 100

Educ, Sci, & Teacher Training 143 4.4 28.4 22.1 21.7 65 36.8 17.1 20.1 8.6 25.4 7.0 44.9 34.5 Hum, Religon & Theology 10.8 3.6 3.2 2.7 20.6 11.0 1.8 1.7 0.3 0.7 14.4 3.5 4.8 3.4 Fme & Applied Arts 2.7 1.6 0.4 0.4 10.0 23 0.7 0.8 03 0.3 0.0 0.0 0.5 1.0

Law 0.7 03 9.5 12.0 1.1 4.1 1.2 0.9 0.6 1.9 2.2 1.4 0.2 1.1 Socal & Beh. Sciences 7.2 3.9 23.0 21.9 3A 5.8 10.8 103 1.2 0.5 0.0 0.0 0.6 0.7 Comm. & Bus. Admin. 38.3 17.9 173 123 8.7 21.1 16.4 145 44.5 17.7 18.8 6.8 13.0 12.0

Mass Comm. & Documentation 23 0.8 0.6 05 1.6 0.6 0.4 0.5 0.2 03 0.0 0.0 03 0.6 Home Economics 0.0 0.0 0.5 0.1 9.6 0.2 0.2 0.2 0.9 0.0 0.0 0.0 0.0 0.0 Service Trades 2.1 2.5 03 0.2 1.1 0.5 5.5 1.4 1.5 0.7 0.0 0.0 0.0 0.0

Natural Sciences 23 4.5 2.6 2.0 3.1 3.8 6.2 7.2 1.0 0.6 103 4.2 1.9 23 Math. & Comp. Sciences 2.4 3.7 03 0.9 3.1 3.4 2.4 2.4 2.9 2.6 1.5 0.6 1.8 1.7 Med. Sc. & Health-Related 3.0 33 25 25 8.7 4.0 33 3.0 9.0 3.2 2.5 3.7 6.8 1.0

Engineering 2.1 42.9 5.7 13.8 2.1 23.0 4.0 19.5 4.3 30.0 17.4 62.6 6.7 10.8 Arch. & Town Plng. 0.7 2.8 0.8 15 0.7 5.0 1.7 5.4 0.4 2.2 5.6 4.4 2.4 2.0 Trade, Craft & Ind. Prog. 53 5.0 03 0.4 03 03 0.6 1.2 9.6 21.0 0.0 0.0 6.4 7.6 Transport & Communs. 0.0 0.0 0.0 0.1 0.2 0.5 03 0.9 03 4.6 0.0 05 0.9 1.8 Ag., Forestry & Fish. 0.0 0.0 43 6.2 2.1 6.3 1.1 35 23 3.6 0.0 0.0 8.8 9.7 Others 5.7 2.8 0.1 0.6 1.9 1.7 65 9.6 0.4 0.7 1.9 5.2 0.0 0.0

Note: Detail may not add to 100 percent because of rounding.

Soue Based on UNESCO (1987). East Asia 225

In higher education, the subject concentration of women in a few countries was even clearer; for example, in engineering,women ranged from a meager 2.5 percent of the total students in Hong Kong to 16.7percent in Singapore. Men also dominated in medicine and health-related sciences,except in the Philippines,and in the natural and mathematical sciences, except in the Philippines and Singapore.

Women's educational choices are both influenced by and influence women's occupational prospects (see Sutherland 1988,p. 483). Where educational options and occupational fields are in close correspondence, the gender differences in choice of subjects may be quite sharp. For example,in the Philippines,95 percent of teachers in primary and secondary schools were women, and nearly three-fourths of the tertiary-level students taking education and teacher training courses were women. This gender bias was also evident in postschool technical training. Lazo (1984) reported that women usually trained for the garments trade, embroidery, clerical work, food and nutritic,u, and food preservation/food processing. Men mostly took supervisorycourses, instructionaltechniques, and trained for suchfields as electricity,automotive mechanics, and radio-televisionmechanics.

Women's concentration on just a few areas of study and training is due to a variety of factors. Social pressures lead girls to enter fields that are occupationallyless remunerative. Women choose subjects that complement both future market and nonmarket activitiessince, unlike men, they foresee some periods of complete 'specialization' in the household as wives and mothers (Ram 1982,p. 211). The lack of access to professionaljobs also influenceswomen's choices for education. In Korea, Indonesia, and Thailand, barely 3 percent of the female labor force was in professional occupations;in Taiwan and Malaysia,the figure was only aboat 5 percent (Kurian 1982). Temporal comparisonsmay indicate, as Don (1984,pp. 119-20)argued in the case of Malaysian secondary education, that utilization of educational facilities by women tended to be "complementary"to that by men over time--female participation decreased when male utilization of economically advantageous education was higher. Such patterns are not unusual.8 More important, however, these patterns seem to be "resistantto changes even under conditions where equalization between men and women has been a prime goal in educational reforms" (Harnqvist 1987, p. 358), as in the case of developing East Asia.

Some people argue that occupational segregationby gender may not be totally disadvantageousto women. In some cases, it may contribute to an increase in female enrollment. For example, growing demand for teachers may induce more women to enroll in teacher training courses. The Islamic tradition of gender segregation may to some extent, as McGrath (1976,p. 37) argued, help promote the employmentof women as teachers because women must teach girls, or as physiciansbecause only women must treat women. On the whole, however, this pattern may not be to women's advantage in the long run.

Factors Affecting Education Reach and Attainment

Detailed gender-specificanalyses of the determinants of schooling in developing countries, includingEast Asia, have been very few. In most cases, gender is introduced as one of several variables in equations explaining enrollment or educational attainment. This method does not shed light on why gender differences in education exist, however. Different researchers have found various factors to be determinants of school participation and attainment. Depending on the purpose at hand and the background of the researcher, studies have emphasized demographic factors, sociocultural factors, or labor market factors. Although an exhaustive list of these factors, classified into a few groupings in order of importance, is beyond the scope of this chapter, a few categories can be identified as follows: (a) economic factors, including labor market characteristics; (b) community and sociocultural factors; (c) school factors; (d) home environment; and (e) demographic factors. These groups :re not necessarily mutually exclusive--somesocial factors have an economic interface, some home factors may have social dimensions, and so forth.

8 See Moore (1987) for documentation of such patterns for a large number of countries. See also UNESCO (1980a and 1985). 226 East Asia

Income Levels The generalbelief is that the higherthe levelof economicdevelopment, as measured,say, by per capita GNP, the hiher the levelof femaleparticipation in schooling.In a cross-nationalstudy, Schultz (1987) confirmedthis belief, findinglarger income elasticitiesof educationfor females than for males. He concluded, ...femaleenrollments tend to increasemore rapidly with income per adult than do male enrollment rates...therise in incomeand the declinein relativeprice of schoolingthat appear to occurat the onset of modern economicgrowth contribute reinforcing gains to the educationalattainment of womenthat exceedthose achievedby men (pp. 450-51).

The levelof higher educationfor femalesdoes not necessarilycorrespond closely to the country'slevel of economicdevelopment, however. As Kelly(1984) concluded, ...women'saccess to educationand the disparitybetween male and femaleenrollments at all levels of schoolingcannot be predictedby the levelor pace of a nation's economicdevelopment. Sex inequalityin educationis not a problemthat willgo awayonce a nation has mu-easedits Gross NationalProduct and built an industrialinfrastructure (p. 83).9

Thesetwo viewsare not necessarilyinconsistent. Per capitaGNP maynot be the best measureof the level of economicdevelopment. For example,higher rates of economicgrowth might require more womento participatein the economy,a conditionthat leads to greater women'sparticipation in education. Several studies have found that poverty,as measuredby income and assets, was critical in decisions regarding female education. Cochrane and Jamison (1982) observedthat parental landholding,an importantindicator of wealthin rural society,was the most importantpredictor of female (but not male) educationalattainment. In rural Malaysia,the low socioeconomicstatus of Chinese women was an importantbarrier to their accessto education,but not so for Chinesemen (see Safilios-Rothschild1980). De Tray (1988) found that familyincome had three times as much influenceon the probabilityof enrollmentof girlsas on that of boysaged 12-18in PeninsularMalaysia. In Indonesia,Chernichovsky and Meesook(1985) concluded that, althoughin each economicstratum schoolenrollments were lower for femalesthan males,female enrollment rates werehigher in the economicallybetter-off and urban segments of the population. Direct responsesfrom Indonesianparents as to why their childrenare not in schoolalso indicate the importanceof incomeas a determinantof schoolinvestments (table 7.15). Economichardship ("no funds") was the singlemost frequentreason given for not sendingchildren to school,but its importanceappears to be equal for daughtersand sons.

9 See also Bowmanand Anderson(1982), Smock (1981), and Deble (1980). East Asia 227 Table 7.15 Distributionof Reasons for Non-Participationin Schoolsin Indonesia cwent)

Ages 7-12 AMes13-15 Ages 16-18 Reason for Not Java Outer Islands Java ltrIandsJava OuterIslands Attending School Male Female Male Female Male Female Male Female Male Female Male Female Had Sufficient Schooling 1.9 8A 0.0 53 5.0 49 5.6 7.8 3.9 10.0 10.7 9.0 No Funds 48.1 49.1 49.3 43.9 51.5 47.7 S6.9 55.8 55.9 49.8 S34 5S.9 Too Difficult 12.5 204 164 17.9 13.1 12A 14.5 11.6 11.8 11.7 11.3 8.3 SchoolToo Far Away 9.6 2.4 0.0 0.0 7.0 8A 4.5 1.1 5.8 5.5 2.8 1.5 Other 27.9 19.8 34.2 33.0 23.4 26.6 18.6 23.7 22.7 23.0 21.8 25.3

Note: Detail may not add to 100 percent because oi rounding. Source: Chernichovskyand Meesook (1985) p. 16.

Low family income may force girls to abandon school because the opportunity cost of sending daughters to school may be higher than for sons. Girls may earn wages in the labor market or help in household work so that their mothers can take up wage labor. By doing so, they also enable their brothers to go to school. Safilos-Rothschild (1980) found that girls were given more responsibility for time-consuming household chores, care of young children, farming (in some regions), and all kinds of nonmarket activities. Girls generally were assigned greater responsibilitiesat an earlier age in productivelabor in and outside the home than were boys. Although educating boys in low-incomecountries can also have a high opportunity cost, that cost was borne less reluctantlybecause parents had higher expectationsfor male education1 . The high opportunity cost of education for daughters and their smahlerrange of economic opportunities as adults are factors in many of these countries--particularlypoorer countries such as Indonesia--in the relatively low participationby women at secondary and higher education levels. Exceptions are the Philippines,Singapore and Hong Kong where a larger share of girls are enrolled at higher education levels.

Labor Market Factors

Conditions in the labor market also influence the family's educational decisions significantly. Gender- based role divisions can determine the education level and choice of subjects in secondary and tertiary schools for sons and daughters. At least four factors in the labor market are important in this context: wage structure, employment/unemploymentlevels, participation in the labor force, and rates of return on education. Very few studies explainingschooling choices estimate the feedback effect of these labor market conditions, however.

10 See de Tray (1979) for evidenceon children's economic contnbutions in Malaysia. 228 East Asia Wage Disamnation 1' According to human capital theory, wage structure and, more importantly, expected changes in this wage structure, play a dominant role in decisionmakingregarding investment in education. In general, wage structures everywhere in the world tend to favor men.12 The East Asian economies are no exception. Table 7.16 presents data on nonagriculturalwages, not specified by educational level, for three East Asian countries. In 1985-86,Korean women earned less than half what males earned; in Singapore, about two- thirds; and in Hong Kong, three-fourths. In a detailed examination in the Philippines, Arcelo and Sanyal (1987) found that males had higher wages in all fields except in those related to "female"professions, such as food, nutrition, dietetics, and, surprisingly,also in law and some liberal arts pursuits. In some industries in Malaysia, the wage differences ran 1:3 in favor of males (Sundaram and Leng 1985, p. 44).13

These male-female wage ratios only reinforce parents' traditionalbias against educating their daughters, and discourage the daughters themselves, in some countries. According to a study on Malaysia (Wang 1982), boys expectea their salaries to be higher than did girls. The less a grouV perceives further education as contributing to earnings, the lower is the tendency to continue in school.1 Differentials in "full"earnings, which include benefits,would be a more appropriate measure of the earnings gap between men and women. Although many modern government sectors pay equal wages to men and women for the same job, men may enjoy greater fringe benefits, such as cost of living and housing allowances, as in Malaysia (Wang 1982).

11 Although differences in gross earnings unadjusted for differences in productivity-relatedfactors cannot be totally attributed to discrimination,the term discrimination is used here in the broad sense that these differences reflect "cumulativediscrimination" (Madden 1975). See also the discussion in Tilak (1987,pp. 148 and 160-63).

12 That gender itself is an important determinant of earnings is well-documented. See, for example,Blaug (1974) on Thailand.

13 See also Singh (1987) and Yue (1987) on Malaysia, and Tonguthai (1987) on Thailand and the Philippines.

14 Rarely does wage discriminationforce women to get more education so that they may earn wages equal to men's; see, for the case of Chile, Schiefelbein and Farrell (1982). For India, Rosenzweig and Evenson (1977) found that male wages and enrollment levels were negativelyrelated, while female wages and female enrollment levels were positivelyrelated. East Asia 229 Table 7.16 Wage Discriminton in Hong Kong, Korea, and Singapore, Non-agriculturalSector

Earnings (local currencies) Coefficientof Country Year Female Male DiscriminationWl

Hong Kong (wages/day) 1982 66.70 87.90 0.3178 1986 98.10 129.80 0.3231 Korea, Rep. (earnings/month) 1977 47.40 107.60 1-2700 1986 208.90 426.90 1.0436 Singapore (earnings/hour) 1980 1.68 2.67 0.5893 1985 2.78 4.04 0.4532

aJThe coefficientof discriminationis, as defined by Becker (1957), (Wm/Wf) - 1, where Wm and Wf refer to male and female wages, respectively. A lower coefficientindicates greater equality.

Source: Based on United Nations (1985).

Emplkyment/Unemplye Current educational choices depend on perceptions of unemployment conditions. Demand for education may be positivelyor negativelyrelated to these perceptions. Higher rates of unemployment may induce more enrollment in higher education because time becomes available,albeit not by choice, for academic pursuits. A worsening employment situation also mav have serious adverse effects on enrollments, more so for girls than for boys. When unemployment becomes widespread, a popular opinion is that "women should withdraw into their homes and leave the available posts for men"; men are seen as the principal breadwinners,women as earners of supplementaryincome or "pin money' (Sutherland 1988,p. 486). Thus, women have little incentive to pursue a career. In China, the CommunistParty's theoretical journal clearly suggests that women's productive role should be only "secondaryand supportive" (Hooper 1984, p. 327).

In addition, better C oyment conditions could reduce enrollments, as the opportunity costs of education rise, or they could stk uilate enrollments as the future returns to educations (in the form of better jobs) increase. Women's demand for schoolingmay 'e directly tieo to their perceptions about their employment probabilities and future remuneration from work. Accordingto Wang (1982),Malaysian women perceived unfavorable employment conditions and believed they would tend to get less well-payingjobs than would males for a given level of education. The range of jobs for women was in fact restricted, partly by discriminatoryhiring practices,parLy by the inadequacyof informationregarding employmentopportunities, partly by society's role expectations for women, and partly by the "male monopoly of the labor market" (Madden 1975). At the same time, men perceived that a wider range of better payingjobs would be open to them. The perceptions of parents and husbands regarding employment probabilities for women also significantlyinfluence decisionmakingregarding women's education. The link between female participation in the labor force and education is well documented. Frequent estimates have been made of education's effect on the probability of being in the labor force (see Shields 1987). Individualsurvey data from the Philippines(Harman 1970;Encarnacion 1974) and Indonesia (Nagib 1986;Corner 1987)clearly showed a direct positiverelationship between females' averageyears of schooling and their participation in the labor force. However, the ways in which the rate of labor force participation 230 East Asia influenceseducation has rarely been examined.Backer (1964 p. 51) arguedthat "womenspend less time in the laborforce thap men, and therefore,have less incentive to investin marketskills." In other words, if the rate of participationin the laborforce by womenis low,women tend to havelower educational levels. Employmentconditions are quite good in those East Asian economiesthat have experiencedrapid technologicaladvances. As shownearlier (table 7.1), the annualrate of growthin employmenthas been above 3 percent in most of these countriesduring the 1960-84period, Indonesiaand Korea being the exceptions.Although education-specific data on unemploymentrates bygender are not available,in general womenexperienced lower rates of unemploymentthan men in HongKong, Korea, Singapore, and Thailand (table 7.17). The situationwas differentin the Philippines;although men had lowerunemployment rates, femaleunemployment was increasing at a slowerpace than that of men (see alsoEncarnacion and others 1976).16 Table 7.17 Unemploymentin East Asia

As % of those in labor force oounvy Year Females Males

China 1983 1.1 0.7 1986 1.0 0.6 Hong Kong 1977 4.2 4.4 1986 2.5 3.0 Korea,Rep. 1977 2.4 4. 1986 2.1 49 Philippines 1977 8.6 2.6 1985 8.2 4.8 Singapore 1977 1.6 2.3 1986 5.5 7.0 Thailand 1977 0.8 1.2 1980 0.7 1.0 Taiwan 1968 11.9 5.3

A/ Officialestimates. Source: Taiwan--Visaria(1980, p. 90); others--UnitedNations (1985).

These macro trends were not closelyrelated to female enrollmentin school. The very low female unemploymentrate in Chinaand Thailandwas positivelyassociated with relatively low femaleenrollment ratios in secondaryand higher education.This lends somesupport to the opportunitycost argument. In contrast, the high female unemploymentrate in the Philippineswas associatedwith a very high female

15 Mincer(1962, p. 68) also observed: "In vie- of the expectedsmaller rate of participationin the labor market, educationof womenis more stronglyfocused on the 'consumption'sphere, and returnsare non- pecuniarythan for males." See also Mincerand Polacheck(1974). 16 Arcelo and Sanyal(1987, p. 141) further showedthat the rate of unemploymentwas higher among graduatesof privatehigher educationthan amonggraduates of publichigher education. In either case, womenhad higherunemployment rates than men. East Asia 231 enrollmentratio in higher education. Becausehigher educationis largelyprivate in the Philippines,the argumentrelating to time availabilitymay not be a credibleexplanation. A possibleexplanation is that, althoughoverall female unemployment rates were high,the rate for the moie educatrdwomen may have been the lowest.

Unequal Rates of Retum on Eation

Accordingto human capital theory, the rate of return, which reflectsboth the costs and benefits of education,provides "signals of direction"for additionaldemand for education. Shultz, in ChapterII of this volume,presents past estimatesof the rates of return to educationby gender.1 Becauseof differences in estimaton methods,the rates of return acrosscountries are not directlycomparable. Nevertheless,the estimatesare stilluseful for comparingrates of return betweenmen and womenwithin each country. In Korea,Taiwan, Thailand, and Indonesia,the rates of return on women'seducation were higherthan those for men. In Malaysia,secondary education yielded - higherrate of return for womenthan for men, but investmentin highereducation paid relativelymore for men than for women,although the differenceswere not large. Thiswas also the case for all levelsof educationin Singapore. Thus,on the whole,women's educationin East Asianeconomies yielded returns that werehigher than or about equal to those on men's education.18 In all these countries,however, female enrollment ratios in secondaryand highereducation were lower than male enrollmentratios, exceptin secondaryeducation in Malaysia. Thus, the rates of return alone failto explainfemale enrollment ratios in East Asia.19 Nevertheless,the rates of return clearly suggestthat most East Asian economiesunderinvest in women'seducation.

School Factors

Schoolsupply and other related factorsare a major determinantof educationalattainment. Schoolsthat are inadequatein resourcesand thosethat fail to providerelevant curricula lead to low enrollment,poor attendance,and underachievementby students. Heyneman and Loxley(1983) in a studyof 29countries and Fuller (1986) in a reviewof about 60 empiricalstudies found that school characteristicswere more importantto levelsof educationalachievement than were socioeconomriccharacteristics of the home.

Availabibltyof Schwos

The availabilityof schoolfacilities is importantin determininglevels of participationby both boysand girls. It may be more importantfor girls,however; parents may not mind sendingtheir sons to a neighboring villagefor schoolbut mayhesitate to send their daughters.As Kelly(1984, p. 86) observed,'the greatest singleindicator of whetheror not a girl will attend schoolmay well be whetherschooling is made both availableand accessible."A studyon Indonesiaconcluded that "wheneducational facilities are availableand accessible(in proximityand cost),daughters are likelyto be givenequal opportunitywith sonse (Scott 1985, p. 13). Provisionof additionalschools after independencein Malaysiaboosted female enrollment

17 Schultzalso discussesthe problemsinvolved, inter ali in gender-basedcomparisons of conventionally estimatedrates of return on education.

18 If allowanceis made for nonmarketwork, as Woodhall(1973) rightly argued, the rates of return on women'seducation are likelyto be much higherthan thoseon men's. In addition,Tilak (1987)has shown that greater labor force participationby womenwould substantially increase the rates of return on their education.

19 This result is not altogethersurprising. The rates of return on educationrarely explaindifferences in enrollmentin educationacross different countries. See, for example,Tilak (1982) for a widercross-country study. 232 East Asia significantly(Hirschman 1979). As mentionedearlier, the Philippines'achievements in femaleeducation are attributablein part to the major expansionof the schoolsystem over the past 50 years. Closelyrelated to the provisionof schoolsis the distanceto schools. "Amongthe most problematicfactors for girlsare the costs of travelto school(in time of hazards),a matter of both logisticsand culturalnorms" (Bowmanand Anderson1982, p. 25). Physicalaccessibility is importantto improvingenrollments. If a schoolis withinthe communityand withineasy walking distance, enrollment is likelyto be high for both boys and girls,and it may be criticalto the enrollmentof girls. In a studyof 400 householdsin twenty- two rural villagesin Thailand,Cochrane and Jamison(1982) found that the distanceto schoolnegatively and significantlyaffected participation in schoolsby both boysand girls. Surprisingly,however, the impact on femaleparticipation was less. In the Philippines,however, the effectwas higherfor females,as e .petcted. Whena schoolwas providedwithin the villageor at a "shortdistance," female enrollment was estimatedto increaseby 3 percent,while male enrollment was estimated to increaseonly by 1 percent(King and Lillard 1987,p. 172). King and Lillardalso reported positiveeffects in Malaysiafrom havinga school in the communitywhere childrenlive. Anotherkey factor is the provisionof completeprimary schools. Schoolswith onlylower primary levels maynot attract as manystudents as woulda completeprimary school. Thisfactor maybe more important in decisionsregarding schoohng for girls than for boysbecause completing the primarylevel would entail travelto a schoolat somedistance. As Johnstone(1976, p. 234)observed, "...parents perceiving the non- availabilityof completeprimary level schooling, are more proneto withdrawtheir children(or not to send them at all) than are parents who live in regionsor areas where completeprimary schools exist." In Malaysia,de Tray (1988)found that childrenresiding in communitieswithout a secondaryschool exhibited lowerrates of primaryattendance than did childrenin communitieswith secondary schools, and this effect was larger for girls than for boys. Distanceto the secondaryschool did not appear to have a significantly differenteffect on girlsand boys,but the lack of convenienttransporiation had a more adverseeffect on femalethan male enrollment.

7)le of School The type of school and qualityof schools also influenceparents' decisionsregarding their daughters' education. For instance,in a traditionalor religioussocial environment, religious schools may be more effectivein enrollingfemale children. The Islamicschools in Malaysiaassure parents that traditionalsocial valueswill be taughtso that their daughterswill make better wives and better Muslims(Don 1984).Female studentscomposed a large majorityin theseschools, indicating that religiousschools were more important for girls than for boys. A relevantconsideration is whethera schoolis single-genderor coeducational.20 Single-sexschools may be seen as both an economicand a socialissue. They mayL.fect differentials in levelof participationand achievementof boys and girls. Again,in traditionalmid&"--class families, many parents prefer all-girls schoolsfor their daughters. Free mixingof boysand girlsin coeducationalschools is regardedas morally unhealthyfor the girls, while single-genderschools provide a "protectiveenvironment" for daughters. Empiricalresearch supports this view on single-genderschools: they are found to be more prestigiousfor socialas wellas furthereducational reasons (Smock 1981). Jimenez and Lockheed(1988) found that even after controllingfor suchfactors as socioeconomichome background and schoolresources, girls in Thailand achievedmore in all-girlschools than in coeducationalschools, while boys did better in coeducational schools. This was due to a varietyof factors,including a better classroomclimate in the all-girlsschools (see also Jimenezand others 1988). In East Asia, however,such schools are few in number. In Malaysia, for example,coeducational schools greatly outnumbered single-gender schools (Don 1984,p. 117).

20 See Lee and Lockheed(1989) for a short reviewof the literatureon this issue. East Asia 233 Femae Teaches

Middle-class traditional families,particularly Muslim ones, prefer that their daughters be taught by female teachers rather than male ones. A shortage of female teachers may be an important educational constraint in these societies. The number of female teachers, however, is a function of the enrollment of women in teacher training courses in secondary and higher education. China and Laos are at the bottom of the ranking when it comes to the share of female teachers in primary school teachers (table 7.18). In 1987 only 40 percent of the primary school teachers in China, and just over one-third in Laos were women. At the secondary level, these two countries are joined by Indonesia and Korea. In the Philippines, where the proportion of female teachers corresponds closelyto female enrollment, male teachers are in the minority. The evidence suggests that patterns of female enrollment in East Asia are associated with the proportion of female teachers, but no microlevelstudies have analyzedwhether having a female teacher encourages girls' schooling. Because this factor is often cited as a potential force in further raising girls' education levels, a more rigorous examination of this hypothesisis in order. Table 7.18 Percentage Share of Female in the Teaching F.)rce

Counhy Year Pnimary Secondary Higher

China 1987 41 29 28 Hong Kong 1987 74 49 240 Indonesia 1987 49 33 18!/ Korea,Rep. 1987 46 31 15•' Laos PDR 1987 35 35 22b/ Malaysia 1987 53 49 224/ Mongolia 1986 na na 39 Philippines 1984 95 95 53F' Singapore 1987 71 52" 21a/ Thailand 1980 49 na 56WJ Viet Nam 1985 70 57 22-/

na = not available.

!/ 1984. h/ 1985. / 1980. /" 1983 e/ 1976 Source: UNESCO (1990).

Schoing Expenditwes Several direct schooling expenditures, including school fees, books, clothing (uniforms), hostels, and transport, have been found to be higher for girls than for boys in many countries. For example, for Indonesia, Chernichovskyand Meesook (1985,p. 12) estimated that fees and other schoolingexpenditures for females were Rupees (Rp.) 190 at the primary level and Rp. 1,090 at junior high school compared to Rp. 132 and Rp. 862, respectively,for males. At the senior high school level, however, expenditures per child were much less for females than for males: Rp. 841 for females versus Rp. 1,425 for males. The reason for this switch at the higher levels may be that boys are more likely to be enrolled in courses that entail more capital expenditures, such as courses that require laboratories. 234 East Asia In the householdcalculus of the total costsand benefitsof education,comparatively higher direct costs at the lowerschool levels can be veryimportant and maybe majorobstacles to the enrollmentof femalesfrom poor families(see also Deble 1980). The expansionof publiceducation in manycountries have paid rich dividendsin the form of significantincreases in girls'enrollment in primaryeducation; reforms related to this expansionindude the abolitionof the fee in lowerprimary grades in Indonesiasince 1976and free primaryeducation in other countries,such as Malaysia.The large-scaletextbook program in Indonesiaand the textbookloan scheme,supplementary feeding programs, and scholarshipsfor the needyat the secondary levelin Malaysiaare believedto havebeen significantfactors in boostingwomen's participation in school. The Philippinesguaranteed equal rights to educationfor both men and women as early as 1901,and Thailandenacted policies for compulsoryeducation as early as 1921. As expected(see UNESCO1980b, p. 29), policiesrelated to achievinguniversal primary education in the East Asian countriesreduced the significantdisparity between males and femalesin education. In Indonesia,special government provisions to directmore resourcesto primaryeducation and a massivesr-hool building program increased enrollment in primary education remarkably. China provides an excellent example of effective policies of decentralizationand local financingof primaryeducation. Also in China,allowing girls to take younger siblingsto schoolinstead of havingto care for themat home(Gansu province) has enabledmany to attend school(World Bank 1988a). Governmentexpenditures have also been directedtowards female students. For example,a larger number of womenstudents receive scholarships than do men in Korea'scoleges and universities:in 1980,58percent of the scholarshipswent to women,compared to 38 percent in 1975(Korean Councilfor University Education1988, p. 22). These favorablefinancial aid policiesmay be partly responsiblefor Korea's significantimprovement in enroDlmentof womenin highereducation.

Sociocaltual Factors In mostdeveloping societies, families place a lowervalue on the educationof their daughtersthan of their sons. In East Asia,this preferenceseems less strong and is declining.McGrath (1976, p. 37) argued,'Most obstaclesto fuDlequality in educationexist only in peoples'minds, in the insubstantial,diaphanous forms of prejudice,traditional beliefs, and culturalstereotypes." In an interestingstudy on Taiwan,Greenhalgh (1985)presented evidence of parentalgender discrimination in providingeducation. The highlypatriarchal Chinesetraditional family system was reflectedin parentaldiscrimination against daughters. Girlshad to sacrificetheir educationand work to financetheir brothers' education. The data indicatedthat as the numberof daughtersincreased, the numberof yearsof schoolingof their brothersincreased. 21 Parentsstill viewedfemale educationmerely as a stop-gapbetween childhood and marriage,and a "womangoes to coDegepartly to increasethe profitabilityof marryinga more desirableman" (Becker 1964,p. 103). In Malaysiaalso, education for girls seems to have been consideredprimarily a way to increase their desirabilityas marriagepartners (Weekes-Vagliani 1980, p. 69). Parentstried to equalizeeducational levels among their sons but not their daughters. In another study,King and others (1986)estimated that if daughterswere treated likesons, female schooling levels would rise by 150percent in rural Java(Indonesia) and by 30 percent in the Philippines,where discriminationwas not that high. Data on post-school investmentin the form of trainingand apprenticeshipshow more markeddiscrimination against daughters in Taiwanesefamilies. One reason for parental bias againsteducating daughters is that by the time the girls are employed,they may be marriedand the parentswould not be able to reap these benefits. Education,by its very nature, has a longgestation period before the economicbenefits are reaped in the form of increasedearnings. In Thailand,however, women took care of their parents in old age, so parents wanted to provideas much educationto their daughtersas to their sons (Safilios-Rothschild1980, p. 338).

21 See Rosenzweig(1975), who argues,in contrast,that a predominanceof malesin a familywould be to the advantageof females. East Asia 235 In the Philippines,'religion and linked culturalpatterns providethe major originsof division!by gender (Smock1981, pp. 17-18). Religionis generallybelieved to be a dominantinfluence on the participationof womenin education?2 Religiousideas and teachingsexert a powerfulinfluence in shapingsociety's values in general and its approachesto femaleeducation in particular. Althoughmost religions in their originalforms do not discouragefemale education, in practiceat least some have done so. The region'smajor religionshave, in turn, factionssuch as the Hinayana,Mahayana and Theravadain Buddhism.Some religions, such as TheravadaBuddhism, are regardedas fosteringtechnical change (Niehoff1964), while some, such as Islam and Confucianism,are believedto hinder progress. Muslimsdiscouraged female education in theory,given their view of the placeof women(Kelly 1984, p. 85). Amongthe economiesin East Asia, the two Muslimcountries--Indonesia and Malaysia--afterChina, have had the lowestrates of female literacyand the highestmale-female differences in literacy. In Islamic societies,as Dasgupta(1988) observed, ...even if literacyis regardedas a good thingfor men,the same maynot applyto women.Female literacymay be thought, at best to be unimportant. Some may even be persuadedthat it is undesirable,for literacycould have the effect of makingwomen km fit for the role they are traditionallyexpected to perform(p. 130).

SomeIslamic states have progressedin educatingfemales, and their advancesare comparableto thosein developedcountries. Malaysia,for example,has made significantprogress in highereducation of women. There, although the relativelyconservative Sunni Muslims were dominantin number in 1985,women constituted44 percent of the total enrollmentin higher education,a rate comparableto that in some Europeancountries. Religious education played a significantrole in the developmentof women'seducation. Islamicschools in Malaysiawere dominated by girlsbecause most parents preferred to send their daughters to religiousschools, which segregate boys and girls. Parents also felt that religiousschools were more harmoniousand that girlsin religiousschools were less likelyto be waywardand wouldmake better wives (see Don 1984). On the other hand, the Islamicschools in Indonesiahave not had a similareffect on women'senrollment, probably bec!Iuse they havebecome "more and more like secular schoolse(Thomas 1988). In general,the Muslimstates in East Asia are distinctlydifferent from the othersin Asia and in the rest of the world. Muslimwomen in East Asia, for example,do not practiceseclusion (Dldah) and by and large are not constrainedby very rigid gender segregation(see Whyte and Whyte 1978). Further, the practiceof religionvaries significantly within East Asia.

Ethnic Odigin Ethnicityis an important factor related to culture,values, and social position,and appears to be an importantdeterminant of educationalparticipation in manycountries. For example,in Malaysia,significant differencesin educationlevels are evidentbetween Malay and Chinesewomen. In Kuala Lumpur,23 percentof workingChinese women have received postsecondary education, while among Malay women the share has been three timeshigher. Also, the proportionof femaleswith secondary and higher education in the Malaywork force has been greaterthan that of the maleMalay work force (see Mazumdar1981, and Weekes-Vagliani1980). Controllingfor familysocioeconomic characteristics, de Tray (1988)found that Chinese and Indians in Malaysiatended to have lower enrollmentrates than Malays,but this inequity seemed to have been less for girls than for boys. King and Lillard (1987),also controllingfor other socioeconomiccharacteristics, found that the enrollmentrates of Malaywomen improved much more than those of Chinesefemales as a result of the educationalreforms begun in the early 1970s. The ratio of

22See U. King (1987)for an analysiscf the waysin whichmajor religions have generally given a lowerand sometimesextremely limited role to womenin education. 236 East Asia

Malay to Chinese enrollment rates at the secondary level increased from 0.6 for cohorts born before 1952 to 1.31 for cohorts born after 1964;at the tertiary level this ratio rose from 0.8 to 1.2.

Several studies have argued that marriase, which is almost universal for women in East Asia, discourages women from continuingtheir education. Systematicdifferences have been found between the educational levels of married and unmarried women, as well as those of unmarried men and women, compared with married men and women, of aViven age group. Single women have achievedhigher levels of education as compared to married women. This would imply that the lower the legal age of marriage, the lower is the level of enrollment of females, partic llarly in secondary and higher education. The scanty evidence is not very clear, however. As. Bowman and Anderson (1982) observed, ...there is no simple trade-off between marriage and schooling, for in most countries the proportions married plus those in school add up to much less than 100percent. Rather, it appears that there are common causes for both early marriage and low school attendance (p. 19).

In Hong Kong, women marry late, work before marriage, and help their parents financially. As a result, parents attach the same importance to their daughters' education as to their sons' (Salaff 1976). Of the six economies in the region for which data are available, the minimum legal age for marriage is lowest in Thailand (1'); it is 16 years in Indonesia and Korea, 18 n Malaysia and Taiwan (Kurian 1982),and 20 in China. The enrollment ratios for women in secondary and higher education have been much lower in Thailand than in Malaysia,but in Korea, the ratios have been higher than those of Malaysia. In China, which has the highest legal age for marriage, enrollment in secondaryand higher education has been among the lowest in the region. Thus, the relationship between the age of marriage and participation in secondary and higher education has not been consistent. The direction of causalityin the relationship between education and age at marriage is also unclear. Actual or desired level of educational attainment may postpone the timing of marriage. More educated women marry at later ages, so marriage does not substantially affect their further education and labor force participation (Montgomery and Sulak 1989). King and others (1986) examined the effect of the level of completed schooling on age at marriage and labor force participation in Indonesia and the Philippines. Among poor urban and rural families in Java, they found that education increased the chances that women would enter the labor market and that they would marry later. Even though education is associatedwith higher socioeconomicstatus and other direct benefits, East Asian men do not necessarily prefer to marry better educated women. A survey of university students in China found that only 28 percent of university male students preferred to marry university graduates (Hooper 1984, pp. 331-32). The others feared that universityeducated women would lose the "traditional feminine virtues of gentleness and devotion." Hooper concluded that in China "higher education might well be detrimental to a young woman's marriage prospects." Dowry practices also have implications for female education. The dowry is an important social factor with economic dimensions. Huge dowries force poor parents to postpone their daughters' marriage and absorb resources that might otherwise be available for educating them. To the extent that women's education is viewed as an asset, it could supplement, or even substitute for, a dowry. For many parents, higher education of their daughters becomes at least a partial substitute for the dowry they have to pay to bridegrooms. In several cases, women's earnings before marriage are also paid into the dowry.

23 See Safdlios-Rothschild(1979) for a comprehensivereview of the literature.

24 See Weekes-Vagliani(1980) for details on Malaysia. East Asia 237 Household Characteristics In the literature,a numberof factors togetherdescribe the socioeconomicstatus of households. Two of these factors--parentaleducation and occupation--areparticularly important to the educationof girls. Parental educationis generallyexpected to influencefemale participationin schoolingpositively. Well- educated parents perceivethe intrinsicand monetarybenefits of educationmore clearlythan do less educated parents. Parents who have experiencedthe education-earningsconnection tend to send their childrento school. Workingmothers particularly may be motivatedto send their daughtersto school. A .Wdyfor Indonesiafound that amonghouseholds with university- or academy-educatedheads, the gender differencein the children'senrollment rates is narrowerthan for the generalpopulation, and in some age groups(or at someeducational levels), females had higherenrollment rates than did males(Chernichovsky and Meesook1985). King and others (1986)found that mother'seducation was an importantdeterminant of children'sschooling in Indonesianand Philippinehouseholds. Among the Malaysin Malaysia,King and Lillard(1987) found that a mother'seducation had a strong,positive effect on daughters'schooling but not on sons', while a father's educationgenerally did not affect the schoolinglevel of children. In the Philippines,they founda positiverelationship between the educationalattainment of both parentsand their children'sschooling. Smith and Cheung(1982) found that gender differencein educationalparticipation variedsignificantly with the father'seducational level. In Taiwan,among the severaldeterminants of female educationalattainment that Hermalinand others (1982)examined, the father's educationturned out to be the most dominant. In their studyon Thailand,Cochrane and Jamison(1982) found that father's (not mother's)educational aspirations for daughtersis importantin determiningtheir daughters'schooling, but not so for sons. The occupationof parents has generallybeen found a significantfactor in their children'seducational attainment,particularly for girls. Hermalinand others(1982) found tha: after parentaleducation, the most importantvariable influencing female educational attainment in Taiwanwas parental occupation: the higher the occupationof theparents in the occupationalhierarchy, the higherthe probabilitythat their daughters wouldgo to school. This is similarto resultsobtained for Indonesiaby King and others (1986),who concludedthat girlswhose fathers had white-collarjobs tended to havesignificantly higher schooling than those whosefathers had blue-collarjobs. In both Indonesiaand the Philippines,children of farmershad muchlower schooling than others. Among the factors that might influencefemale educationalparticipation, family size has attracted wide attention amongresearchers.26 In larger families,each child receivesless individualattention and other resourcesfrom parents. Larger familiesmay not be able to affordto send all their childrento schooland prefer to send the boys. In large families,then, givena preferencefor boys' educationand higherschool costs for girls,girls are less likelyto be sent to school.In addition,in a large family,girls' householdwork would increaseconsiderably. Earlier research appears to indicate a persistentlynegative correlation betweenfamily size andchildren's educational performance: children from small families tended to perform better in schoolthan did those fromlarger families,partly because they had more financialresources and receivedgreater parental encouragement(King 1987,p. 375). Aggregate patterns, however,do not clearly support the general hunch that family size and female educationalparticipation are inverselyrelated. Of the countriesfor whichdata on householdsize are available,the Philippineshas the largestaverage family size, 5.9; yet its femaleenrollment ratio in higher educationis also the highestin the region. Indonesia,with an averagehousehold size of 4.8, amongthe lowestin the region,is associatedwith the secondlowest level of femaleenrollment in secondaryand higher

25 This occupationalhierarchy places white-collarworkers at the top, blue-collarskilled workers next, followedby unskilledlabor and farm workers. 26 Also possibleis that parentaleducation influences family size. Considerableresearch is availableon the effectsof educationon demographicfactors, such as fertilityrates and populationgrowth. See, for example, Hermalin(1974) on Taiwan,and Goldstein(1972) on Thailand. 238 East Asia education. Household size in Malaysiais also low-4.8--and its female enrollment rates are not particularly high in comparison with other countries. Thailand, however,which is at the bottom with respect to female enrollment ratios in secondary and higher education, has a mean household size of 5.7, the second highest.

The results of investigations at the family level are similarly mixed. Studies in Taiwan (for example, Hermalin and others 1982, p. 269) and in Indonesia and the Philippines (King and others, 1986)concluded that family size was not significantlyrelated to the educational levels of siblings, or of females in particular. In rural Thailand, however, the presence of young children in the household was found to reduce the participation of both boys and girls in school particularlyat the upper primary level (Cochrane and Jamison 1982).

Urban Residence That educational attainment is biased against rural children is well-known;rural-urban differences are well documentedin the literature. These differencesare particularlyglaring for women. For example,female literacyrates in China are 47 percentin rural areasand 74 percentin urban areas. The differencesare less in other countries, but urban rates are still noticeablyhigher: in Indonesia,52 percentof rural femalescan read versus76 percentof urban females,and in Malaysiathe rural literacyrate is 52 percent againstan urban rate of 74 percent. Schultz(1987) found in his cross-nation study that urbanization was not a significantvariable in explaining the differencesin enrollmentrates at the primarylevel for either boysor girls,but urbanizationbecame an importantfactor at the secondarylevel and was more importantfor femalethan for maleenrollment. Smith and Cheung(1982) found that in the Philippinesurban-educated parents made more favorabledecisions regardingtheir daughters'education than did those (especiallyfathers) with a rural background. Country-leveldata do not unanimouslysupport an urban-ruraldifference in propensityto educatefemales. HongKong and Singapore,which are almostentirely urban, have lowerfemale enrollment than do many other economieswith respect to both secondaryand higher education. The Philippines,with only one- third of its populationin urban areas, has muchhigher enrollment ratios than less rural countries. China and Indonesia,however, have both low levelsof urbanizationand low femaleenrollment ratios.

Summaryand Conclusions On the whole, educational opportunities for women in East Asia are good in comparison with other developing regions. Literacy and school enrollment of women are fairly high and in some respects comparableto levelsin developedcountries. The East Asian economiesdiffer significantly,however, in their social, economic,political, demographic, and educationalcharacteristics. China, Indonesia,and Malaysiaare relativelyless developedin terms of educationalattainment than the rest of the region,while economicallydeveloped Korea and the less well-developedPhilippines are way ahead.

The educational indicators for East Asian women are mixed. For example, female enrollment ratios producedone rank order for the countries,while the share of women'senrollment in total enrollment produced a slightlydifferent rank order, and rate of growth in female enrollment produced yet another. On the whole, however, the relative rank order of countries for female education using different indicators varied onlymarginally.

.. iatioas in social, economic,cultural, political and historicalconditions explain to some extent the differencesin levelsof femaleeducation among East Asian countries.Studies on pedagogicalfactors, such

27 See Jamison and Lockheed (1987) for similar results on Nepal. East Asia 239 as textbooks,relevance of curricula,and methods of teaching,usually did not separate their empirical analysesby gender. Genderwas introducedonly as a shift factor in the equationsestimated. This reAewof the East Asian experienceleads to the followingconclusions: * The educationalexperiences of most of the economiesin the regionsuggest that coloniallegacies in educationcan be overcomeeasily and educationdeveloped aidckly,and that genderequality in education is achievable. * The Philippines'experience clearly suggeststhat econom:- dev"lopmentis not a prerequisitefor expandingeducation for women.Carefully planned and executededucational and economicpolicies and programscan enhancefemale participation in schoolsquite remarkably. * Socialcustoms, such as early or late marriage,do not systematicaUyaffect the enrolment of womenin education. * Demographicfactors, such as householdsize, are not necessarilyrelated to femaleenrollment. * Althoughthe literature placesmuch of the blame for the low enrolment of girls on culturalfactors, includingreligion, this viewhas no strongbasis. The Malaysianand Indonesianexperiences suggest that religions,such as Islam,need not stand in the wayof ensuringequality of educationalopportunities for women. The educationlevels of womenin these two countriesare not muchlower than men's relative to educationin other countries. No religion--Christianity,Islam, Buddhism or Confucianism--hasbeen a major hurdle in narrowingthe gender gap in education. T-hesecondusions are in a sense reassuring. Although by tradition, women have been subject to discriminationat home,school, and in the labor market,nothing about these intangibleideas and beliefs is immutable.Rather, traditionshave been responsiveto changesinduced by economicdevelopment and governmentpolicy. Meaningfulpublic policies have paid rich dividendsin improvingwomen's educational levelsin this region. Policiesfor free and universalprimary education, improving school quality to increase promotionand survivalrates, and establishmentof single-genderschools are onlya few examplesof the positivesteps that emergefrom this reviewof the experienceof developingEast Asia. Few meaningful educationpolicies and programsfaiL Educationprograms that reduceparent's expenditureson schooling, such as publicprovision of textbooks,scholarships, and supplementaryfeeding programs, have had a very significanteffect on femaleenrolment. Whenprovided with the appropriateincentives, girls and women enter schoolsand completeeducation levels close to those achievedby men. 240 East Asia AppendixTable 7.1. Gross Enrollment Ratios in Taiwan (pent of relvaw populaion)

Year Age-Group Females Males Total

1968 5- 9 70.2 68.6 69.4 10 - 14 76.6 85.6 81.1 15 - 19 29.3 43.1 35.9 20 - 24 6.4 17.1 10.5 1974 5- 9 71.9 74.3 73.1 10 - 14 93.0 96.9 95.0 15- 19 45.5 61.0 53.3 20 - 24 9.4 18.5 12.8

Source: Visaria (1980) p. 212. East Asia 241 References

Anderson, J. "The Relationship between Change in Educational Attainment and Fertility Rates in Taiwan." Studies in Family Planning 6 (March): 72-81, 1975.

Arcelo, AA., and B.C. Sanyal Employment Opportunitiesand Career Opportunitiesafter Graduation: The Philippine Experience.IIEP/UNESCO study. Manila: Fund for Assistance to Private Education, 1987.

Balassa, B. "The Lessons of East Asian Development:An Overview."In "Why Does Overcrowded, Resource- poor East Asia Succeed--Lessonsfor LDCs?" Anthony M. Tang and James S. Worley (eds.), Economic Developmentand CulturalChange, 36 (April) (supplement): 5273-90,1988.

Becker, G.S. Economics of Discrimination.Chicago: University of Chicago Press (second edition 1971), 1957. Human------. Capital: A Theoreticaland EmpiricalAnalysis with SpecialReference to Education. New York: Columbia UniversityPress (second edition: National Bureau of Economic Research, 1975), 1964.

------. "Discrimination,Economic." In IntemationalEncyclopedia of Social Sciences, D.L. Sills (ed.) 208-10. London: Macmillan and the Free Press, 1968.

Benavot,A. "Education, Gender and Economic Development:A Cross-nationalStudy." Sociology of Education, 62 (January): 14-32, 1989.

Bird, C. Bom Female: The High Cost of Keeping Women Down. New York: Simon and Schuster, 1971. Blaug, M. "An Economic Analysis of Personal Earnings in Thailand." Economic Development and Cultural Change, (October): 1-32. (Reprinted in M. Blaug. The Economics of Education and the Education of A Economist Hants, Eng.: Edward Elgar, 1987, pp. 301-31.),1974.

Borcelle, G. Jobs for Women:A Pleafor Equalityof Opportunity,Technical Education, VocationalTraining and Employment. Paris: UNESCO, 1985.

Boserup, E. Women's Role in Economic Development. New York: St. Martin's Press, 1970.

Boudon, R. Education, Opporunity, and Social Inequality:Changing hospects in Westem Society. New York, Wiley-Interscience,1974.

Bowman, MJ. "The Human Investment Revolution in Economic Thought."Sociology of Education, 39 (Spring): 111-37, 1966.

Bowman, MJ., and CA Anderson, "The Participation of Women in Education in the Third World." In Women's Education in the Third Worktk ComparativePerspectives, G. Kelly and Carolyn Elliott (eds.), pp. 311-30. Albany. State University of New York Press, 1982.

Chamie, M. "National, Institutional, and Household Factors Affecting Young Girls' School Attendance in Developing Societies."Processed. Washington, D.C: International Center for Research on Women, 1983.

Chernichovsky,D., and OA. Meesook, "SchoolEnrollment in Indonesia."World Bank Staff WorkingPaper, 746. Washington, D.C.: World Bank, 1985.

China, Department of Planning, Ministry of Education. Achievement of Education in China Beijing People's Education Press. 1986. 242 East Asia Cochrane,S.H. 'Educationand Fertility:An ExpandedExamination of the Evidence."In Women'sEducation in the Third WorJd ComparatvePerpectives, 0. Kellyand Carolyn Elliott (eds.), pp. 311-30.Albany: State Universityof New YorkPress, 1982. Cochrane,S.H., and D.T.Jamison, *Educational Attainment and Achievement in RuralThailand." Inho4ictivily Assessmentin Education.New Directionsfor Testingand Measurement,15. Anita A. Summers(ed.), pp. 43-60.San Francisco:Jossey-Bass, 1982. Cochrane,S.H., D. O'Hara, and J. Leslie,'The Effectsof Educationon Health."World Bank Staff Working Paper, 405.Washington, D.C.: World Bank, 1980. Collier, P. "Womenin Development:Defining the Issues."Processed. PPR Working Papers, WPS 129. WashingtonD.C.: World Bank, 1988. Corner,L. "FemaleLabor ForceParticipation and Earningsin Indonesia!In Women'sPardcipaton in Asia and the Pacific,pp. 131-59. Bangkok:United NationsEconomic and SocialCommission for Asia and the Pacific, 1987.

Dasgupta, A.K. Growth Developmentand Welfare:An Essay on Levels of Living. Oxford: Basil Blackwefl,1988. Deble, I. The School Education of Girls. Paris: UNESCO, 1980. de Tray, D. "Children'sEconomic Contributions in PeninsularMalaysia." A RAND Note. Processed.Santa Monica,Cal.: RAND Corporation,1979. "Govermnent--.- Policy, Household Behavior, and the Distributionof Schooling:A CaseStudy of Malaysia" In Researchin Population Economics: A ResearchAnnual 6, T.P. Schultz (ed.), pp. 303-36. Greenwich, Conn.:Jai Press Inc, 1988. Don,F.H. "Educational Opportunities for Girlsin MalaysianSecondary Schools." In Womenin Educaton:World Yearbookof Education1984 SandraAcker (ed.), pp. 110-22.London: Kogan Page, 1984. Dube,L. Studieson Womenin SoutheastAsia. Bangkok-UNESCO, Regional Office for Educationin Asia and Oceania,1980. Eliou,M. "Equalityof the Sexesin Education:And NowWhat?" In "SexDifferences in Education, Comparative Education,23 (specialnumber 10,Patricia Broadfoot and MargaretB. Sutherland,eds.): 59-67, 1987. Encarnacion,Jr., J. "Fertilityand LabourForce Participation:Philippines 1968." Working Paper, 2. Population and EmploymentResearch Project, World Employment Programme, International Labour Organization, Geneva,1974. Encarnacion,Jr., J., GA Tagunicar,and R.L. Tidalgo,"Unemployment and Underemployment."In Philippine Economic Problemsin Perspective,pp. 136-77.Manila: Universityof the Philippines,Institute of Economic Developmentand Research,1976. Fuller,B. "RaisingSchool Quality in DevelopingCountries: What InvestmentsBoost Learning?" World Bank DiscussionPaper, 2. Washington,D.C.: World Bank,1986. Goldstein,S. "TheInfluence of Labor Force Participationand Educationon Fertilityin Thailand."Populaton Studies, 26: 419-36, 1972. Greenhag S. "SexualStratification: The Other Side of 'Growthwith Equity'in East Asia."Popuation and DevelopmentReview, 2 (June): 265-314,1985. East Asia 243

Harman, AJ. "Fertility and Economic Behavior of Families in the Philippines."Processed. Santa Monica, Cal: RAND Corporation, 1970.

Harnqvist, K. "Social Demand Models." In Economics of Educaton: Research and Studies, George Psacharopoulos (ed.), pp. 356-63. Oxford: Pergamon Press, 1987.

Hermalin, A. "Empirical Research in Taiwan on Factors UnderlyingDifferences in Fertility."Studies in Family Planning.5 (October): 314-24, 1974.

Hermalin, A., JA. Seltzer, and C-H Lin, 'Transition in the Effect of Family Size and Female Education Attainment: The Case of Taiwan."Comparative Education Review, 26 (June): 254-70, 1982.

Heyneman, S.lF.,- . Jamison, and X. Montenegro, "Textbooksin the Philippines;Evaluation of the Pedagogical Impact of a NationwideInvestment," Educational Evaluation and PolicyAnalysis,6 (Summer): 139-50,1984. Heyneman, S.P., and W. Loxley, 'The Effect of Primary School Quality on Academic Achievement Across Twenty-nine High and Low-incomeCountries." American Joumal of Sociology,88: 1162-94,1983. Heyzer, N., (ed.). Missing Women: DevelopmentPlanning in Asia and the Pacific. Kuala Lumpur: Asian and Pacific Development Centre, 1985.

Hirschman, C. "Political Independence and Educational Opportunity in Peninsular Malaysia."Sociology of Education, 52: 67-83, 1979.

Hooper, B. "China's Modernization: Are Young Women Going to Lose Out?" Modem China, 10 (July): 317- 43, 1984.

Huq, M.S. Education and DevelopmentStrwegy in South and SoutheastAsia. Honolulu: East-West Center Press, 1965.

Jamison, D.T., and M.E. Lockheed,"Participation in Schooling: Determinants and Learning Outcomes in Nepal." Economic Developmentand CulturalChange, 35 (January): 279-306,1987.

Jimenez, E., and M.E. Lockheed, "The Relative Effectiveness of Single-sex and Coeducational Schools in Thailand." PPR Working Paper, WPS 29. Processed. Washington, D.C.: World Bank, 1988.

Jimenez, E.; M.E. Lockheed, and N. Wattanawaha, "The Relative Efficiencyof Private and Public Schools: The Case of Thailand." WoridBank Economic Rewew, 2 (May): 139-64,1988.

Johnstone, JA. "The Locationand Distribution of Primary Schoolsin the Asian Region."Bulletin of the UNESCO Regional Office for Education in Asia, 17 (June), 1976.

------. "Indicators of Indonesian Provincial Education Systems." Badan Penelitian Dan Pengembangan. Pendidikan Dan Kebudayaan, 1978.

Kaneko, M. "Education and Labor Force Composition in Southeast and East Asian Development." The DevelopingEconomies, 12 (March): 47-68, 1984.

"---The--Educational Composition of the World's Population: A Database." EDT 29 (revised edition). Education and Training Department Discussio. ?aper. Processed. Washington, D.C.: World Bank, 1987. Kelly,G. "Women's Access to Education in the Third World. Myths and Realities."In Women and Education: World Yearbookof Education 1984 Sandra Acker (ed.), pp. 81-89. London: Kogan Page, 1984. 244 East Asia

.Setting------. State Policy on Women's Education in the Third World: Perspectives from Comparative Research." In "Sex Differences in Education," Patricia Broadfoot and Margaret B. Sutherland (eds.), ComparativeEducation, 23 (special number 10): 95-102, 1987.

Kelly,G., and C. Elliott, (eds.). Women's Education in the Third World. ComparativePerspectives. Albany State University of New York Press, 1982.

King, E.M. "The Effect of Family Size on Family Welfare: What Do We Know?" In Population Growth and Economic Development:Issues and Evidence, N. Gale Johnson and Ronald D. Lee (eds.), pp. 373-411. Madison, Wisc.: University of Wisconsin Press, 1987.

King, E.M., and LA. Lillard, "Education Policy and Schooling Attainment in Malaysia and the Philippines." Economics of Education Review, 6: 167-81,1987.

King, E.M., J.R. Peterson, S.M. Adioetomo, LJ. Domingo, and Syed, S.H. "Change in the Status of Women Across Generations in Asia." Processed. Santa Monica, Cal: RAND Corporation, 1986. King, U. "World Religions, Women and Education." In "Sex Differences in Education," Patricia Broadfoot and Margaret B. Sutherland (eds.), ComparativeEducation, 23 (special number 10): 35-49, 1987.

Korean Council for University Education. Equity, Qualityand Cost in HigherEducation. Bangkok: UNESCO Principal Regional Office for Asia and the Pacific, 1988. Kurian, G.T. Encyclopediaof the Third World. New York: Facts on File, Inc., 1982 Lazo, L.S. "Work and Training Opportunities for Women in the Philippines." Processed. Islamabad: ILO- APSDEP, 1984.

Lee, V.E., and M.E. Lockheed, "The Effects of Single-sexSchooling on Student Achievement and Attitudes in Nigeria." Processed. Washington, D.C.: World Bank, 1989.

LeVine, RA. "Influences of Women's Schooling on Maternal Behavior in the Third World." In Women's Educationin the Third World. ComparativePerspectives, G. Kelly and Carolyn Elliott (eds.), pp. 283-310. Albany: State University of New York Press, 1982.

Levy,M.B. "Determinants of Primary School Dropouts in DevelopingCountries." Comparative Education Review, 15 (February): 44-58, 1971.

Lockheed, M.E., S.C. Vail, and B. Fuller, "How Textbooks Affect Achievement in Developing Countries: Evidence from Thailand,"Educational Evaluation and PolicyAnalysis, 8 (Winter): 379-92, 1986. 1 idden, J.F. "Dis.rimination--A Manifestationof Male Market Power?" In Sex, Discriminationand Division of Labor, Cynthia B. Lloyd (ed.), pp. 146-74.New York: Columbia University Press, 1975.

Mazumdar, D. UrbanLabor Market and IncomeDistribution: A Study of Malaysia.New York: Oxford University Press for the World Bank, 1981.

McGrath, P.L. The UnfinishedAssignment. Equal Educationfor Women. World Watch Paper, 7. Washington, D.C.: World Watch Institute, 1976.

Meesook, OA. "Financing and Equity in the Social Sectors in Indonesia." World Bank Staff Working Papers, 703.Washington, D.C.: World Bank, 1984.

Mincer, J. "On-the-job Training: Costs, Returns, and Some Implications." Joumal of Political Economy, vol. 8, no. 2 (part 2 supplement): S76-S108,1962. East Asia 245 Mincer,J., and S. Polacheck,"Family Investments in HumanCapital: Earnings of Women."Joumnal of Political Economy, 18 (1) 3p-24, 1974. Montgomery,M.R. "Female First Marriage in East and SoutheastAsia: A Kiefer-NeumannModel." Discussion Paper, 510.New Haven:Economic Growth Center, Yale University,1986. Montgomery,M.R., and D.S. Sulak,"Female First Marriagein East and _outheast Asia: A Kiefer-Neuman Model."Joumal of DevelopmentEconomics, 120:33-52, 1986. Moore,KM. "Women'sAccess and Opportunity in HigherEducation: Toward the Twenty-frstCentury." In "Sex Differencesin Education."Patricia Broadfoot and MargaretB. Sutherland(eds.), Comparative Education, 23 (specialnumber 10):23-24, 1987.

Nagib, L. FactorsRelating to Female Work ForcePaticipation in Three CentralJavanese Communities.Jakarta: NationalInstitute of Economicand SocialResearch, Indonesian Institute of Science,1986. NationalCommission on the Role of FilipinoWomen. Filipino Women in Education.Manila, 1985. Neher,C.D. "SexRoles in the Philippines:The AmbiguousCebuana." In Womenof SoutheastAsia; PennyVan Esterik (ed.) pp. 154-75.Occasional Paper, 9. NorthernIllinois University, Center for SoutheastAsian Studies,1982. Niehoff,A.H. "TheravadaBuddhism: A Vehiclefor TechnicalChange." Human Organization, 23: 108-12,1964. Oshima,H.T. "Humar Resourcesin East Asia's SecularGrowth." In "WhyDoes Overcrowded,Resource-poor East Asia Succeed--Lessonsfor LDCs?"Anthony M. Tang and James S. Worley (eds.), Economic Developmentand CulturalChange, 36 (April) (supplement): S103-22,1988. Papanek, H. "Class and Gender in Education--EmploymentLinkages." ComparativeEducation Review, 29 (August):317-46, 1985. Patel,SJ. "EducationalMiracle in Third World,1950 to 1981."Economic and Political Weekly, 20 (August3): 1312-17,1985.

Postlethwaite,T.N., and R.M. Thomas, (eds.). Schoolingin theASEANRegion: Pnmary and SecondaiyEducation in Indonesia, Malaysia,the Philippines,Singapore, and Thailand Oirord: Pergamon Press, 1980.

Psacharopoulos, G. Returns to Elucation: An Inte.-nationalComparison. San Francisco:Jossey-Bass, 1973. "---..Returns to Education:A FurtherInternational Update and Implications." Joumal ofHuman Resources. 20 (Fall):584-604, 1985. Psacharopoulos,G., and A.-M.Arriagada, "The Educational Composition of the LabourForce: An International Comparison."Intemational Labour Review, 125 kSeptember-October):561-74, 1986.

Psacharopoulos,G., and M. Woodhall,EducationforDevelopment:AnAnalysis of Investment Choies. New York, OxfordUniersity Press for the WorldBank, 1985. Ram, R. "SexDifferences in the Labor Market Outcomesof Education."In Women'sEducation in the Third World ComparativePerspectives, G. Keliyand CarolynElliott (eds.), pp. 203-07.Albany: State University of NewYork Press,1982. Rao, V.V.B.'Income Distribution in East Asian DevelopingCountries." Asian-Pacific Economic Literatura, 2 (March): 26-45, 1988. 246 East Asia

Rosenzweig,M.R. "ChildInvestment and Women."In Sex, Discriminationand Divisionof Labor, Cynthia B. Uoyd (ed.), pp. 269.91.New York: Columbia UniversityPress, 1975.

Rosenzweig, M.R., and RE. Evenson, "Fertility, Schooling and Economic Contribution of Children in Rural India." EconomeaicA 45 (July): 1065-79,1977. Rosenzweig,M.R., and T.P. Schultz, Fertilityand Investmentsin Human Capital: Estimates of the Consequences of Imperfect Fertility Control in Malaysia."Bulletin No. 87-1. Processed. Minneapolis: Department of Economics, Economic Development Center, 1987.

Safilios-Rothschild,C. Sex Role Socializaton and Sex Discrimination:A Synthesis and Critiqueof the Literamre. WashingtonD.C.: Department of Health, Education, and Welfare, National Institute of Education, 1979. .The-.....-. Role of the Family. A Neglected Aspect of Pove-ty." In Implementing Progiumsof Human Development,Peter T. Knight (ed.), pp. 311-72.World Bank Staff Working Paper, 403. Washington,D.C.: World Bank, 1980.

Salaff,J.W. "The Status of Unmarried Hong Kong Women and the SocialFactors Contributing to Their Delayed Marriage. .opulationStudies, 30: 13-34, 1976. Salvador-Burris, J. "Modernization, Social Class, Childrearing anu Cognitive Performance: Their Inter- relationships in an Urban Communityin the Philippines."Ph.D. dissertation, Universityof Chicago. Cited in LeVine 1980, 1977.

Sanyal, B.C. HigherEducaton and Employment:An Intemnw#onalComparatve Analysis. London: Falmer Press, 1987.

Schiefelbein,E., and J.P. Farrell, "Women,Schooling, and Work in Chile: Evidence from a LongitudinalStudy." In Women's Educaton in thc Third World.Comparative Perspectives, G. Kelly and Carolyn Elliott (eds.), pp. 228-48. Albany: State Universityof New York Press, 1982. Schultz, T.P. "School Expenditures and Enrollments, 1960-80:The Effects of Income, Prices, and Population Growth." In Population Growth and Economic Developmenc Issues and Evidence, D. Gale Johnson and Ronald D. Lee (eds.), pp. 413-76. Madison: Universityof WisconsinPress, 1987. -Women-- and Development:Objectives, Framework and PolicyInterventions." Processed. New Haven: Department of Economics, Yale University,January 1989.

Schultz, T.W. "Investmentin Human Capita'.American Economic Rev w, 51 (March): 1-17, 1961.

-----. Investment in Hunan Capital. The Role of Educaton and ,j Research. New York: Free Press, 1971.

Scott, G.L. 'Indonesian Women and Development."Processed. Washington, D.C.: World Bank, J85.

Selowsky, M. "A Note on Pre-school Investments in Human Capital in Developing Countries." Economic Developmentand CulturalOange, 24 (July): 707-20, 1976. ---. wTheEconomic Effects of Investment in Children: A Surveyof the Quantitative Evidence."In Child DevelopmentInfomw*ion and the Formaton of Public Policy:An IntemationalPerspective, T.E. Johnson (ed.), Springfield,IIL: Charles C. Thomas, 1982.

------. Women's Access to Schooling and the Value Added of the Educational System: An Application to Higher Education." In Women and Poveny in the Thin! Worl4 Mayra Buvinic,Margaret A. Lycette, and William Paul McGreevey (eds.), pp. 177-97.Baltimore: Johns Hopkins UniversityPress, 1983. East Asia 247

Shields,N.G. 'Female Labor Force Participationand Education:Developing Countries." In Ihe Intma(ti Encyclopediaof Educador Rsearh and Studies, Torsten Husen and T. Neville Postlethwaite (eds.), pp. 1854-61.Oxford: Pergamon Press, 1987. Singh,J.S. 'The World of Work.lIn UnivmeityEducaton and Emph*vnentin MoLakqsiA.A. Aziz,C.S. Buan, L.K Hock,and B.C.Sanyal (eds.), pp. 147482.HEP Research Report, 66. Paris:International Institute for EducationalPlanning, 1987. Smith,P.C., and P.P. Cheung,"Social Origins and Sex DifferentialSchooling in the Philippines."In Women's Educaion in the Third WordS Compwe Perspectves G. KeRyand Carolyn Elliott (eds.), pp. 51-67. Albany:State Universityof New York Press, 1982. Smock,A.C. Women's Educaton in DevelopingCounbies: Opponwddes and Outcomes. New York: Praeger, 1981. Stromquist,N.P. "Empowering Women through Knowledge: Policies and Practicesin InternationalCooperation in BasicEducation." Report preparedfor UNICEF. Processed.1986. Sundaram,J.K, and T.P.Leng, "Not the BetterHalf: Malaysian Women and DevelopmentPlanning." In Missing TWomen:Development Planning in Asia and the Pacifiw,Noeleen Heyzer (ed.), pp. 1-72. Kuala Lumpur: Asian and PacificDevelopment Centre, 1985. Sutherland,M "Womenin HigherEducation: Effects of Crisesand Change." HigherEducation, 17 (specialissue on women,Sheila Slaughter, ed.): 479-90,1988. Tang,A.M., and J.S. Worley,(eds.). "WhyDoes Overcrowded,Resource-poor East Asia Succeed--Lessonsfor the LDCs?" Economic Developmentand CQhurd Change,36 (April) (supplement), 1988. Thomas,R.M. "TheIslamic Revival and IndonesianEducation." Asian SunW, 28 (September):89,-915, 1988. Thomas, RM., and T.N. Postlethwaite,(eds.). Schoofingin East Asia: Forcesof Change:Fomnol and Nonfomwal Education in Japan, the Republicof China rhePeople's Republic of Ouna, South Koea, North Korea,Hong Kong and Macau.Oxford. Pergamon Press, 1983. Tidalgo,R.L.P. "The Integration of Womenin Developmentand PhilippineDevelopment Planning." In fissing Women:Development Planwng in Asia and the Pacific, Noeleen Heyzer (ed.), pp. 354405. Kuala Lumpur: Asian and PacificDevelopment Centre, 1985. Tilak,J.B.G. "EducationalPlanning and InternationalEconomic Order." ComparaveEducadton 18 (2): 107- 21, 1982. Economics of Inequalivtin Educaion. New Delhi: Sage Publications, 1987a. .Returns------.to Investmentin Women'sEducation." Sixth Annual Conference of the Societyof the Study of Regional Disparitieson "Womenand Access to Resources."Processed. Mysore: Central Food & TechnologicalResearch Institute, 1987b. ------ComparativeDevelopment Indicators. Research Project on The PoliticalEconomy of Poverty,Equity and Growth.RPO 673-73.Processed. Washington, D.C.: World Bank, 1988. "ComparativeDevelopment Indicators." In Essas on Povet, Equit, and Gwt, G. Psacharopoulos (ed.). Washington,D.C.: World Bank (in press) 1990. Tonguthai,P. "Womenand Workin Thaibnd and the Philippines."In Women'sEconomic Partcipaton in Asia and thePaciflc, pp. 191-219.Bangkok: United Nations Economic and SocialCommission for Asia and the Pacific,1987. 248 East Asia

UNESCO. ComparativeAnalysts of Male and Female Enrolbnentand Iltiteracy.Current Surveys and Research in Statistics. Paris: UNESCO, 1980a. Wastage------. in Pnmary and GeneralSecondwy Education: A StatisticalStudy of Trends and Patterns in Repetitlonand Dropout. Current Studies and Research in Statistics. Paris: UNESCO, 1980b. ------Female Partcipationin HigherEducation: Enrollment Trends, 1975-1982 Current Surveysand Research in Statstics. Paris: UNESCO, 1985. ------. Statistical Yearbook Paris: UNESCO, 1987 and 1990. Trends------. and Projectionsof Enrolment by Level of Education and by Age, 1960-2025.Paris: UNESCO, 1989. Towards------. Universalizon of PrimaryEducation in Asia and the Pacifc: Regional Overview.Bangkok: UNESCO Regional Office for Education in Asia and the Pacific, 1985. UNESCO-ROEAP. The Drop-out Problem in Primary Education: Some Case Studies. Bangkok: UNESCO Regional Office for Education in Asia and the Pacific, 1984. United Nations. UN StatisticalYearbook 1985. New York: United Nations, 1985. Van Esterik, P., (ed.). Women of Southeast Asia. Occasional Paper, 9. Center for Southeast Asian Studies, Northern Illinois University,Center for Southeast Asian Studies, 1982. V.isaria,P. "Povertyand LivingStandards in Asia: An Overviewof the Main Results and Lessons of Selected Household Surveys."LSMS WorkingPaper, 2. Washington, D.C.: World Bank, 1980. Wang, B.-L.C. 'Sex and Ethnic Differences in Educational Investment in Malaysia: The Effect of Reward Structure."In Women'sEducation in the Third World.Compratve Perspectives,G. Kelly and Carolyn Elliott (eds.), pp. 68-87. Albany: State Universityof New York Press, 1982. Ward, B.F. (ed.). Women in the New Asia Paris: UNESCO, 1963. Weekes-Vagliani,W., in collaborationwith Bernard Grossat. Women and Development:At the Right 7imefor the Right Reasons. Paris: Organisation for Economic Co-operation and Development, 1980. Whyte, R.O., and P. Whyte, "Rural Asian Women: Status and Environment."Research Notes and Discussion Paper, 9. Singapore: Institute of Southeast Asian Studies, 1978. Woodhall, M. "Investmentin Women: A Reappraisal of the Concept of Human Capital."Intemational Revew of Education, 19 (special issue: "The Education of Women,"Evelyn Bromhead, ed.), pp. 9-29, 1973a. "Economic------. Returns to Investment in Women's Education."Higher Education 2: 275-99.(Reprinted in Economics and Education:A Reader. Carolyn Baxter, PJ. O'Leary and Adam Westoby,eds., pp. 92-111, 1977.) New York: Longman/Open UniversityPress, 1973b.

World Bank. Indonesia: Wagesand Employnent. World Bank Country Study. Washington, D.C.: World Bank, 1985. ------. China: Growth and Developmentin Gansu Province.World Bank Country Study. Washington, D.C.: World Bank, 1988a. East Asia 249

Social-----.Indicators of Development198& Washington, D.C.: World Bank, 1988b.

------. WorldDevelopment Report 1989. Washington, D.C.: World Bank, 1989. Yue, C.S. "Women's Economic Participation in Malaysia."In Women's Economic Participationin Asia and the Pac{fic pp. 163-90.Bangkok United Nations Economic and Social Commission for Asia and the Pacific, 1987. EducatingWomen 251 Chapter 8. Educating Women: Lessons from the Past

Rosemay T. Bellewand ElizabethM g

Introduction Expandingeducation, especially basic education,has been an objectiveof educationpolicy in developing countriesover the past two decades. The reasons for this objectiveare clear. Basic educationis often considereda right whichnations have a responsibilityto guaranteeto each generation. And, the benefits to educationare by now wellestablished. The evidenceis overwhelmingthat educationraises the quality of life;it improveshealth and productivityin marketand non-marketwork, increases individuals' access to paid employment,and ol' -n facilitatessocial and politicalparticipation. The evidenceis alsoclear that these benefitsare especiallylarge in the case of girlsand women.Educated womenhave smaller families,fewer of their childrendie in infancy,and the childrenwho surviveare healthier and better educated. Educatedwomen are also better equippedto enter the paid labor force whichis criticalto the survivalof the manyfemale headed householdsin developingcountriesY" It is not surprising then, that nationswith higher levels of femaleschool enrollment in the past todayshow higher levels of economicproductivity, lower fertility, lower infant and maternal mortality,and longer life expectancythan countriesthat have not achievedas high enrollmentlevels for girls (Schultz1989). The previouschapters indicate, however, that manyparents and societiesunderinvest in girls' education. What role can publicpolicy have in raisingthe levelsof femaleeducation? Compulsory education laws and tuition-freeprimary education clearly have not been sufficient.Direct intervention through programs have been necessaryand have sometimessucceeded. Past researchprovides few hard and fast rules to guide governmentsin their efforts,but it does suggestthat where enrollmentin primaryschool is low, efforts shouldbe targetedat expandingcomplete primary schooling as opposedto secondaryor highereducation. Not onlyare the.rates of return highestat the primarylevel but underthe best of circumstances,at least fiveto eight yearsof primaryschooling are necessaryto acquirethe readingand math skillsessential for operationalliteracy and numeracy(Lockheed and Verspoor1990). Based on this information,governments wouldbe wiseto investin basiceducation, in frmal and nonformalschool settings. In countrieswhere all boysand girlsare enrolledin primaryschool--such as manyin LatinAmerican and East Asia-interventions shouldbe directed at reducinggender differentialsin enrollmentand attainmentat the secondaryand tertiarylevels. Beyondthese directions,little can be said withmuch certainty, particularly about the most cost-effectivemeasures to boost girls' attendance. Motivatedby the uncertainty,we set out to identifythe approachesundertaken by governments,nongovernmentalorganizations, donor agenciesand communitiesto raise girls' and women'sattendance in educationalprograms. This chapter discusses these approaches, distills the lessonslearned from experience, and examinesthe conditionsunder which the strategiesseem to workor faiL(See table8.1 for an overview of the approachesincluded.) The surveyis by no meansexhaustive; it selectsthose interventionsfor which performancedata are available. Despite this selection,our discussionof some approacheswill seem impressionistic.Most initiatives have not been evaluatedcarefully to permitstrong conclusions about their effectiveness.Therefore, at this time,advancing females' education in certainsettings requires proceeding with best guesses. It involvesdesigning strategies based on what has worked well under similar

We are gratefulto MaureenPetronio who wrote most of the boxesin the chapter. In Jamaica,for example,36 percentof al householdsare headedby women;in El Salvador40 percent are headed by women(World Bank 1988a). 252 EducatingWomen circumstancesand based on what theoreticallymight work. It requires experimentationand carefuF monitoring. We do not prescribe a single strategy for any country or group of countries often, a combinationof interventionsis needed. Howgovernments, communities and donors approachthe issue willdepend on the country-specificcontext, including the existingsupply of schools,prevailing cultural and social norms, famiiis' incomesand productiveactivities, women's opportunities for paid work, and the qualityof education. With these caveatsin mind,we begin withthe obvious.

Table 8.1 Summaryof Interventionsto Raise Female Enrollment

Past Approaches County Descwtpdonof inmevew iewed Year began

Increase school supply Bhutan Built 'extended' primary school classrooms 1988 in rural areas (World Bank funding) Egpt Built primary schools in rural areas 1981 Mali Build and renovate school buildings 1989n/ Yemen Established vocational centers for women 1987 Build appropriate school Bangladesh Built primary schools and teacher faciHties training centers 1985 Mali Provided communityschools with pedagogic support; in new WB project, medersas also to receive pedagogic materials 1989&/ Kenya Secularizedcurricula in Koranic schools to 1985 attract more students Pakistan Secularized curricula in mosques 1979 Provide sanitation and water facilities in schools and build boundary walls (Sindh) 1990W/ Recruit female teachers Nepai Trained rural females with secondary 1971 education as school teachers; those without were trained to qualify Pakistan Recruited female teachers in rural areas 1984 and trained them there; provided female teacher residences Somalia Established teacher incentive systems Yemen Built separate urban primary teacher training institutes for women, and pilot institutes 1975 in rural areas to attract rural women 1987

Lower cost of uniforms Bangladesh Distributed free uniforms to primary girls 1981 Pakistan Abolish required uniforms in rural areas 1990/ Provide scholarships Bangladesh Offered scholarshipsto girls in secondary 1982 schools (USAID funding until 1988;Worid Bank) Guatemala Offered scholarshipsto girls in primary 1987 grades (USAID pilot funding; future World Bank funding) India Offered scholarships to girls in primary school (govemment funding) Nepal Offered scholarships to girls in primary school (govemment funding) Early 80s Establish day-care centers China Established worksite day-care centers and preschool centers, and sibling care at primary schools Mid-80s Colombia Built community-basedcenters ("Hogares de 1987 Bienestar Infantil')

...continued Educating Women 253 Table 8.1 Summaryof Interventionsto Raise Female Enrollment ...continued

Past Approaches Couny Descrption of unerventonsreviewed Yearbegan

Adapt labor-saving home Burkina Paso Distributed labor-savingmachines to 1967 technologies encourage nonformal education of women (UNESCO sponsored) Nepal Disseminated fuel efficient stoves 1977 Design flexible school Bangladesh Introduced programmed instruction in 1980 schedules selected rural schools Colombia Programmed learning in *escuelasnuevas" 1975 El Salvador Introduced program instruction Late 80s Indonesia Introduced multigrade teaching Late 70s- Liberia using self-taught learning materials early 80s Philippines taught learning materials Build safety nets Bangladesh Built lower primary schools in rural areas; 1983 known as BRAC schools (funded by UNESCO and Norway) India Established nonformal evening schools for 1979 out-of-school youths (supported by govemment and UNICEF) Promote gender-neutral Bangladesh Revised textbooks to improve perception instruction of women's roles in familyand society 1988 China India Same as above 1980s Kenya Educating community Mali Launched media campaignsto advertise value of education of girls 1989A/ Morocco Developed materials and extension service promoting girls' education 1989W/ Delay childbearing Guatemala Used girls' scholarship program to reward avoidance of prepancy 1987 Improve girls' nutrition Jamaica Provided school breakfast program Offer training in non- Chile Built vocationalcenters to train middle- 1968 traditional occupations level male and female technicians Morocco Established industrial and commercial 1979 training program for men and women Tanzania Established training centers near primary schools for unemployedfemales 1975 Yemen Built vocational training centers 1987 Alleviate poverty Bangladesh Established women's income-earningprograms 1970s

Note: W Although the project has been launched, the specific intervention pertaining to girls or women may not have started yet. 254 EducatingWomen Past Approaches ImprovingAccess to Schools Studentscannot attend schoolwhen places are in short supplyor whenschools are locatedfar from home. Recognizingthis, planners have developed various low-cost expansion strategies to expandaccess and bring schoolingcloser to home. These effortshave resulted in experimentationwith multigradeclassrooms, double-shifting,and with feeder and satelliteschools at the primarylevel (See Box 8.1), with radio education and correspondencecourses at the postprimarylevels, and with literacyprograms for adults. Educational opportunitiesfor femaleshave also been increasedby eliminatingdiscriminatory admissions practices and institutingquotas that reserveplaces for them in educationalprograms.

U 6. _IF to ,h l&1 ToodFarst Wl::."ofromtamontas E tded

The landis'rugged.. .Ri.vers fow from tall mountainsto valeys coveredwith dense forests. Villages are remoteand widelyscattered. Schoolfacilities are scar,.overcrowdedand sbabby. Teachers are in short supply. If childrengo to''schol at'al, theymuwalk Oni or findboarding accommodations near -'one-d Bhutan's1,47 primary school. .Be'a Buuta'n'i ita never 'hada formalcensus, enrollment rates are estimates;they suggest that female:eonimrisew *ly thiirty-rivepercent of the twentypercent enrollmentrate in primaryschools.:

...To increasefemale partictn Sch the goerme i'ofBhutan is planing to provide -miiti.gradeeducation fom there-Primay eIel thrigh Clas L extendedclassroom model (ECR). -ECs or. lower'primmaiysco.ol il arouid DevelopmentService Center Schools(DSC) ;c-willhavea boardig fcles fM ildren in Cla V -VI. Childrenwho ,sucf pass Css T l feed'intothe- iC. Th gonmeutw recruitand pay for teachers :Incomnmities- tat can provide EC'Rfailities nd'"assue-a minimm of 100 students. Bhutan believesECRs will incre-ase femalefenrollrbentbytdigthe if6rboarding schoolsat the lower g'ade or-mitories at the :upp'eiarylevels ilM' ite parentsto send daughtersto school tf ears fersotresi t a u n a d .byinsuring that childrenwho : completeprimary -school will have the opp'nit>'to'otinue their education. s'Wties:World Bank(1988b), World a (9.

These strategiesto expandaccess are necessaryto increasegirls' enrollment, but one lessonwe learn from the supply-sidemeasures is that they are not alwayssufficient. When the demandfor girls' educationis low, familieswill not send their daughtersto school,even if one is available.The followingexperiences of Egypt, Mali and Yemenillustrate this. Egypt,in its effortsto expandprimary education to rural children,built 400 new primaryschools in rural areas between1981 and 1987.The increasedavailability of spacespermitted more childrento enroll,and the proportionof school-agegirls enrolled increased from 56 to 74 percent;boys' enrollment increased from 94 to 100 percent. Evaluatorsof the expansioninitiative concluded that the existingdemand for girls' educationhad alreadybeen met, and that additionalschool construction in the originalsites woulddo little to attract the girlsstill out of school(Robinson and others 1987). EducatingWomen 255 A similar situationoccurred in the Koulikororegion of Mali where a school expansionprogram was launchedand where small multigradeschools are common. Accordingto a report, overallenrollment actuallydeclined by 1.5 percenta year between1982 and 1986despite a 3 percentincrease in the number of schools,a 14 percentexpansion in the numberof dasses,and a 21 percentrise in the numberof teachers (Haughton,1986). Female enrollmentfell faster-by 2.6 percent each year in that period tMinistryof Education,Mali, n.d.). The report concluded.

There is probablyrelatively little unsatisfied demand for vublicschooling, as it currentlyexists, in this region. In that case an upper limit on net public schoolenrollments of about a quarter can be expectedin rural areas if the onlypolicy pursued is expansionin the numberof schools(Haughton 1986).

The inadequacyof expandingplaces in educationalprograms as a strategyto raisefemale enrollment is also evidentin the expansionof vocationalprograms. In the People'sDemocratic Republic of Yemen, the governmentaccepted, as policy,a commitmentto increasethe role of womenin economicdevelopment. This commitment,coupled with a foreseencritical shortage of skilledtechnicians and clericalworkers, led the governmentto establish a networkof 14 VocationalTechnical Centers to meet the needs of the industrial,agricultural and cwmmercialsectors. The projectprovided for 15,20and 75 percentparticipation of femalestudents in the industriaLagricultural, and commercialcenters, respectively. The outcomewas disappointing.By 1984, women comprised only four percentof commercialstudents (7 out of 166)and zero percent of agriculturaland industrialstudents. Twoproject evaluations identified several reasons for the lowfemale attendance. They pointed to the same conclusion:project designersdid not take into considerationthe preferencesof Yemenigirls and their families,cultural norms, nor the economy. Femalespreferred employment in manufacturingand fishing whichprovided reasonable incomes without a diploma,and those who did attend secondaryor post- secondaryschool did not work in agricultureor industryunless it was in the Ministryin Aden. Parents outsideAden also opposed co-educationwhile early marriageprevented many girls from continuingin school(UNESCO 1985a, World Bank 1987c). These experiencesdemonstrate that simplyexpanding educational programs may be insufficientto increase girls' enrollment.For programsto be fullyutilized, the demandfor educationmust emanate from families and the community.Where parentsare concernedabout the physicaland moral safetyof their daughters, wherethe direct and opportunitycosts of attendanceare too high,and wherethe benefitsto educationtoo few, schoolexpansion policies will only be effectiveif they are accompaniedby policiesthat lower the cultural,direct, or opportunitycosts of educationand/or raise the benefits.

Building CulturallyAppropriate Facilities Schoolsmust conformto communities'cultural standards, especially the standardsof proprietyto which femalesare held. In parts of North Africa,the MiddleEast, East and SouthAsia and Africa's Sahelian region,girls' and youngwomen's activities are governedby socialpractices that restricttheir presencein publicplaces and their interactionwith males. In theselocations, parents may insist that malesand females be separatedand theymay he more concernedwith the availabilityof closedlatrines than theyare withthe supplyof desksand chairs.6 / Pakistanhas respondedto these concernsby buildingboundary walls around girls'schools (World Bank 1987a,1987b). Bangladesh has respondedby providingsanitary facilities which has had a positiveinfluence on community,teacher and student attitudestoward school and addressedan importantparental objectionto girls' attendance(World Bank 1985a).

V A surveyof 2,000Pakistani parents reported that they did not mind the absenceof desksand chairsin girls' schools,but two-thirdscriticized the absenceof latrines(Culbertson and others 1986). 256 Educating Women

Some countries have also responded to parents' concerns about propriety by supporting the expansion of Koranic schools and activelyrecruiting and training female teachers. The evidencesuggests that low quality programs may limit the success of Koranic schools and that increasingthe number of female teachers is a promising strategy to raise female enrollment. Koranic schools are under the control and supervision of the Imam, a revered member of the community. They have strong traditional roots, and provide a proper and sheltered environment and a religious education that is more acceptable to "traditional"parents. Historically,Koranic (or mosque) schools offered instruction only in the Koran and Islamiya. Mali, Pakistan, Bangladesh, and Kenya have supported the accreditation of Koranic schools by introducing the primary school curriculum and a trained teacher to supplement religious educationY Mauritania experimented with introducing math and reading in pilot Koranic schools via radio broadcasts and by providing learning materials and supervisorysupport from the inspectorate (World Bank 1983a). The Gambian government also hopes to improve female school attendance by raising the quality of education in Koranic schools (madressas). The government is working with Muslim organizationsto establish a school calendar,introduce a broader curriculum,and provide better trained teachers (World Bank 1990b). The results of these efforts are mixed. In recent years, girls' enrollment in Mali's Koranic schools (medersas) has grown rapidly. Medersas currently enroll 23 percent of all primary students, and girls account for 47 percent of them, compared to 32 percent of enrollment in govermnent primary schools (World Bank 1989d). In Pakistan, parents have not responded as enthusiasticallyas expected to their government's initiativeto expand mosque schools. In 1986,girls constituted only 30 percent of students in those schools compared to 32 percent in government primary schools. Supporters of the expansion say that mosque schools provide places that could not ordinarily be provided by public schools. However, others question the benefits, claiming that the quality of education provided in mosque schools is lower than that provided in government primary schools (See Box 8.2).

IV See Haughton 1986;World Bank 1989d;Harley 1979;Warwick, and others 1989; World Bank 1987b; UNESCO 1985b;Eisenmon and Wasi 1987. EducatingWomen 257

-13ox821s "Cleanand Safe' Enough?: PakistanWsMosque Schools There is no , no fan is vey depressigand dreazyandsuca Th.eis o toiet, no weeperin any of th sdzools Whe dwy needa latrine,the girls have to go home duritg school hours, wastinga considuableamount of time. HeadmistressPakistan Primary School In grim termsa Pakistanheadmistress describes' her school.Twenty nine thousandother schoolsare shelterless;sixteen thousand have only one room. Someare so far awaythat childrenmust walk four kilometersto get to them, Worse,67 percentof the teachersare male. Theseare seriousdrawbacks to parents who thinkit is not 'respectable"for their daughtersto attend schoolif there is no female teacher,if they must walkalone, if there are uo boundarywalls or latrines. School conditions,coupled with the overall low status accordedwomen, conspire to limit girls' educationalopportunities. In 1985/86,the enrollmentrate of girlswas onlythirty-two percent, the 'eight lowestrate in the world. To encouragefemale enrollment,the Prime Minister in 1986 introduceda FivePoint Program for Economicand SocialDevelopment which called for the opening of 26,700mosque schools. The Ministryof Educationdescribed the featuresof the mosqueschool in the followingstatement: The'Mosquewill be used as a placeof learningfor children,for out of schoolyouth and for adults. In additionto Islamiyat,the childrenwill study the modern curriculafor primaryschool ... In order to teach modern subjects,a primaryschool teacher willbe -appointedin such school[sic] who in cooperationwith the Pesh Imamwill teach children and adultsat hours convenientto the community.Free booksand teachingaids would [sic]be suppliedto childrengoing to mosqueschools. This will ensure rational utilization of the mosqueand re-establishits traditionalrole of spreadingthe lightof knowledge ;'(Warwick and others 1989). The'Ministrybelieved parents wouldenroll daughtersin institutionsthat affirmedcultural traditions and had longprovided religious instruction to both boysand girls The imam,a respectedreligious 'figure in the communityiwould allayparents fears about sendingdaughters to classestaugbt by strge men. Sheltered,clean facilitieswith fresh water, and within easywalkdng distance woud' provide' respectablebenvironments for girls stillsubject to female seclusion. Since 1985some 26,700mosque schbols providing education up to grade three }ave been opened. ' Opinios regardin ther successare mixed. Whileby June of 1986they had succeededi enrolling '630,0o00puils, only'30 percent were girls, a slightlysmaller percentagethan those enrolledin governmentschools.- A'1989 study of mosqueschools in the Sindprovince revealed an evensmanler percentageof girlsenrolled (26.4 percent). Of mote woncernare questionsabout the qualityof educationprovided. A studyof mosqueschools by the HarvardInstitute lfinternational Development, cites a districtofficial who clainis*Imams are poor teachersbecause they ate illiterate(Warwick and others 1989).' The report goes on to say * Mosque schoolscan be rated high on financialefficiency, cultural acceptability,and quantitativesuccess, low on the capabilityof implementorsand doubtfulon the quality of schloolingprovided (p. 26). 'aCean and safeXmay be enough to encourageparents to enroll daughters they mightnot have enrolledotherwise, but is it enough? Sources: Teacher'sResource Center (1989);World Bank (1988c); Warwick and others(1989); World Bank (1989c). 258 EducatingWomen Recruiting Female Teachers Interviewsand anecdotalevidence in somecountries suggest that increasingthe numberof femaleteachers willboost girls' enrollment. Femaleteachers are in short supply,however, especially in Africancountries. These shortagesarise partlyfrom the requirementsneeded for admissioninto teachertraining programs. Thoughthe r-quirementsare minimal--sometimesas little as an eighth-gradediploma--the majority of womenstill do not pos..essthem. The predominantlyurban location of teacher trainingfacilities also hindersthe attendanceof rural girls. Therefore,women who do becometeachers are more likelyto be urban residentswho are often unwillingto acceptposts in rural areas whereliving and workingconditions are less desirable;where housingand medicalfacilities are inadequate;where good qualityschools are lacking;where food and dothing suppliesmay be limited;where piped water, electricity,and modern householdtechnologies are absent; and wheresingle women may fmd it difficultto meet desirablemates (Ankrah-Dove1982; Seethmaru and Ushadevi1985). To enlargethe pool of femaleteachers, some countrieshave modifiedtheir uniformsalary schedules by providingcompensating differentials in the form of housingsubsidies and free travel to teachers' home towns(Dove 1986; Murnane 1987). Othershave deployed recruiters to rural areas. Neithercompensating differentialsnor local recruitmentalone, however,substantially increased the number of femaleteachers in Pakistanand Somalia. Pakistanattempted to attract femaleteachers by buildingresidences in rural areas where severalyoung womencould live together. The residenceswere unpopular except in Baluchistanwhere they were occupied by married couples. In the other provinces,they remainedunoccupied because socio-cultural attitudes discriminateagainst single women living alone (World Bank 1987a). Somalia attempted to recruitrural girls to teacher training programshoping that they wouldwant to teach in rural areas after their training. Becausethe onlyteacher training institute in the countrywas locatedin the capitalcity of Mogadishu,girls had either to travellong distances daily or to movecloser to the city. Rural parentswere also reluctantto completelyrelease their daughtersfrom domesticresponsibilities and refusedto send them. Therefore, womenwho did attend the trainingprogram were from areas surroundingMogadishu, a city that already had a surplusof teachers(USAID 1989). These strategiesimplicitly assumed a certain degree of mobilitywhich girls and women did not have. Pakistan addressed this mobilityconstraint by introducinga teacher trainingprogram that combined recruitinggirls from rural areas and trainingthem there, closeto their homes. The programbegan in the Punjabprovince in 1984where the governmentintroduced primary teacher training in unitsattached to local secondaryschools. Of the 90 PrimaryTeacher Certificate units started, 80 were exclusivelyfor women. Tlheunits were instrumentalin raisingthe proportionof femaleteachers. In 1985-86,85 percent of the teachersin trainingat the unitswere female (5,040 total), compared to only19 percentat the normalschool and 38 percentat the GovernmentColleges for ElementaryTeachers. In that year alone,the unitstrained 67 percentof all new femaleteachers (World Bank 1987b). Locatingthe trainingclose to home not only weakenedparental oppositoin but, by eliminatingthe need for boardingfacilities, it wasalso lesscostly ban the conventionalprogram.

1/ Other countriesmay discriminatein favorof urban teachers,though. In Cameroonand Pakistan,for example,urban teachersare paid a salarypremium as compensationfor the highercost of livingin a city. 0/ The costper traineein the home-basedprogram, amounted only to about Rs. 14,000(or Rs. 19,000with central office costs) in 1988,compared with Rs. 23,000in the conventionalprogram (Governmentof Pakistan,undated). EducatingWomen 259

Box 83 Recautia in Nepal

Posters,booklets, newspapers nd radio programsdeluged rural villageswith information targeted at sociallyand economicallydisadvantaged families in an effortto motivatethem to takeadvantag of a PrimaryTeacher Traini Prram in Neal. The programaimed to promoteequal educational opportunityfor wome n and men and idkatfitd women as keyagents in increasing access to education. Its majorstrategy was to train roupsof rural, irls as primaryschool teachers. Hostelswere constructedto makeit possiblefor irls fm consivativefamilies in remoteareas to enrollin the program. The programfuncti6oed on two leve'L Atuthe frst level,girls with Secondary School Levig Certificates,were trained at campusesmattacheto Tribbuvan University. Hostel accommodations were"provided,- l aontbly s d, travelexpenses, medical care materials and tutorial assistance.The year-ong program iofeted courses -i ProfessionalEducation, Methods, and General Education.To relate studiestiowomenuFroles ii commnunitylife, supplementary programs in health, nutritionand gardening wre provided.the hostes. Becausethe teachertraining centers were part of TribhuvanUniversity; replicatint was possible without major structural changes. The prqect was expandedfrom its initialsite at the Polchaa Campusin 1971,to Dhankutain 1973,Nepaljung in 1976,and Jumlain 1978.: At the secondlevel, gSis wiho ha not attainedmiore -than tenth-gradeeducation were lodged at feederhostels and senattoarb-y secondaryschoo vhcre theycould acquire the skillstbat would qualifythem for enryJito the teacher traningprigram. They too were providedwith monthly stipends,travel e micbneits and tutoral assistance. Undera directivewfromtheMinistry ofducaon. lDistrictPducation Officers were instructed to give, priority to thfewomena woha3d gr'auat f:rom'the.teacher trainin program in assignments-to teachingposts in priary,shool Followup o . ps at campussites provided monitoring and "Eresher' coursesf -ast d,... L Betl*een17 and -ttr d: - reshare of femaleteachers increased frio 3 percentin 1972to.8: prcenn 9r v perent of the newteachers employed cowi.n fromthe four smal a .Intesamepriod, ryscho ermenticreased from 16.8percent to 28 pec. Thei--- bnt o recruitment,dua enty levels,subsidiion, and.home teathing p tpsp i ten iniadding females to Nepal'steaching force. Soture: UNDP (1 82).- 260 Educating Women

Although local recruitment and local training appear to be effective at increasing the supply of female teachers in rural areas, lowering minimum educational qualifications, actively recruiting girls from rural areas, subsidizingtheir secondary and teacher education, as well as providinggirls with the option of being posted in a school near home, are features of a coherent strategy. Nepal successfullyimplemented such a program for rural female teachers (See Box 8.3). A similar program was introduced in the Yemen Arab Republic in 1987 where three teacher training centers were set up as temporary sections of existing postprimary schools to provide teacher training to rural girls with a sixth grade education. In this program, buses transport participants to and from each center, and each trainee receivesa stipend. Upon completion of the three year course, the new teachers are deployedin local schools and the program moves to another area. The first course was attended by 80 rural girls and only two had dropped out after the first two years of classroom instruction.

Lowering Direct Coss

Location and propriety are not the only things parents consider when deciding if they should school their daughters. The costs of schooling also influence parents' schooling decisions. Although public schooling often implies free or subsidized tuition, parents still incur the costs of transportation, uniforms, books and school supplies, and schools may also request cash or in-kind donations. These expenses can be prohibitive to poor parents, especiallyat the secondarylevel where tuition is often high. Bangladeshand Pakistan have responded to these cost constraints by lowering the cost of uniforms; Bangladesh and Guatemala have responded by introducing scholarshipprograms for girls.

Lowering the Cost of Uniforms

Providingfree uniforms did not substantiallyraise girls' enrollment in BI7gladesh, however. Beginning in 1981, uniforms were to be distributed to 500,000 girls aged 6 to 10. The intended recipients were daughters of landless agricultural workers, fishermen, and other low-incomegroups. As it turned out, only 150,000to 200,000 girls received school uniforms. Anecdotal evidence suggests that girls' enrollment increased slightly,but the scheme was discontinuedafter only two years because manufacturers were unable to meet quality standards, principals distributed uniforms to pupils who were ineligible, and wealthier parents withdrewtheir daughters from school because they had been excluded from the program (UNESCO 1989;World Bank 1980a, 1980b).

In the Sind region of Pakistan,where school uniforms are compulsory,a different approach to the problem is planned. Instead of providing free uniforms, the region will experiment with abolishing the required uniforms in rural areas (World Bank 1990c). This strategy is less costly and more easily administered, but whether the experiment will reduce the direct costs enough to raise girls' enrollment remains to be seen.

Provi&g SchowhiId

Bangladesh has had greater success with its scholarshipprogram for secondary schoolgirls (classes6-8) than it had with its free uniform program. The scholarshipprogram was established in 1982by the Bangladesh Association for CommunityEducation, a local non-profit organization. By the beginningof the 1988 school year, 20,085girls had benefitted from the program, and the benefits were multiple (See Box 8.4). Due to the encouraging results of the project, the government announced in early 1990 that it would consider

O/ The final year is spent on supervisedteaching.

2/ This represents 10 percent of all girls in this age group. Only 54 percent of school-age girls were enrolled at that time. The project was funded by the World Bank, UNICEF, and bilateral donor agencies. Educating Women 261

waiving all fees for girls during the first three years of secondary school and would consider providing scholarships to girls in the higher grades.

Assessments of Guawaiala' scholarship program for primary school girls are equally encouraging. This program, modeled after the Bangladeshscholarship program, was piloted in the Indian antiplano where only 53 percent of school-agedgirls attend primary school and only 17 percent complete the cycle. ILbegan in 1987 with one village and 50 scholarshipgirls, and later expanded to 12 villages. By 1988, the families of 600 girls between the ages of 7 and 15, most of whom were enrolled in grades three, four and five, had received a payment of 15 quetzales (US$4) a month for ,each daughter who did not become pregnant and who attended classes at least 75 percent of the time.0' Since parents paid no tuition and books were provided free in schools, the monthly scholarshippayment partially compensated parents for other scLo01- related expenses and for the loss of their daughter's time. The project was so successfulin retaining girls in school that over 90 percent of the scholarship girls completed the year (Ministry of Education, Guatemala, 1989). The government now plans to fund 550 new scholarshipsfor 11 additional communities. Government-sponsoredscholarship programs were also implemented;n India and Nepal, but little is known about their impact (UNESCO 1986). In the early 1980s,Nepal pilot tested a scholarship program in two of the poorest rural areas. For three years, all girls who attended primary school received a small sum of five rupees per month. Although its importance to the cash-poorparents was never evaluated, data on girls' enrollment suggest that the program did not increase the proportion of girls attending school. However, those who came did stay in school longer. Whether this was the only positive outcome of the program is unknown;nevertheless, the government is willing to invest additional resources in a scholarshipprogram. It recently budgeted a program whereby fivepercent of all girls in everydisuict would receive a scholarship (Butterworth 1989).

The experiencesof Bangladeshand Guatemala with scholarshipprograms are encouraging,however the use of scholarship programs as a strategy to boost girls' enrollment raises several questions that only country specific research and implementation experience can answer. For example, is a scholarship program financiallysustainable for the period of time or coverage necessary for it to have an impact? Can a country afford to support a growing number of female students even if the cost per student is low? Who should receive scholarships? What is the appropriate amount of the award--shouldit cover only tuition or should it include also other school-related expenses? For how many years should they receive scholarships? Should awards be given more selectivelyover time? What is the most cost-effectivetargeting approach? Answers to these questions require an understanding of the level of demand for girls' education and the resources households and government are willingto allocate for it.

/ The project is administered by a local NGO, Guatemalteca de Educacion Sexual (AGES). A village woman "promoter"verifies attendance prior to paying parents each month. 262 EducatingWomen

..Box 4 My Daughter Won't Need a Dowry- - weBangladeSh Female Senday Educatio Scholhbip Proam AnjumanAs r acholarship to attend Chitoib High school. She was thirteen yeus old. She has ine brothersad siters. Her father completedprimary school and is a rice famer, her- mother is illiterate, Her friend,Majeda, was alreadymaried and exting a baby. rural poor Sharat upazila,Comilla District Bangladesh, Majeda was the rule rather than the exception.She, unlh her brothers,was an economicburden to her familyat fourteen,she wasmarried and shipped to her in-laws. Give parental attitudes about daughter it is not surprisingthat the 1981cens reported the averageage of ma 'gtobe 16. years for female as comparedto 23.9 yearsfor males and that only'4.7percent of femiles had some secondaryeducation as comparedto 10.6percent fot males.. This.information is, even less suprising m a countrywhere the primaryreason for not sending cidren to schoolis poverty,where fe.wer than two percentof secondaryschools are government operated and where,in addtion to tuition,parents must bear the costs of traosportation,books, statioy supplies;uiorms, admissionfees, exam fees, poor fundand snacks. No wonderMajeda!s familywas anxiousto Shipher off to in-lawswhere she could fulfillher role as obedientwife and fertile mother. Majeda can be e8pectedto bear sevenchildren, contributing to an alarmingthree percentannual rate of populationgrowth. In an effort to curb that growth,the BangladeshAssociation for CommunityEducation (BACE), piloted the female SecondaryEducation ScholarshipProject in Januar of 1982. Sparked by populationliterature documenting the positiveeffect of secondaryeducation on decreasingfertility, the project'saim was to encouragegirls to enter and continueseconday school,thereby delaying marriageand increasingcontraceptive use. Initiallyall femalestudents in gradessix through ten who livedin the Sharastiupazila were eligible for the scholarshipswhich reduced tuiton by half. In Januaryof 1985the projectwas extendedto the Gopalganjupazila and, in the same year, a selectioncriteria was introducedthat limited scholarshipsto girlsfrom families who earnedless than 1,200taka ($47.00)a month. The selection criteria provedto be problematicinsofar as fieldworkers had difficultyfinding a sufficientnumber of girls who came from familiesearning less than the cutoff income;staff was burdened with reviewingand verifyingapplications; and mid-levelbut influentialmembers of the community,whose daughtershad previously qualified for scholarships,were antagonized.Notwithstanding the negative impactof the selectioncriteria, the overallsuccess of the project resultedin its expansionto four additionalupazilas. hX y.eptember of 1988,20,085women had benefitedfrom the program.Anjuman Ara was one of them. As a result of her expe.rience,Anjulman says she won'tget marriedbefore twenty. She'l haveonly two childrenand slle'Uuse birth controlpills. Anjumanand other scholarshipgitls have increased the oercentageof femaleenrollment in projectarea secondaryschools from 27.3percent before the projectbegan, a figuresimilar to the nationalaverage, to 43.5percent in 1987,more than doublethe nationalaverage. The secondaryschool female drop-out rate in the projectarea alsodeclined. from 14.7percent before the projectbegn to 3.5 percentin 1987. Anjumanand her friendsencourage their sistersto attend schooland so, primaryschool enrollment is up as welL Anjumanattended schoolregularly in accordancewith scholarship provisions mandating seventy-five percent attendance. Shedid well in English,Bengali, Religion, Social Studies and Science. She also learnedabout banking and savingsthrough ma the scholarshipaccount the projectset up in her name;twice a month, accompaniedby a teacher,Anjuman could go to the localbank and withdrawher scholarshipmoney. Anjuman,in fact,wants to be a banker. Anjuman'sadditional and timelytuition payments improved the status of the schooL It nowhas a toilet and moneyfor equipmentand supplies. Knowingthey will receiveregular salaries,more qualifiedteachers have cometo the highschool, stabilizing a once transientstaff. EducatingWomen 263

And how does Aujumaniifamiiy fel at ll this? `Winessher mother'srespouse. I couldnot gv duc.ati my elder dauhter who wasmared wbilein: class3. I had to. -ao 'Thanks to th Nolaship I receivedman- offers of marriage d coud: pickand oe.ii demndedd-dry. .Someevenwanted to e MYexpee I couldmany mydaughtr to a householdwhich is socy superiorto min. Shev1$its me: more oe than - myelder daughter nd bring me preseuts. Evidentl myyoun daiughter is happier than my vlder daughter to whom I coud jnot give poper education.

At a costof $44.43a yeart retrasfrmd AumAr Xiomfaily i to famr asset..

SourceS: Ather (1984),Maruin (198$), Thein (1988).

Lowering Opportunity Costs

The resourcesthat householdsare willingto allocatefor girls'education include not onlycash outlaysbut also the girls' time. Womenin developingcountries spend large amountsof time performinghousehold chores. Girls often share this workwith their mothers;they care for siblings,prepare meals,carry water and firewood,or eam an incomefrom outside jobs. Therefore, it maybe necessaryto lowerthe opportunity cost of schoolingto increasegirls' participation.There are at least four waysto lowerthis cost. First, scholarshipprograms can ease the barrier high opportunitycosts create by offeringmonetary compensationto parents for the loss of their daughters' time. Second,allowing girls to bring younger siblingsto school,establishi , dvycare centersnear schoolbuildings, or introducingsimple technologies can lowerthe amountof timegirls spend at work. Third, the formalschool schedule and instructionaltime can be made more flexibleand consistentwith girls' work schedules. Fourth,alternative or "safetynet" .chools can provide schoolingopportunities at times duringthe day that girls can attend. These four approachesare far from exhaustive,but each has been implementedwith varying degrees of success.

EstablMhingDay-Cf entwe

Colombiaand China are activelyexpanding day-care in communities,at schoolsand at worksites. In Colombia,where one-fifthof the pooresthouseho!ds are headedby singlemothers, and 44 percentof poor childrenaged 7-11 do not attendschool, the communityday-care program -- Hogaresde BienestarInfantil - - has freed manygirls and womento attendschool or join the workforce.The community-basedcenters werefirst establishedin 1987. They are locatedin poor neighborhoodsand serve about 400,000children under the age of seven. At the centers,children are providedwith 60 percent of their dailynutritional requirementsand are supervisedby "community"mothers, selected by women in the communty. The communitymothers are trainedin nutrition,health, hygiene, and recreation,are paid a salaryand receive assistancein obtaininghome improvementloans. By 1992,Hogares de BienestarInfantill hopes to reach 1.5 million children. 264 Educating Women

China, in its efforts to increasefemale enrollment, has established day-care centers at schools and worksites. In Ghansu province, for example, girls in 20 to 30 schools are allowed to bring their younger siblings to class, and worksite day-care centers for employed mothers have improved girls' enrollment in urban areas (Coletta and Sutton 1989). China has a!so expanded preschoolswhich not only relieve girls from child care during the day but also provide an educational advantage for the younger siblings(Lockheed and Verspoor 1990).

Modifng H;me Twcnlgie

In many cases, simple improvements in home production technology can rescue hours of time, time that could be spent in school. In the hills of Nepal, for examj o, where deforestationhas resulted in scarce wood supply,women walk for miles, sometimes focwhole days, into the forest searchingfor leaves,branches, and twigs for heating and cooking. One hill woman of the Magar ethnic group says:

Once it wasn't difficultto find wood on the ground. But now there is not even enough left over to fill one headload [35 pounds], unless you walk for miles and miles... As it is now, I must bring my daughter to help collectfuel and fodder, so she often skips school to help me... If fuel gets even more scarce, I will have to take my daughter out of school completely so she can help me with my other tasks... The better woods are getting much scarcer. I must collect other species that burn poorly... Some woods make my eyes burn and give the rice a bitter taste my husband can hardly stand (Molnar 1987, p.4).

So, the government of Nepal disseminatedfuel efficient, smokeless,wood-burnin3 stoves to 15,000Nepali families as part of a 1977 Forest Act, designed to check deforestation, improve the forest cover, and increase the amount of fuel, fodder and timber available for subsistence. Forest department staff trained female "stove promoterse who, in turn, taught village women how to use the stoves, and how to conserve fuel. A study of families using the stove found an average wood savingsof 28 percent, reducing the haul by about 2,000 pounds of wood per female a year (Molnar 1987).

The introduction of labor-saving technologies may be sound a policy in some settings, but it may not guarantee greater school attendance. Even when the technologies are affordable and appropriate, i.e. simple, adaptable to the conditions of the community, and easily maintained, factors other than time may prevent schooling. In Burkina Faso, for example, a nonformal education program for women, initiated in 1967, introduced mechanical grain mills, accessible water wells, and carts for hauling wood in villages located in three geogiaphicalzones (McSweeneyand Freedman 1980). The labor-savingtechnologies did not boost girls' enrollrmentin the project areas, probablybecause the schools were still remote. They did, however, lower the amount of time women needed for certain chores, but instead of attending literacy classes with tbe time saved, women used the time for other work -- like preparing better meals or weaving.

Adoptg FRble School Scheduls

Programmed instruction or programmed learning is often advocated as a way to make formal programs more responsiveto children's work schedules, to equalize the quality of instruction across children, and to lower the penalty for absenteeism. It is also a low-costway of expanding school places because it facilitates multigrade teaching. With programmed instruction, the curriculum is organized in sequential units so that students can learn at their own pace. EducatingWomen 265 ThePhilirpines, Indonesia and Bangladesh(Project Impact), and Liberia (Improved Efficiency of Learning), Colombiaand El Salvador(Escuela Nueva) have experimentedwith programmedlearning. Despite the rational underpinningsof these programs,their outcomessuggest that programmedlearning may be disadvantageousfor girlswhen learning materials require more individual attention from teachers and when additionalhomework time is required. Bangladeshand Liberiaillustrate the problem. In 1980,Bangladesh introduced pgrpgrammed instruction under Project Impactin 18 rural schools,with plans to extend it to 300 schools.yIIn Liberia,multigrade teaching and learningwas introducedin five schoolsin 1979and was expandedto 15 schoolsin 1984under the ImprovedEfficiency of Learning(IEL) project. Both programsintroduced self-taught learning materials organized into modules;Project-Impact reliedon teachersand studentsin grades3 to 5 to guide youngerstudents through their lessons.AJ The outcomes of the two projects were generallyunfavorable. In Bangladesh,the program neither increasedenrollment nor reduced dropout rates. Despitegovernment efforts to elicit support for the programthrough home visitsby districteducation officers, parents were not convincedthat olderstudents could properlydirect the learningof their children. The project did not make learningless teacher- dependentas planned becausethe instructionalmodules, though intended to be self-learning,were not. They requiredthe supervisionof a teacher or literateparent; both were in short supply. The resultsin Liberiawere equally discouraging. An evaluationof the projectcompared gender differences amongthird, fourth, and fifthgrade studentsin Math and Englishachievement in the IEL schoolsto those found amongstudents attending conventional schools and those attendingschools that participatedin a textbookdistribution project. The studyfound that, althoughIEL studentsscored higher than either of the comparisongroups,iL gender differenceswere greatestin the IEL schoolsand least in the scLoolsthat receivedtextbooks. A possibleexplanation for this is: The improvedopportunity to learn is paralleledby increaseddemands on students,particularly in terms of studytime to completeassignments. In Liberia,as in manydeveloping countries, numerousdemands are placedon daughters.These expectations may have limited their ability to take full advantageof the enrichedlearning environment (Boothroyd and Chapman1987).

Colombia'sescuelas nuevas seem to have been more successful,perhaps because they are more accommodatingof children'swork schedules,or perhapsthey are less teacherdependent and more easily allowself-instruction, thereby penalizing students less for absenteeism.It may also be that programmed instructionis more effectivewhen parentsthemselves are literate-- adult literacyin Colombiais sixtimes higher than in Bangladeshand 2.5 times higher than in Liberia. Consideringthe reported successof Colombia'sescuelas nuevas, investigating the conditionsunder which programmed instruction benefits girls is an area that merits further research(See Box8.5).

2/ The expansionwas assistedunder the WorldBank's Second Primary Education Project (World Bank 1985a). J In the IEL project,teachers instruct students in groupsof 15 in gradesone to three. Childrenthen break into groups of five for peer practice sessionsto ensure content mastery. In grades four to six, learningtakes place primarilyin peer groupsof 5 to 8 childrenand, at times,there are opportunitiesfor independentself-learning. IV Theyscored 13.5percent higher in math, and 5.5 rorcent higherin English. 266 EducatingWomen

Bo 85 WorkTodaydi^y owCobi'ts EscuelaNueva

a remotvillap aoungirwetsdowneu4flo..of.t her but. Her parents are in the fields :ax.istin gain. A babyi girltd op r pai lifts the infantfrom hiismat. Thisyoung .i l: like so many other" iddv i ct make it to schooltoday. The school --scedue in htt ate- idiWlto hewfailys survival.When she does got tostl, e maybD fAibhid he-rcasmtes. Facedwith r8tention, she is lkelyto drop-out, Forirls whodo.nt . eite leisureto spendsio huot inischool every day, Colombia's Escuela Nuevaprogram offirs tlaopo CWscaived as a wvajtoprovid a fullfive yprimary educationto rural -areaswhere sa r e and ld resoUrces ted conventionalone teacher/onegrade a-o s, te program :s ofi launched 976;by July of 1989,it wasoperating in 15,000 : ral bchobl.Notlyb thesEla Nuevaincreased the relevanceof primaryeducation in poor :ruralcommunti,it hasb rovedstudent achievement, enhanced girls' self-esteem and reduced drop-outd repetor ra Basedoii te inl ofmli-rade teachingand flexiblepromotion, the EscuelaNueva : presupposesstudent ibsences drn periodsof agictural activities.It allowsstudents to resume hstdies aftie-suchabsn csthroug)t the use of semi-programmedmaterials arranged in -quetioal in itT ma leani relevant, the curiculm contentis ruraloriented and eadilyadated to the ir stancesof a particlarcommunity. It promotesproblem solving rather :thanirote ke g as s s applywbat they study to the homeand the community.'Resource corns aow student to worklone or in smallgroups on assignmentsappropriate to their grade -andstae bf pr A udred bookbbrary provides resource material to encouragea project apprbi.chto learning.With the assistanceof popularlyelected student 1leaders,; the teacheris able to handleup to fiv grad_csmeoly. Andthe studentis ableto care for her brothersand sistes oneday nd learnbetter ways of doingit the next. Sowtes:World Ban-*' .8d), (1982).

Buiig Safey Nets Nonfornalprograms offered before or at the end of the workingday providean alternativefor children whuare unableto attendday school. Although morning or eveningschool may not be an idealapproach to raisegirls' levels of education,it is one strategythat has workedwell in Indiaand Bangladesh. India'snonformal evening education program aimed to bringschool dropouts back into the primary educationmainstream and to givethose who had neverattended a chanceto enroll. Eveningclasses for youthsaged 9 to 14were staffed by teachers drawn from and trainedin the localcommunity. Two years into the program,1,431 students were enrolled and 1,040were girls. The convenienttime schedule, communitycontributions to the schoolcenters, and recruitmentand trainingof localteachers were key ingredientsto the successof this program(See Box 8.6). EducatingWomen 267

lO -8.6Night S3ofor NineYa Olds: NonformalEducation in Mah

In 1978,only 63.5 of Indi'sgs agedsir to eleven,were in shool comparedto 97.3percent of the boys. The proportionof Irlsin the upper ary grades was less than half that of boys (245% gls; 48.7%boys). Mostof the hildreneither did notenter Shool or droppedout soonafter entering. They camefrom poor familieswhere they shoulderedburdensome household chores or workedoutsidethe bome to add to the familyincome. They did not have time for a sk hour day schooLA further problemwas Indiaessingl-point school entry system, whereby a childcould enter school oiyat Grade 1,evenif s/he was eight or nine years o. Com to e young .lot, older childrenfltnoortable and left. The isistence on single-pointentry and fill time attendance has been '`ig hurdle in unlversaliing primary education for children from poor rural families,who coustituteabout 70 percentof the age groupconcerned. Of the, two-thirdsto three fourths areegirls. A`nonormal educationprojet, desind by The Indianitute of Education, discovred that when as- were scheduledat night, after choreswere mpleted,preadolescent girs woud go to school,even if it meant challengingghosts and man-eatingtigers The project,supported by centraland state governments,and the localauthority of the Pune t)istric began in Novemberof 17 igty-six carooms, providedby the communities,were established in seventeenvillages. Enrollt was restrictedto illiteratechildren from tine to fourteenyears of *age. Classes, limitedt twentystudents, were ungradedand based on the principleof mastery learning.They were scheduled300 days a yearfrom 7:00P.M9.00 P.M. To build inkageswith the community,teachers were recruitedfrom the villagesor hamletswhere the classeswere.located. They-were paid a smallstipend of 2,000rupees ($250)a year. Sixty-fourmen and twety-four women,most from the farmingcaste, with somesecondary education, were trainedby membersof the projectteam. The men teacberswere caled *BhauM (elder brother); womenteachers were *TaiP (elder sister). The nomenclaturewas decidedupon by teachers,community leaders and project officersbecause it underscoredthe responsibilityof the educatedrural youthto passdown education to the next generation. Wuringthe weeklong trainingsessions, emphasis was placedon preparing and usingteacher/learning materials that would be relevantto rural life, and also encouragepeer- group teaching. Annualexams were replaced by an evaluation"fairW called Dal Jatra. Teachersand childrenfrom all proJectcasses met in a centralvillage, where, for half the day,they playedgames, sang, danced and listenedto stories;the latter half of the day they read, calculatednumbers, told stories,and answereda quiz. In Jauauy of 1981,among the 1,431enroUed students, 1,040 were girls. A year later, 196 girlsand 86 boyshad droppedout of the project,a ratio far lowerthan the usual 50 to 70 percentdrop out rate in Class1. Girls attendancewas muchhigher than expected.To understandwhy girls attended regularly,project staff interviewedthe girls and their parents. Both saidthat attendingschool would help the girlslearn to keep accounts,to writeand read letters,to managedaily transactions; learning wouidmak them wise."Parents added that the girlscould marry an educatedyoung man. Parents and daughtersagreed tbat the most convenienttime for classwas in the eveningafter the girls had finishedtheir days worL Source: Naik (1982).

A similar program was launchedin Bangladeshin 1983. There, the BangladeshRural Advancement Committeeestablished 23 pilotschools in selectedrural areas that offeredthe first three years of primary education. Schoolsessions ran two and a half hoursa day,271 daysa year. Parents,on the basis of their workschedules, decided whether the schoolin their communitywould be open in the earlymorning or late 268 EducatingWomen afternoon. Sixty-sevenpercent of the teacherswere female. The demand for this programwas so great that parents tried to enroll "tinytots whocould barely hop on one foot' (Mallon1989). So great was the enthusiasmthat by 1988,731 centerswere operatingand 21,903child-zn were enrolled. So successfulwas the programin attractinggirls that they comprised63 percent of the enrollment,less than one percent droppedout, and 83 percentwent on to contlauetheir educationat governmentprimary schools (Begum and others 1988).

Raising the Benefits The cost of schoolingis one factor that parentsconsider when makingtheir schoolingdecisions, and most past effortsto raise femaleenrollment have focused on loweringschooling costs. Strategiesthat raise the benefits to females'education can be used as alternativesor complementsto cost reductionstrategies, howeverfew past efforts have targeted this objective. Yet, if parents do not perceivethe productivity .._nhancingeffects of educationon non-marketwork, in the eyesof manyparents, girls may benefit less from educationthan boys. And, girls may in fact gain less from their educationif they achieveless than boys when in school,and if they studyin areas that do not facilitateemployment or earn them a reasonable income. The lowerbenefits that derivefrom theseconditions have led to recommendationsthat education and informationcampaigns should be used to advertisethe benefitsof education,and that educationcould yield larger benefits if schoolsprovided meals and gender-freetextbooks, and if girls and womenwere trained for occupationsthat facilitatetheir employmentand earn them a reasonableincome.

Eduatin Me Commuit Fromthe parent's perspective,girls maybenefit less from educationthan boys. Where girlsare expected to becomemothers and wivesand work on the familyfarm, time spent in schoolis time taken awayfrom work and from learningskills considered relevant to these roles. In Somalia,for example,mothers are committedto trainingtheir daughtersto performall householdand domesticactivities at an early age (USAID 1989). Schoolswill not teach them how to dry dung sticksfor fuel or grind spicesfor cooking. Similarly,parents inl villages of Maliare not convincedthat schoolingwill help girlsbecome better farmers. The fewparents who sendtheir daughtersto schooldo so becauseschooling is obligatory,not becausethey perceiveany life advantagesfor their daughters(World Bank 1983b). Educationand informationcampaigns have been successfulin raisingthe demandfor healthand family planningservices in developingcountries,U 2 but theyare rarely used to promotegirls' education.Mali and Moroccoare exceptions. In Mal, media campaignsadvertise the value of educationas an investment (WorldBank 1989d).In Morocco,materials are beingdeveloped to promotegirls' education;they will be distributedby extensionworkers who visit rural communitiesto encouragelocal participationin the constructionand maintenanceof primaryschools (World Bank 1989f).

Pmvidftg School Feedi Prnms Even if parents were awareof the many non-marketreturns to girls' education,girls may gain less from educationthan boys. In some countries,females are more likelythan males to be undernourishedor malaourished.In the Matlabthana of rural Bangladesh,for example,malnutrition rates are substantially higheramong femalechildren and mortalityrates for girlsexceed boys' rates by an appalling50 percent

W In Egypt,for example,through a campaignlaunched by the Ministryof Health in 1984,television, radio and newspapershave played a decisiverole in boostingpublic understandingof, and demand for, oral rehydrationsalts and vaccinations. Today, knowledge of oral rehydrationtherapy is almostuniversal among mothers,and over 80 percent of Egypt'syoung children are immunizedagainst the six main childhood diseases(UNICEF 1990). EducatingWomen 269

(D'Souzaand Chen 1980 ).P/ Girls are also more likelyto be malnour!hedthan boysin Pakistan,India and Guatemala,and a studyof 94 LatinAmerican countries showed that girlsaged 0 to 4 met a significantly lowerpercentage of their age/weightmeasurements than boys of the same age (Schofield1979, Safifios- Rothschild1979).H/ Iodine and iron deficienciesare also more prevalentamong women. A reviewof gender differencesin iodine deficiencyshowed that in 50 of 53 paired gendermeasurements from 17 countriesthe incidenceof goiter was greater among females(Simon and others 1990),and at least one studyshowed a stronger associationbetween goiter and 10 scores in girls than in boys (Levinand others, in press). Data on hemoglobinconcentration also indicatedthat at least half of all women,and sometimesthe entire female population,is anemic(Levin 1986). The evidenceis clear that such nutritionaldeficiencies place childrenat risk in school. Malnourished children are less active,less attentive,less motivated,and less responsivethan their better nourished counterparts.They perform significantly lower on assessmentsof achievement,10, psychomotorskills, and social-personalbehavior. They are absent from schooland repeat grades more often. Hungryand iron deficientchildren have shorter attention spans; iodine deficient children are slowerat _rocessing information and suffer from impairedvisual-perceptual and motor coordination(Pollitt 1 990).7 Giventhis evidence, schoolfeeding programs (SFPs) are often advocatedas an means to reduce abserteeism,and improve children'sability to benefitfrom instruction by removinghunger or nutritionaldeficiencies. They are also often suggestedas an incentiveto raise girls' enrollmentand attendanceby offsettingsome of the costs of attendingschool. Despitethe rat;onal underpinningsof SFPs,most evaluationsin both developedand developingcountries have been unsuccessfulat establishinga relationshipbetween school meals and school enrollment, attendance,retention, or achievement.1/A reviewof SFPsis beyondthe scopeof this paper, but reviews by Levinger(1984), Pollitt (1990)and Halpern and Meyers (1985)point to two major reasons for the inabilityof past -esearchto establishfirm conclusionsabout the effectivenessof SFPs. First, most evaluationsof SFPs are methodologicallyflawed, having been conductedon pre-existingnon- experimentalprograms which precludes adequate control of confoundingfactors. Second,students' health and nutritionalstatus, the program'sdesign, and the social,economic and schoolenvironments all interact in producingthe observedoutcomes. A stimulatingenvironment can compensatefor the effectsof hunger and malnutrition,and greater intellectualdevelopment can be achieved when diet as well as the psychologicaland socialenvironments are enriched(Levinger 1984). An unstimi:latingschool environment may also negate the educationalbenefits of SFPs. If so, school meals will accomplishlittle unless accompaniedby improvementsin the qualityof education(Halpern and Meyers1985). The designof the

/3 Mortalityrates are for childrenaged one to four.In-depth dietary surveys also show that malesconsume more caloriesand protein than femalesof all ages even when nutrient requirementsthat varyby body weight,pregnancy, lactation, and physicalactivity are considered(Chen and others 1980). / In contrast,girls do not appear to be at a disadvantagevis-a-vis males in anthropometricstatus in the majorityof Sub-SaharanAfrica (Svedberg1990).

1/ Also, severeiodine deficiencyis often accompaniedby cretinismand deaf mutism;chronic vitamin A deficiencyimpairs vision,often causingblindness, preventing afficted childrenfrom attendingformal schools. M/ See Powelland others(1983); Halpern and Meyers(1985); Halpern (1986); Levinger (1983, 1984); and Pollitt(1990). One exceptionis a schoolbreakfast program in Jamaicdwhere 115 undernourished school childrenwith a mean age.of 12.5 yearswere given a governmentschool meal at the beginningof the day. Breakfesthad no effecton the chldren's weight,height, or reading and spellingscores but it improved arithmeticscores and attendance. 270 EducatingWomen SFP may also mitigatethe benefits of school meals and weaken their potential to influenceparents' schoolingdecisions if SFPsare inappropriatelydesigned for the community.Levinger, for example,notes that in po -r communitieswhere the opportunitycosts of schoolingare highand the benefitsunclear, SFPs may be more effectivein raisingenrollment when food rations are large enoughto be viewedby parents as a significantincome transfer. Conversely,in communitieswhere opportunitycosts are not high and where educationhas a clear economicbenefit, SFPs are likelyto have Uttleimpact on attendanceor enrollment. Complimentarynutrition education,and school-basedmicronutrient supplementation and dewc,mingprograms can alsoboost the effectivenessof SPFs (WorldBank 1991). Althoughresearch has not successfullyestablished a relationshipbetween SFPs and schooling,enough is knownabout the relationshipbetween the childs'health and nutritionalstatus and his/her attendancein school and ability to learn to warrant appropriatelydesigned experimentalprograms to boost girls' attendance. To be effective,however, SFPs shouldbe designedwithin a broader interventionto address schooland environmentalfactors that alsocontribute to learningdeficiencies. Moreover, evidence suggests that nutritionalintervention may be more effectivein promotingmental development when given to pre- schoolchildren. A two-yearsupplementation program in Kingston,Jamaica for 129 childrenaged 9-24 monthsresulted in significantlyimproved locomotor skills among stunted children (Grantham-McGregor and others 1991).

Trdanig for Non-traditonal Occupaions Yet, evenif the healthand nutritionalneeds of all childrenwere met, girlsare stilllikely to gain less from their educationthan boys. In the labor market,females face wageand employmentdiscrimination which conspireto limitthe economicbenefit that they can expectto receivefrom their education. Girls' career choicesalso lead them towardeducational programs that are unlikelyto equipthem to substantiallyincrease their earnings.To facilitatetheir accessto well-paidoccupations, earmarking secondary and postsecondary scholarshipsfor girls in areas that prepare them for occupationsin growthsectors of the economyis a strategy that has been successfulin many industrializedcountries. A second option is to provide occupational/vocationaltraining directly linked to employmentwith a strong recruitmentand guidance component. Past experiencessuggest that vocationaltraining programs without these characteristicsare unlikelyto attract females. The Post-PrimaryTechnical Schools Programme establishedin 1975 in Dar es Salaam typifies an unsuccessfuloccupational training program. Spurredby concernwith an increasingnumber of dependent unemployedprimary school girls, the governmentof Tanzaniaestablished twelve training centers on the premisesof existingprimary schools. For girls, the centers offered home economicssubjects such as cooking,housecraft, needlework, child care and laundry.The programdid not guaranteeemployment, and as a result, the centerswere greatlyunderutilized, enroling only37 girls comparedwith their capacityto enroll 240 (U.N. EconomicCommission for Africa 1984). In Chile,the absenceof a recruitmentcampaign and guidanceundermined an otherwisesound six-year experimentalpilot vocationalprogram to train middle-leveltechnicians. The pilotprogram, which began in 1968,was introducedinto two schoolsin Santiago-- La Cisterna,a girls'technical school and La Renca, a mechanicstraining school for boys. Both schoolswere made co-educationaland course offeringswere broadened. At La Cisterna,chemistry, computer programming and data processing,textiles, and bilingual secretarialtraining were addedto the curriculum;chemistry and electronicswere addedat La Renca. Close cooperationbetween pilot schools and industrywas alsoestablished for curriculumdevelopment, practic trainingand job placement. The results of the pilot were decidedlymediocre. Althoughthe programsdid expand girls' access to training,boys' enrollmentin all programsincreased faster than girls' enrollment. By 1973,the chemistry programat la Renca enrolled88 girlsand 214boys, compared with 47 girlsand 37 boysin 1969. Over the same period,boys' enrolment in electronicsgrew from 87 to 499,while girls' enrolment increasedfrom 9 to only42. Similarpatterns were observed at La Cisternawhere the majorityof girlscontinued to enroll EducatingWomen 271 in secretarialcourses. Project evaluatorsattributed the low femaleenrollment to the lack of guidance counselingin primaryschools and the absenceof recruitmentefforts (Ferreri 1974,Pilain 1975). In contrastto thisprogram, recruitment and counselingstrategies were key ingredients to the successof the Industrialand Commercialtraining program in Morocco,which encouraged women to competewith men for admissioninto trainingprograms which they had been reluctantto attend previously(See Box8.7).

EAwing Gender-Neu&MiInstrucion

It is often arguedthat girls' decisionsto enter low-peA,traditionally female occupations are reinforcedby teacherswho maintainstereotypical notions of girls' poor performancein science,math and other "nontraditional' areas and by textbooksthat depictwomen in low-mobility,low-paid occupations (Finn and others 1979,Whyte 1984). These stereotypes,the argumentgoes, dampengirls' aspirationsand, therefore, discouragetheir attendanceand achievement.Such observations have led to the conclusionthat educational programswould yield largerbenefits for girlsif teacherswere made aware of the stereotypesthey hold, if they modifiedtheir interactionsaccordingly, and if gender-neutraltextbooks were developed. There is no empiricalevidence from developingcountries to supportor refute the hypothesisthat teachers' interactionswith female studentsand gender bias in textbookcontent discouragegirls' attendanceor achievement,and we are unawareof effortsto sensitizeteachers to longheld stereotypesor to modifytheir behaviorin the classroom. This notwithstanding,several countries have initiatedlarge-scale projects to make textbooksgender-neutral and to broaden the roles in which women and girls are depicted. Bangladesh,for example,is revampingtextbooks to showwomen in roles other than the traditionalroles of mother and housewife.Kenya (Herz 1989),China, and India havealso implementedsimilar programs, but no informationon their impactis available(UNESCO 1985c). If the argument is correct, such textbookchanges should make a difference. But relative to other investmentoptions, the cost-effectivenessof revampingtextbooks to raise femaleeducational attainment and achievementis questionable,particularly in countrieswhere females status is low and wheretextbooks of anykind are in short supply.As countriescontinue to developcurricula and instructionalmaterials more sensitiveto their circumstances,it isreasonable to expectthat theywill simultaneously adjust their portrayals of girls and women. But in the short run, simplyimproving the qualityof educationfor all childrenby puttingbooks into the hands of childrenand teachingmaterials into the hands of teachersmay accomplish muchmore. Abundantresearch evidence shows that childrenwith textbooks learn morethan thosewithout books (Heyneman and others 1978; Fuller 1985), and that children with higher levels of learning achievementstay in schoollonger. Moreover,textbooks may benefitgirls more than boysif girlsreceive less of the teachers'time and less instructionalsupport at home,and if their outsidework necessitates more frequentabsences. Whether girls do in fact benefitmore than boysfrom improved textbook availability has not been well-researched.One studyon Peru, however,did find that the availabilityof textbookshad a larger positiveimpact on the educationalattainment girls than it did on boys,W suggestingthat parents decisionsto schooltheir daughtersmay be moresensitive to the qualityof educationprovided than are their decisionsto educatetheir sons.

W The availabilityof readingor math books wasassociated with 0.5 yearsadditional education for males and 0.7 years for females(King and Bellew1989). 272 EducatingWomen

Box 8.7 Moroo's lIdustrial and Commerdal Job Trining Program The composition-ofteb labor force 'n Moroccois chagng as many women,driven by economic :necessity,enter the abormarket in searchof paidworkl In responseto a 75 percentincrease in the ntuniberof woen lookng for wor'k,a 20percent female unemployment rate, and severeshortages of trsed technicias andsklled industrialworkers, Morocco's Office of TechnicalTraining and Job Devment (OF??!) intated the Industrialand CommercialJob Trainig Programfor Women. The fiyr program began'in 1979in Casablancaand Fez, foUowinga reviewof nonformal educationprograms for womenand a studyof women'straining needs and resources.Young women withat least 12 yearsfdeducation were recrited to the commercialcenters for traaing in accountin -and'secretarial skd those with9 years of schoolingwere recruitedto the industrialcenters for trainingi raftingeletricit, andelectronics, courses which they had been reluctantto atteid in the --past. : :... . -Togn admissioninto the:programs, women competed with men on the national entrance examiniaton,but womenwere sometiies givenprioriy for admissioninto certain specializations '0T ma,de'special efforts to informwomen about the examinationand to encouragethem to apply NState owne'dradio and televsionstations, which advertised the programsand examinationschedule, *specifically informed the:public that womenwere eligibleand should be encouragedto apply. Notificationswere -also publised in newspapersand distributedto all secondaryschools. OFPPTiprovided counseling for womento guidethem into areas that best suitedtheir aptitudesand preferencesand,4to1facilitate employment after graduation,OPPI I was to establisha formaljob placemen-tservice. Although the formal servicenever made it off the drawingboard, an informal mechanismwas introduced.OFPPI's informalstrategy consisted of convincingemployers to accept women for th'eir datory two month apprenticeship. The informalmechanism worked well. Typically,employers were pleased with the apprentices,offered them permanentjobs, and requested more female apprenticesthe followingyear. A 193 -evaluationshowed'that female enrollment in the projectcenters had reachedtheir targets. Enrollmentincreased from 0 to 99 in the industrialprogram, from 20 to 51 in construction,iand from 30 to 144in commerce.Dropout rates werecomparable for malesand females and employmentrates were hg. Seventypercent of womengraduates were placedin jobs, a rate much higher-'thanfor women with the same levelof formaleducation but no vocationalskills training. But, best of all graduateswere earningas much,if not more, than theycouldhave earned in the publicsector, where sala'riesare basedon yearsof formaleducation and attainmentof the baccalaureatediploma, which none of them had. ..Sourwes.Lycette (1985, 1986); USAID (1978, 1983).

Improvingfirst the overallquality of educationmay be the mostproductive investment option to attract and retain girls in schools. Investmentsin subsidizinggirls to attend low quality schools or establishing alternativelow qualityeducational programs, as illustratedby Pakistan'smosque school effort, are unlikely to attract or retaingirls. Moreover,since girls are more likelyto drop out than boys,they shouldbenefit disproportionatelyfrom qualityimprovements. Educating Women 273 Alleviaft poverty No discussionof female education in developingcountries is complete if it does not address the role poverty plays in undermining efforts to improve it. Absolute poverty is a condition of many children in India, Pakistan and Bangladesh. These three countries alone account for 40 percent of child deaths in the world, 45 percent of the world's malnourished children, 35 percent of children out of s-cool, and over 50 percent of children living in absolute poverty (UNICEF 1990). Simply lowering the costs of education or raising its benefits will probably do little to rescue them. Their survivaldepends on the education of their parents and on their parents' ability to support them. But often, income needs prevent women from participating in literacy,fmily planning, health and nutrition, and skili training programs that aim to improve their living standards.) Attracting women to these programs demands efforts even greater than those required to attract their daughters to school. It requires an immediate opportunity to increase their income. Poverty reduction programs in Bangladesh illustrate this.)9J In the aftermath of the CivilWar and the famine of 1974,Bangladesh faced problems of crisis pro?ortions. Families were destitute, literacy was low, and the population was growing at an alarming 3 percent a year. In response, an integrated and multi-targetedWomen's Program was launched in 1975 to raise the incomes of rural mothers and thereby motivate them to voluntarily accept family planning, acquire reading and writing skills, and learn more about health, nutrition, maternal and child care.A/ The program had three major components.

One component -the vocational training centers -targeted war-affectedmothers and children. The centers taught women weaving,tailoring, embroidery, knitting, jute craft, and food preparation skills. They provided free day-care, education, food and medical attention for the women's children, and a "scholarship"of 2 to 2.5 Takas a day for attending. When the women completed the program, they were invited to work in affiliated production centers. The second component - the Women's Cooperatives - increased the earning capacity of agricultural laborers by exploiting agro-based skills that women presumably already had. The cooperatives offered loans, training, and nutrition and literacy classes. Completing the program were The Mother's Centers which targeted wives of landless laborers.-'/ The centers offered programs in health, literacy, home economics,family planning,and recreation. Classes were held four to five afternoons a week and programs vere from six months to one year in duration. To attract women, some centers also offered skill training.

L A discussion of poverty alleviationprograms is outside the scope of this chapter. Readers are directed to The World Development Report 1990 for a review of these programs.

^-/ Descriptions of these programs were drawn from Gerard and others (1977),Rahim and Mannan (1982), World Bank (1985b, 1986a, 1989e),Canadian International DevelopmentAgency (1985),UNICEF (1977), Huq and Mahtab (1978),Rahman (1977),Alauddin and Faruqee (1983).

W At the same time, there were numerous other women's projects such as cooperatives,cottage industries, Food for Work, and a variety of income generating training projects run by government and NGOS and supported by many international organizationsincluding OXFAM, UNICEF, Canadian CIDA, the World Bank, NORAD, and USAID, to name just a few.

WJSome of the Women's Centers effectivelyappealed to low-middleand middle-incomewomer 'however. 274 EducatingWomen The outcomesof the three programshave been positive.P/Several evaluations reported that a majorityof participantshad raised their incomesand that familyplanning acceptance rates amongparticipants were almostdouble the nationalaverage g Due to the indisputableoutcomes, the programscontinue to expand. There are currently100 productionunits in 20 upazilaswhich employ11,000 women, 1,500Women's Cooperativesin 100 upazilaswith 60,000 members, and 1,600Mother's Centers in 40 upazilaswith 40,000 members. Cumulatively,the programshave recruited an estimated230,000 family planning acceptors.w Bangladesh'swomen's programs demonstrate that earning an incomeis necessaryfor poor women to participatein educationalprograms. The moneyearned also booststheir self-confidenceand givesthem greater decision makingauthority at home and more control over reproductivedecisions. An Indian woman,for example,when askedwhether her participationin an associationof vendorshad anyeffect on the husband'sattitude toward her, responded:'Yes, my husbanddoesn't beat me any more and he can't leaveme, becauseI'm the one who bringsthe loan from the bank."(UNESCO 1976 p.5)

Improvingthe Chances for Success

Sinple and Affordable Approaches are more Sutainable Many recommendedmeasures for increasinggirls' educationare not new, but most have not been incorporatedinto the regulardynamics of the educationsystem and its planningcycle. They remain as pilot or experimentalprojects, frequently sponsored by short-termsupport from localand internationalNGOs and donor agencies.Many are short-lived;most are highlylocalized activities. That theyhave not generally becomean integralpart of the planningprocess belies the commitmentand concemexpressed by more and moregovernments about thisissue. To gainthe sustainedsupport of educationand financeofficials, these programsmust be both administrativelyfeasible and cost-effective,and theymust be consistentwith other objectivesin the educationsector. Interventionsshould be simpleto implementso as notto tax undulythe administrativecapacity of education ministriesand local educationofficials who are imbuedwith the task of monitoringthem. For example, althougha few have succeeded,scholarship programs tend to place a greater burden on administrative resources. Determiningwho gualifies,how muchio grant, and how to monitorthe gratees can be quite complex.Moreover, successful programs may require combining several interventions at once or in tandem in order to address differentbarriers. For example,establishing schools closer in rural communitiesmay not be sufficientto raise girls' enrollmentif parents see no benefit from formal educationsuch that simultaneousinformation campaign (perhaps through extension services) may be necessary. In addition,programs should be affordable. Some programsare expensiveand cannot be financially sustainedfor long. In the poorest countrieswhere nationalbudgets are often severelyconstrained by competingdemands, meeting this criterionis essential. For example,building boarding schools for rural girlsin the primaryor postprimarylevel can entaillarge construction and maintenance costs and thus cannot

22J Unsuccessfulcenters were those where womencould not substantiallyincrease their earnings.This occurredwhere the demandfor the productwas slack, where quality control was weak, where no marketing mechanismswere availableto marketthe products,and whereno income-earningskill was offered. In some of the Mothers' Centers for example,trainng was limitedto producingjute craftswhich the slackened internaland externalmarkets could not absorb. Thesecenters were attractive to middle-classwomen with leisure time to spare for educationaland recreationalactivities.

Wf A UN studyalso showeda 48 percentreduction in fertilityamong members of the MothersCenters.

2/ The estimatesby programare: VocationalTraining Centers - 112,l00,Women's Cooperatives - 20,000, and Mother's Centers - 98,000. Educating Women 275

serve as the cornerstone of government efforts to increase the enrollments in rural areas. Cheaper alternatives need to be explored. Sharpening the targeting of programs to those with a real need also increases cost-effectivenessby reducing total program costs and ensuring that those who benefit the most are reached. But, even targeted subsidy programs should have a well-definedlifetime at the end of which an evaluation is needed. How funds are spend can be critical. Inadequate recurrent expenditures result in badly maintained buildings, crowded classrooms, and shortage of qualifiedteacher. and textbooks. As discussedabove, these conditions lead to specific barriers to female edu^ation. Limited government funds, cuts in the social sectors, and rapidlygrowing youth populationsexacerbate this problem. Developinginterventions--whether constructing school facilities or modifyingschool curricula--raises issues about allocating resources when they are available: What level of resources are needed to achieve objectives? How should programs be financed? Additional funds achieved through a rednction in spending for other social programs that also benefit women and girls, for example, would be coanter to the objectives of the policy. Further, which programs should receive support from central government? Which should be financed and administered locally? The success of policy reforms and progra as depends criticallyon clear policy goals and strong central leadership, but the vitalityof past programs has also relied profoundly on local support. It is important to recognize that the choice of policy is not always rational or based on the benefit-cost calculations of government. Often, they are based on the strength of social or communitypressure. Both in urban and rural areas, communityorganizations exist that can be mobilized for a variety of support programs for local schools. Hence, resources for female education can be expanded by cultivating communityor local support for programs. Support does not necessarilymean financingcapital investments or major recurrent budget items; it may simply mean assistance in administeringand monitoring programs. Devoting administrative energy to the issue of raising girls' and women's education so that it receives attention at all levels of government is essential to achievingthis goal. Local organizationsthat work with women and the poor usnally know a great deal about the communities they work and live in.>' They can provide insights to guide policy efforts in their areas. They can also assist in identifyingtarget groups, in assessing their specificneeds and resources, and in determiningthe willingnessof communities to undertake specific actions.

Knowlege ia Still Key: Monitor, Evauate and Inforn Many tasks lie ahead for governments, education policymakers,and development specialists, but it is important to understand the nature and causes of the problem. Information is essential for accurately diagnosingthe barriers to females'education and identifyingwhich programmatic strategies work and which do not. Ministries of Education do not often gather, tabulate or report data on enrollment, dropout, and repetition by gender. Gender-specificdata collected over a number of years is essential to monitoring the situation and evaluating progress. Comparing school data across geographic regions and demographic groups are often necessary to dissect the problem and isolate trouble areas for targeting policy changes. As the preceding chapters have shown, the answer to why the gender gap in education exisis lies in a complex mix of economic and cultural factors that stem from the home, schools, communities,and nations at large. From anecdotes, simple research, and practical experience, one can easily assemble a long Ulsof such factors. And such lists already exist. The challenge that remains is determining which barriers are key to specificsettings or subpopulations,and what specificpolicy measures are appropriate and affordable. For example,it is not enough to know that parents car- about whether there are female teachers in schools. The following questions must also be addressed. What share of women in the teaching force in primary schools is necessary to raise the enrollment of girls? Is a 50-percent female share necessary to increase girls' enrollment and achievement to the same level as boys'? Or is the presence of at least one female

W For example, women's organizations are very active in several Asian countries. 276 Educating Women

teacher in primary schools in trouble areas adequate? Comparing the cost-effectivenessof competing measures implies ranking the budgetary cost of achieving a given level of benefit associated with each measure. It requires an understanding of parents', children's, and communities' attitudes, perceptions and preferences. For this purpose, a variety of data collection methods, includinghousehold, community and school surveys, are required. But for developing countries which often lack the equipment, personnel and money to support large data collection or to maintain reliable basic education data, it is unrealistic to expect systematic information gathering activities to be a government priority. Surveyscan use up considerable time and funds to design and implement, and demand a cadre of skilled scholars to analyze and interpret the data. However, this is an area where assistance from NGOs, donor agencies, and communities can be useful. International NGOs and donor agencies can sponsor the transfer of knowledgeand experience regarding surveydesign and field implementation, and the sharing of instrument design and data entry software across countries, thus, reducing the cost of data collection to any specific country. Involving communities and local organizations can also help ensure the reliability and completeness of data collected, and exert sufficient political pressure on government for a timely analysis and reporting of such data. Ongoing local projects on girls' and women's education are a valuable source of information on the determinants of enrollment, attendance and completion. A thoughtful and systematicevaluation of their performance could provide key insightsinto these determinants. In addition, local women's organizations already frequentlypossess varied information on the education and training activities of women, but need to channel these to appropriate government planners.

Bmader Pociew Matter

Intervention through specificallytargeted programs is ncither the only nor necessariy the most cost- effectiveway by which the government can influence female education. Broader education policies matter. Though they may not focus on girls' education, these policies can affect girls differently than boys. For example, the more the government increases its support for primary education, the greater the relative beneat to girls, because girls are more likelyto quit school after the prmary level. Improvements in the quality of rural primary schools can have a larger marginal impact on girL' learning because girls are less likely than boys to enroll in better quality schools farther away from home. In turn, the goal of extending basic education to all cannot be achieved without targetting girls, particularly those in rural areas. Policies outside the education sector also affect women's schoo!ng. The delivery of family planning programs and child care services can alter the lives of youngwom .profoundly. Familyplanning programs can make it possible for them to postpone childbearing and continue their secondary education. As discussed in earlier chapters, teen pregnancy is frequently onc reas.on why girls drop out of high school. Because childbearing is still the primary responsibility of women, child care servies make employment outside the home feasible, thus, indirectly increasing the appeal of higher education for women. And, employmentpolicies designed to overcomethe social barriers to women working in higher payingjobs will induce not only larger school enrollments, but also better performance in school by girls.

Concluding Remarks

We set out to identifyapproaches undertaken by various groups to relax, circumvent,or eliminate barriers to girls' and women's education in developingcountries, and to investigatethe effectivenessof these efforts. The investigation was limited and shaped by information on, and suggested strategies to raise, female participation in educational programs contained in education-related documents and literature reviewson barriers to fenr'les' education. As such, it omitted potential strategies to expand access or raise the demand for females' education about which no information was accessible. Three such approaches, for example, are eliminating school policies that bar the enrollment of uregnant or married girls, encouraging girls to delay pregnancy and marriage, and instituting hiring quotas to counteract wage and cmployment discriminationin the labor market. EducatingWomen 277

The effortswe found have two strikingand perhapsinterrelated characteristics. First, most effortsbegan and ended as pilot projectswith short-term funding and implementationsupport provided by donorsand non-governmentalorganizations (NGOs). They have rarely been an integralpat of nationaleducation developmentplans and haverarely resulted from national education policy-making and planningprocesses. Second,although we locatedabundant materials related to thisreview, the dearth of evaluationswas striking and the absenceof any cost informationwas evenmore discouraging.Given this, few strong conclusions can be drawn about the relative effectiveness,particularly the cost-effectiveness,of variousmeasures designedto raise girls' and women'sattendance in educationalprograms. Notwithstandingthe absenceof soundevaluations, this reviewdoes showthat some strategieshave been effective;others bave failed. Somehave yielded mixed results, and for others,no evidenceexists to support or contradicttheir effectiveness(See Table 2). Parentshave respcnded positively to monetaryincentives in the formof scholarshipsand to culturallyappropriate facilities and femaleteachers where cultural norms restrictfemales' presence in publicplaces and their interactionwith males. Alternativeor safetynet schools have also been attractiveto girlswho missedthe chanceto attend regular primaryschool and for those whosework schedules conflict with the regularschool day. Raisingthe benefitsof educationby improving schoolquality, as wellab training women for occupationsin growthsectors of the economywhen combined with strong recruitmentand placementefforts also appear to be promisingstrategies. In contrast, distributingfree uniforms,and vocationaltraining that is not directlylinked to employmentand where recruitmentis weak,have not yieldedthe expectedresults. Table 8.2 A Summary of the Effectivenessof Strategies to Improve Girls' and Women's Education Based on Country Experiences

InsufficientEvidence to Drawa Objective Effective Not Effective Conclusion

Lowerthe cost of Culturally Distributingfree Programmed education appropriate uniforms instruction facilities Female teachersflome production technologies Scholarships Day care Alternative Schoolfeeding (morningor programs evening)schools Raise the benefits Vocationaltraining Vocationaltraining in Revamping of education for growthsectors economy'snon-growth curriculaand books of the economy sectors, -ot directly to introduce when directly linkedto employment broader roles for linkedto and no recruitmenteffort females employmentand Information withstrong campaign recruitment 278 EducatingWomen There is,however, little information on -':eh tojudge the effectivenessof programmedlearning, revamping cuirriculaand textbooksto introducebroader roles for females,home technologies, day care, schoolfeeding programs,and informationcampaigns. There has not been sufficientexperimentation with these strategies to identifythe circumstancesunder which they maybe effective.Similarly, simply expanding school places is ofteninadequate when the culturaland monetarycosts are too highor the benefitsto few.Understanding the conditionsunder whichexpanding opportunities for femalesmay increase levels of attainmentand the strategiesthat need to accompanya supply-sideapproach merits further researchand experimentation.

ResearchLs also required on the importanceparents and girls place on the qualityof educationwhen makingtheir schoolingdecisions. Low qualityeducational programs may not only undermineotherwise soundstrategies, as illustratedby Pakistan'smosque school program, but by raisingthe qualityof programs alreadyout there, fewerspecial efforts may be requiredand availableresources can be more effectively targeted towarddifficult subpopulations. The well-establishedbenefits of educatingfemales, the differencesin contextsacross countries,and the paucityof informationon best strategiesto raise femaleenrollment provide a clearjurtification for special effortsin researchand experimentation.In this regard,several tasks lie ahead for governments,education policymakers,and developmentspecialists seeking to improvefemale education. Developingan effective approachrequires setting goals, identifying target groups by examiningcurrent patterns of femaleenrollment by geographic,demographic and incomegroups, determining which barriers are keyto specificsettings or subpopulations,developing sustainable, con-effective strategies, determiningthe level of investment necessaryto attainspecified goals, and ensuring financial support, monitoring programs, and evaluating their outcomes.As such,improving female education should be a topicof nationaleducation policy concern, one that is integratedinto educationdevelopment plans to ensure adequateplanning, monitoring, evaluation, and financialsupport. Developingcountries, howo;ver, often lack the resourcesto maintaininformation systems with gender- specificdata disaggregatedby geographicand demographicgroups on school enrollment,di op-out, repetition,and achievement.Even fewerhave the resourcesto devoteto household,school or community surveysto deterrmineparents', girls', and communities'attitudes, perceptionsand preferences. Yet, governments,however, can supportthe exchangeof knowledgeand experienceabout informationsystems, surveydesign, implementation, and data entry softwareto reduce the cost of data colection to any one country.There are alsoother sourcesof informationinto whichgovernments can tap. Localorganizations that workwith women and the poor, for example,usualy know a great deal aboutthe communitiesin which they work. They can be an invaluablesource of assistancein guidingpolicy, identifying target groups, assessinga community'sneeds and resources,and determiningits willingnessto undertakespecific actions. Ongoinglocal projects on girls'and women'seducation are alsosources of informationon the determinants of femaleparticipation and attainment, and on administrativelyfeasible approaches. A systematicevaluation of their performancewill greatlyimprove governments' and donors' capabilityto designand implement effectiveprograms. EducatingWomen 279 References

Alauddin, M. and R. Faruqee. Popuaton and Famiy Planing in BJigIadesh. Washington D.C.: World Bank StaffWorking Paper, Number557. Washington, D.C.: World Bank, 1983. Ankrah-Dove,L 'The Deploymentand Traiing of Teachegsfor RemoteRural Schools iii Less-Developed Countries.' Intemaional Review of Education, 28 (1):3-27, 1982 Arriagada,A-M. "TheEffect of Job Trainingon PeruvianWomen's E loymentand Wages."The World Bank,Population and HumanResources Department. Washington,D.C.: World Bank, 1989. Ather, SA. 'US-AID ScholarshipProgramme for Female Students in Sharasti Upazila, ComiLla," Evatuation RepontNo. PD-AAY-603,Dhake, 1984. Begum,K, S. Akhter, and S. Rahman. 'An Evaluationof BRAC'sPrimar; EducationProgramme." Dhaka: Instituteof Educationand Research.Report submittedto the World Bank,mimeo, 1988. BoothroydRA. and D.W. Chapman,"Gender Differences and Achievementin LiberianPrimary School Children." IntemationalJournal of EducationalDevelopment, 7 (2):99-105,1987. Borcelle, G. Jobs for WomernA Pleafor Equality of Opporunity, TechnicalEducaton, VocationalTraining and Employment. Paris: UNESCO, 1985. Bowman,M-J. and C ArnoldAnderson. 'The Participationof Womenin Educationin the Third World." New York:The Ford Foundation,1978. Brozek,J. "BehavioralEffects of Energy and Protein Deficits." Proceedingiof the Intemnatonaland NutritionConference. Washington,D.C.: The NationalInstitutes of Health, 1979. Busdllo,1. "FemaleEducational Attainment in LatinAmerica." Educationand EmploymentDivision, Populationand Human ResourcesDepartment BackgroundPaper Series No. PHREE/89/16. Washington,D.C.: World Bank, 1989. Butterworth,Barbara. Ministry of Educationin NepaLPhone conversation 11/10/89.

Canadian Inte,national Development Agency.Report on the Wonen's ProgwrmsPopulation III Bangladesh (Phase1). Ottawa,Canada, 1985. Chavez, A. and C. Martinez. 'SchoolPerformance of SupplementedChildren From a Poor Rural Aream Symposiafrom the InternationalCongress of Nutrition,pp. 393-402,1981. Chen,L.C., E. Haq and S. D'Souza. "SexBias in the FamilyAllocation of Foodand Health Care in Rural Bangladesh."Population and DevelopmentRewew, 7 (1): 5S.70,1980. Coletta, N. J. and M. Sutton. "Achievingand SustainingUniversal Primary Education:International Experience Relevant to India." Populationand Human Resources Department, WPS 166. Washington,D.C.: World Bank,1989. Culbertson,R. et al. 'Primarykducation in Pakistan."Washington, D.C.: Development Associates, 1986. Dove,L A. Teachersand TeacherEducation in DevelopingCountries. London: Croon Helm, 1986. D'Souza,S. and L. C. Chen. "Sex Differentialsin Mortalityin Rural Bangladesh."Population and DevelopmentRevew, 6 (2):257.70,1980. 280 Educating Women

El-Sanabary, N. "Determinants of Women's Education in the Middle East and North Africa: Illustrations from Seven Countries." Education and Employment Division Background Papers Series No. PHREE/89/14. Washington, D.C.: World Bank, 1989. Eisenmon, T. and A. Wasi. "Koranic Schooling and its Transformation in Coastal Kenya." Intemational Joumal of Educational Development, 7 (2):89-98,1987. Ferreri R. "Acceso de las Mujeres a las Carreras Tecnicas." Paris: UNESCO, 1974. Finn, J.D., L Dulberg, and J. Reis. "Sex Differences in Educational Attainment: A Cross-National Perspective." HarvardEducation Review, 49 (November), 1979. Fuller, B. "RaisingSchool Quality in DevelopingCountries. What InvestmentsBoost Learning?"Discussion Paper No. 2. Washington, D.C.: The World Bank, 1986. GaDer J., and others. "The Influence of Early Malnutrition on Subsequent Behavioral Development." Journal of the Ameiican Academy of Child Psychiatry,22 (1):8-15, 1983. Gerard, R., M. Islam, and H. Jahan. Trainingfor Women in Bangladesh. Dacca: UNICEF, Women's Development Program, 1977. Grantham-McGregor, S.M., CA. Powell, S.F. Walker, and J.H. Himes. "Nutritional Supplementation, PsychosocialStimulation, and Mental DeveAopmentof Stunted Children: The Jamaican Study," The Lancet. 338 (8758):1-5,July 1991.

Halpern, R. "Effects of Early Childhood Intervention on Primary School Progress in Latin America." ComparativeEducation Review, 30 (2):193-215,1986. Halpern, R. and R. Myers. "Effects of Ealy Childhood Intervention on Primary School Progress and Performance in the DevelopingCountries." Report prepared for the Bureau for Program and Policy Coordination, USAID. Michigan:High Scope Educational Research Foundation, 1985.

Harley, B. "Mohalla Schools." UNICEF Education Briefing Paper 2, 1979. Huq, J. and M. N. Mahtab. Report on Income-GeneratingSkills for Women. Dacca: International Labor Organization,Asian Regional SkillsDevelopment Program, 1978. Herz, B. "Womenin Development: Kenya's Experience."Finance and Development,June, pp. 43-45, 1989.

Heyneman, S.P., and WA. Loxley. "The Effect of Primary School Quality on Academic Achievement Across Twenty-NineHigh and Low Income Countries."American Joumal of Sociology,88 (6):1162- 94, 1983.

Haughton, J. "EducationalFinance and Reform in Mali."mimeo. Washington, D.C.: World Bank, 1986.

Hyde, ki ImprovingWomen's Education in Sub-Saharan Africa." Education and Employment Division Background Papers Series No. PHREE/89/j5. Washington, D.C.: World Bank, 1989 ILO/UNESCO. "Joint ILO/UNESCO Committee of Expci is on the Application of the Recommendation Concerning the Status of Teachers." Report of the Fifth Session, Gtn.aeva,October 18-27, 1988. Khan, S. R. 'Barriers to Female Education in South Asia." Education and Employment Division Background Papers Series No. PHREE/89/17. Washington, D.C.: World Bank, 1989. Educating Women 281

King, E. and R. BeHew. 'The Effects of Peru's Push to Improve Education."Policy Planning and Research Working Papers No. WPS 172. Washington, D.C.: World Bank, 1989. King, E. and A. Hill, Eds. Women's Education in DevelopingCounties: A Review of Barniers,Benefits and Policy. Washington, D.C.: Johns Hopkins UniversityPress, (forthcoming). Lembert, M. and I. Nieves. "Women's Technical Training in Latin America: Lessons from Experience." WID Working Paper No. 20. Cambridge, MA: Women and International Development, Joint Harvard/MIT Group, 1987. Levin, H. M. "Economic Dimensions of Iodine DeficiencyDisorders." Prepared for The Pevention and Controlof lodine DeficiencyDisorders, Stanbury J., J. Dunn, and B. Hetzel, eds., 1986 Levinger,B. "SchoolFeeding Programs in Less Developed Countries: An Analysisof Actual and Potential impact." Washington,D.C.: USAID, Office of Evaluation,Bureau for Food and VoluntaryAid, 1983. 'School.------. Feeding Programs: Myth and Potential." Prospects:Quarterly Review of Education, 1984.

Lockheed,M., A. Verspoor, and others. ImprovingPKimay Edurion in DevelopingCountnes. Washington, D.C.: World Bank, 1990. Mallon, N. "The Community Classroom." WorldEducation Reports, No. 28, Spring, pp. 15-18,198 9. Martin, L. G. "Evaluation of the Bangladesh Female Secondary Education Scholarship Program and Related Female Education and Employment Initiatives to Reduce Fertility." Washington, D.C.: International Science and TechnologyInstitute, Inc., 1985. McGregor, S. "The relationship Between Developmental Level and Different Types of Malnutrition." Human Nutntion, 36:319-320,1972. McSweeneyand Friedman. "Lack of Time as an Obstacle to Women's Education: The Case of Upper Volta." Co.nparativeEducation Review, June, pp. S124-S139,1980. Ministry of Education, Guatemala. Ministry and World Bank mimeos, 1989.

Ministry of Education, Mali. School enrollment data, n.d. Molnar, A. "Forest Conservation in Nepal: EncouragingWomen's Participation." Seeds, No. 10, 1987. Moock, P. and J. Leslie. "ChildhoodMalnutrition and Schoolingin the Terai Region of Nepal."Education and Training Discussion Paper Series No. EDT 17. Washington D.C.: World Bank, 1985. Murnane, R. "EducationPolicy ad Economic Incentives:Tailoring Incentivesto Desired Responses." Paper presented for the World Bank Seminar on Teacher Costs and Effectiveness,April, 1987.

Lycette, M. "An Evaluation of Industrial and Commercial Job Training for Women in MAorocco:Project No. 608-0147." Washington, D.C. : International Center for Research on Women, 1985. "The...... Industrial and Commercial Job Training for Women Project in Morocco." Washington, D.C.: USAID, 1986. Nag, M., C. Peet, and B. White. "An AnthropologicalApproach to the Study of the Economic Value of Children in Java and Nepal." CurrentAnthropology, 19 (2): 293-306,1978. 282 Educating Women

Naik, C. 'An Action-Research Project on Universal Primary Education - The Plan and the Process.' In Women'sEducation in the ThirdWorld ComparadvePerspectives, Kelly, Gail P. and C. Elliott,eds. Albany State Universityof NewYork, 1982. Pilain,A. C. WomenEducation, Equality: a Decadeof Expelment Paris: UNESCOPress, 1975. Po11itt,E. "EarlyChildhood Intervention Programs in LatinAmerica." Report presentedto the Officeof LatinAmerica and the Caribbean. New York: Ford Foundation,1979. Malnut,idonand Infectionin the Classroom.Paris: UNESCO. 1990. Popkin,B. and Lim-Ybanez."Nutrition and SchoolAchievement." Social Science and Medicine,16:53-61, 1982. Powell,C. and others. "AnEvaluation of Givingthe JamaicanGovermment School Mean to a Classof Children." HumanMNtiition, 37:381-388, 1983. Rahim, M. A. and M. Mannan. An Evaluton of the Women'sVocational Training Programme for Popuaion Educaton and Contl Dacca:External Evaluation Unit, PlanningCommission, 1982. Rahman,J. ^SocialWelfare Mothers Club." Dacca: Government of the PeoplesRepublic of Bangladesh, SocialWelfare Department, 1977. Richardson,S. A. 'The LongRange Consequencesof Malnutritionin Infancy A Studyof Childrenin Jamaica,West Indes."Topics in Pediatics,2:164-176, 1980. Robinson,W., N. Makary,and A. Rugh."Fourth Annual Report of the Studyof USAIDContributions to the EgyptianBasic Education Program." Washington, D.C.: Creative Associates, 1987. Safilios-Rothschild,C. "Accessof Rural Girls to PrimaryEducation in the Third World:State of the Art, Obstacles,and Recommendations."Washington, D.C.: Office of Womenin DevelopmentOctober, U.S. Agencyfor InternationalDevelopment, 1979. Schofield,S. Developmentand the Problemsof VllageNutrition. London: Croon Helm, 1979. Schultz,T. P. "Benefitsof EducatingWomen." Washington, D.C.: World Bank, Background Papers Series, Educationand EmploymentDivision, Population and Human ResourcesDepartment, 1989. SeetharamuA.S. and M.D. Ushadevi. Education in RuraAreas.New Delhi: Ashih Publishing House, 1985. Simon,P. A., D. T. Jamison,and M. A. Manning "GenderDifferences in Goiter Prevalence:A Review." Los Angeles:UCLA, February, 1990. Svedberg,P. "Undernutritionin Sub-SaharanAfrica: Is there a Gender Bias?" Joumalof Development Sudies, 26 (3):469486,1990. Teacher'sResource Center. "A Study of MosqueSchools in Sind with Special Reference to GirlsEducation." Paper prepared for the World Bank.Karachi, 1989. Thein,T-M. "Evaluationof the FemaleEducation Scholarship Program Supported by the AsiaFoundation." September,1988. Tilak,J. "FemaleSchooling in East Asia."Education and EmploynentDivision Background Papers Series No. PHREE/89/13.Washington, D.C.: World Bank,1Q89. EducatingWomen 283 UNDP. 'Nepal: Teacher Training (Equal Access of Women to Education)."Project Fmdingsand Recommendations."Paris, 1982. UNESCO. "TheContinuing Effort for Women'sRights." UNESCO Features No. 697. Paris UNESCO, 1976. ---. -People'sDemocratic Republic of Yemen.Completion Report of the SecondEducation Project Credil 865-YDR."Paris: UNESCOEducational Fmancing Division, 1985a. 'Trainingof EducationalPersonnel Focused on Girls and Women."Report of a Sub-regional Workshopfor the Trainingof EducationalPersonnel Focused on Girls and Women. Katmandu, Nepal 19-27September 1984. Bangkok.UNESCO Regional Office for Educationin Asia and the Pacific,1985b. ------. "TowardsEquality of EducationalOpportunity." Report of the Visitsof the RegionalPanel on the Educationof Girls,27 May- 10June. Bangkok:UNESCO Regional Office for Educationin Asia and the Pacific,1985c. ------"Education of Girls in Asia and the Pacifi.: Report of the RegionalReview Meeting on the Situationof Educationof Girls for Universalizationof PrimaryEducation." Bangkok: Regional Office for Educationin Asia and the Pacific,1986. Statstcal YearbookParis, 1988. 'BangladeshCompletion Report Frst PrimaryEducation Project Credit 1054-BD.0Paris, 1989.

UNICEF. Repoit of a FeasibilitySurvey of Productive/Income Generatng Activites for Women in Bangladesh Dacca, 1977. WithinHuman Reach. A FutureforAftica's Children. NewYork, 1985. TheState of the Word's Children. New York, 1990. U.N.Economic Commission for Africa. 'Trainingand EmploymentOpportunities for Out-of-SchoolGirls in Dar Es Salaam."Addis Ababa, 1984. U.N. StatisticalOffice. Wistatdata. 1988. USAID. Project GrantAgreement Between the Kingdomof Moroccoand the UnitedStates of America for Industrialand CommercialJob Trainingfor Women." Waslngton, D.C., 1978. "Industrialand CommercalJob Trainingfor Womenin Morocco,Fnal Repor." Washington, D.C., 1983 ---. Policy Research Initiahve, Somalia TeacherIncentive Systems: Final Report. Washington, D.C.: Improvingthe Efficiencyof EducationalSystems Project, 1989. Warwick,D., F. Reimers and N. McGinn. 'The Implementationof Reformsin the PrimarySchools of Pakistan."Harvard Grauuate Schoolof Education,1989. White, K, and others. IntegrahngWomen into DevelopmentPrograms: A Guideforlmpkmentahon for Latin Americaand the Caribbean.Washington, D.C.: International Center for Researchon Women,1986. Whyte,J. 'EncouragingGirls into Scienceand TecLnologySome EuropeanInitiatives. Paris: UNESCO, l984. 284 EducatingWomen World Bank. "StaffAppraisal Report BangladeshFourth EducationProject." South Asia RegionalOffice ProjectsDepartment. Washington,D.C.: World Bank, 1980a. .Report-...... and Recommendationof the Presidentof the InternationalDevelopment Association to the ExecutiveDirectors on a ProposedCredit to the People'sRepublic of Bangladeshfor a Fourth Education(Primary) Project." Washington,D.C.: World Bank,1980b. .-----.'Colombia Staff Appraisal Report. Subsector Project for Rural BasicEducation." Washington, D.C.:World Bank,1982. "Project------.Performance Audit Report, MauritaniaFirst EducationProject (Credit 459-MAU)." OperationEvaluation Department. Washington, D.C.: World Bank,1983a. "The Demandfor Health andSchooling in Mali:Results of the Communityand ServiceProvider Survey."Research Project 672-72. Washington,D.C.: World Bank,1983b. "Staff---. Appraisal Report BangladeshSecondary Primary Education Project." Washington,D.C.: World Bank,1985a.

'-..'' ".StaffAppraisal Report BangladeshThird Populationand FamilyHealth Project."Washington, D.C.:World Bank, 1985b ------.'Project PerformanceAudit Report BangladeshFirst PopulationProject (Credit 533-BD)." Washington,D.C.: World Bank,1986.

------. "ProjectCompletion Report Fourth Education Project (Credit 892-Pak)." Washington, D.C.: World Bank,1987a. ------. "StaffAppraisal Report: Islamic Republicof Pakistan Third Primary Education Project." Washington,D.C.: World Bank,1987b. ------"Project Completion Report People'sDemocratic Republic of YemenSecond Education Project (Credit 865-YDR)."Washington, D.C.: World Bank,1987c. 'Jamaica------. Summary Review of the SocialWell-Being Program." Human ResourcesOperations Division,Country Department m, LatinAmerica and CanbbeanRegion. Washington, D.C.: World Bank,1988a. "Staff.----.Appraisal Report, Bhutan Primary Education Project." Populationand HumanResources Division,Country Department I, Asia RegionalOffice. Washington, D.C.: World Bank,1988b. "Pakistan:.------.Education Sector Strategy ;view."Report No. 7110-PAKWashington, D.C.: World Bank, 1988c. "Colombia..------. Social Programs and PovertyAlleviation An Assessmentof GovernmentInitiatives." LatinAmerica and the CaribbeanRegion. Washington, D.C.: World Bank,1988d. ------. "BangladeshProposed General EducationProject StaffAppraisal Report". Washington,D.C.: World Bank, 1989a. "Bhutan------.Development Planning in a Unique Environment."Washirgton, D.C.: World Bank, 1989b. ------. Women in Pakistan:An Economic and SocialStrategy. Volume 1. Report No. 8009-PAIL Washington,D.C.: World Bank, 1989c. Educating Women 285

------. "Report and Recommendation of the President of the International DevelopmentAssociation to the Executive Directors on a Proposed Development Credit in and Amount Equivalent to US$26.0 Million to the Republic of Mali for an Education Sector Consolidation Project." Washington, D.C.: World Bank, 1989d.

------. "Project Completion Report Bangladesh Second Population and FamilyHealth Project (Credit 921-BD)." Washington, D.C.: World Bank, 1989e.

------. "StaffAppraisal Report Kingdomof Morocco Rural Primary Education Project." Population and Human Resources Division, Country Department 11, Europe, Middle East and North Africa. Washington, D.C.: World Bank, 1989f.

------. World DevelopmentReport. Washington, D.C.: World Bank, 1990a.

------. "StaffAppraisal Report The Gambia Education Sector Project." Washington,D.C.: World Bank, 1990b.

------. "Staff Appraisal Report of the Sind Primary Education Project." Population and Human Resources Division, Country Department 1, Europe, Middle East and North Africa. Washingxon, D.C.: World Bank, 1990c.

------"*A Dialogue on School-Feeding." The LAC Connection, Newsin Human Resources, vol3, No. 3., 1991.