Terence Roopnaraine, and

With contributions from

Natalia Smith, Elif Altinok, Nurfer Çelebioğlu, Sema Cemal, Wahid Quabili, Tugba Atalar, Suzulay Hazar, Ozlem Agaoglu and Selahattin Erhan

Submitted to the

General Directorate of Social Assistance and Solidarity Prime Ministry, Republic of

Prepared by the

International Food Policy Research Institute 2033 K Street, N.W., Washington, D.C. 20006, U.S.A.

In collaboration with the

AGRIN Co. Ltd. Bestekar Sokak 30/3, Kavaklidere 06680, Ankara, Turkey

March 26, 2007

The International Food Policy Research Institute (IFPRI) and its collaborator, the AGRIN Company Limited, gratefully acknowledge the General Directorate of Social Assistance and Solidarity (SYDGM) for funding the Impact Evaluation of the Conditional Cash Transfers Project, under which this final evaluation report has been completed. The authors are especially indebted to Yadigar Gökalp, Director of the Social Risk Mitigation Project (SRMP), for her advice and support. We thank Süha Barlas, Ümit BaĢaran, Elif Güden, Derya HaĢemoğlu, Nazile Kademli, Gökhan Karatepe, Yalçın Kaya, Müge NiĢancı, Tomris OkĢar, Ekrem Serin, Hamdi Tomaç, Sevtap Turan, SavaĢ Yılmaz, and Feridun Akgöbek of the SRMP for their cooperation. We are also grateful to Cahit Bağcı from the State Planning Organization, and Mr.Mustafa Acar, from Kırıkkale University, for their comments on the study design. For the Quantitative Assessment, managers and staff of local Social Solidarity Foundations provided invaluable information and we are grateful for their support. The study would not have been possible without the cooperation of the 2,905 families from 26 who patiently answered numerous questions during the comprehensive household survey, which is the basis of this study. Our special thanks go to those survey respondents. We are indebted to enumerators and supervisors of the OPTIMAR for their dedication and hard work in carrying out the household survey. For the First Qualitative Assessment, 36 Social Solidarity Foundations managers and staff provided invaluable information and helped with arrangements to contact other stakeholders, and we are grateful for their support. The study would not have been possible without the cooperation of 557 people who have shared their views with us to help evaluate the Conditional Cash Transfer program. Appreciation is expressed to Yusuf Ziya Ozcan, Oyku Yalcin, Yasemin Surmeli, and Fisun Tatligil for their participation in the collection of qualitative information and editing the report. For the Second Qualitative and Anthropological Study, managers and staff of local Social Solidarity Foundations in Diyarbakir, Ergani, Samsun, Tekkekoy, Van, and Gurpinar provided invaluable information and assistance, and we are very grateful for their support. We also thank the muhtars and other government officials and local leaders, doctors, nurses and other health professionals, and teachers and school managers, who took time to speak with us and help us to understand local experiences with the program. Above all, we are grateful to the 87 families from six communities in the provinces of Diyarbakir, Samsun, and Van—men, women, and children—who welcomed our field researchers into their communities and homes for several months, and shared detailed aspects of their lives and experiences. Without their generous time, candidness, and insights, we could not have hoped to have come up with these findings; we only hope that the results can contribute toward improvements in the CCT and other social programs in ways that benefit them, their communities, and poor communities elsewhere. At IFPRI, we thank the director of the Food Consumption and Nutrition Division, Marie Ruel, for her overall guidance. We thank Nelly Rose Tioco, Jay Willis, and Marinella Yadao for their help with the technical production of all reports.

ii

ACKNOWLEDGMENTS ...... ii

EXECUTIVE SUMMARY ...... vi

1. INTRODUCTION ...... 1 1.1 Background ...... 1 1.2 International Experience on CCT Programs ...... 1 1.3 Evaluating the CCT Program in Turkey ...... 2

2. SALIENT FEATURES OF THE CCT PROGRAM IN TURKEY ...... 5 2.1 Social Risk Mitigation Project and CCT ...... 5 2.2 CCT Targeting Mechanism...... 6 2.3 Beneficiary Targeting and Monitoring of Eligibility Status ...... 6 2.4 Payments ...... 8

3. METHODOLOGY AND THE DATA ...... 9 3.1 Quantitative Assessment of Program Impact ...... 9 3.2 First Qualitative Assessment ...... 11 3.3 Second Qualitative Assessment ...... 12

4. INSTITUTIONAL ISSUES AND CONSTRAINTS AFFECTING PROGRAM PERFORMANCE ...... 15 4.1 Sources of Program Information ...... 16 4.2 Knowledge of Benefit Packages ...... 16 4.3 Understanding of Conditionalities ...... 17 4.4 Strengthen Communications Systems...... 18 4.5 A Community Liaison...... 18 4.6 Summary of Constraints on Program Implementation ...... 20

5. PROFILE OF SURVEY HOUSEHOLDS...... 22 5.1 Household Characteristics ...... 22 5.2 Consumption Patterns ...... 27 5.3 Assets and Amenities ...... 30 5.4 Labor Force Participation ...... 30 5.5 Private Transfers and Remittances ...... 32

6. SYNERGIES BETWEEN CCT AND OTHER PROGRAMS ...... 33 6.1 Complementary Support for Education ...... 33 6.2 Complementary Health Support ...... 35 6.3 Participation in Other Social Assistance Programs ...... 35

7. TARGETING PERFORMANCE ...... 37 7.1 Targeting Performance: Distribution of CCT beneficiaries and Rejected Applicants Across Income Groups ...... 38 7.2 The Targeting Process at the Community Level...... 40 7.3 Selection Criteria, Local Understandings, and Perceptions of Fairness ...... 41 7.4 Comments on the Proxy Means Test Scoring Formula ...... 43

iii 8. IMPACT ON EDUCATION ...... 45 8.1 Educational Attainment: Descriptive Analysis of Quantitative Household Survey Data ...... 45 8.2 Assessing the Impact on Education ...... 48 8.3 Factors Explaining Education Decisions: Overview of Findings ...... 52

9. IMPACT ON HEALTH AND NUTRITION ...... 60 9.1 Vaccination ...... 60 9.2 Illness ...... 62 9.3 Nutrition ...... 64

10. IMPACT ON PREGNANCY ...... 67

11. CCT EXTERNALITIES: HOUSEHOLD DYNAMICS AND WOMEN‘S STATUS .... 72 11.1 Adult Work Patterns ...... 72 11.2 Child Labor Patterns ...... 73 11.3 Use of the CCT Money ...... 74 11.4 Intrahousehold Relations ...... 75

12. CONCLUSIONS AND RECOMMENDATIONS ...... 79

REFERENCES ...... 87

APPENDIX A: TABLES ...... 89

APPENDIX B: FIGURES ...... 102

APPENDIX C: RESEARCH QUESTIONS AND INSTRUMENTS FOR THE SECOND QUALITATIVE AND ANTHROPOLOGICAL STUDY ...... 105

FIGURES

7.1 — Distribution of CCT education and health-beneficiary households, by HBS expenditure deciles...... 38 8.1 — Factors affecting education decisions by regions ...... 53

TABLES

3.1 — Application and eligibility ...... 9 5.1 — Characteristics of survey households ...... 22 5.2 — Literacy and migration status of household head ...... 23 5.3—Percentage distribution of population by gender and age group ...... 24 5.4— Characteristics of CCT education beneficiary households, by regions ...... 25 5.5— Characteristics of CCT health beneficiary households, by regions ...... 26 5.6 — Food budget share ...... 27 5.7 — Nonfood budget share ...... 28 5.8 — Calorie consumption, shares, and costs ...... 29 5.9 — Selected household asset ownership ...... 30

iv 5.10 — Selected dwelling facilities ...... 31 5.11 — Labor force participation, population age 15 years and above ...... 31 5.12 — Private transfers and remittances received ...... 32 6.1 — Assistance received from government programs (excluding the CCT program) ...... 36 6.2 — Benefits received from nongovernmental organizations (NGOs) ...... 37 7.1 — Distribution of CCT beneficiary and non-beneficiary applicant households, by expenditure decile ...... 39 8.1 — School enrollment rates...... 46 8.2 — Students who never missed school in November 2005 ...... 47 8.3 — Academic performance of students in 2005...... 47 8.4 — School enrollment rates for education beneficiary households, panel survey results .. 48 9.1 — Immunization of children of age 6 and below ...... 60 9.2—Immunization of children of age 6 and below, panel survey results ...... 61

v

The Conditional Cash Transfer program in Turkey is a targeted social assistance program that is designed to provide support to the poorest 6 percent of the Turkish population. CCT is a sub-component of the Social Risk Mitigation Project (SRMP) of the General Directorate of Social Assistance and Solidarity (SYDGM). The specific objectives of the CCT in Turkey are to increase school attendance rates for the poor, decrease dropout rates, increase immunization coverage, and enhance the utilization of health facilities for the 1.1 million target beneficiaries. The program uses an indicator-based administrative targeting method to identify ultra- poor households with children aged 0-6 years, school-aged children 6-17 years, and child- bearing-age women. This final evaluation report on the impact of the CCT program is a synthesis of three studies: The First Qualitative Assessment (Kudat et al. 2006), completed in early 2006; the Quantitative Assessment (Ahmed et al 2006), completed in late 2006; and the Second Qualitative and Anthropological Study (Adato et al. 2007), completed in early 2007. We request readers to go to these original sources for more information. At the outset, it is important to note that the evaluation study encountered several methodological, techincal, and physical constraints. These constraints affected the efficacy of the impact assessment mainly in two ways: (1) In order to measure program impact, it is necessary to construct a counterfactual measure of what might have happened without the program. The most powerful way to construct a valid counterfactual is to randomly select beneficiaries from a pool of equally eligible candidates. However, a randomized approach was not feasible for the CCT evaluation, as the program had already been implemented before the evaluation, and it is national. Thus, there is no immediately obvious comparison group in areas without the program. As a result, the evaluation had to employ somewhat weaker, nonrandomized approaches for impact assessment. (2) The impact assessment was based on a cross-sectional household survey and a two-round panel survey. The surveys were delayed due to problems in locating sample households, the Bairam holidays, and winter conditions that affected the speed of the field survey work. The delay in carrying out the first round of the panel survey shortened the elapsed time between the first and the second survey rounds, which was not sufficient for measuring the impact of the CCT program on a number of outcomes. Further, some institutional factors limit the potential impacts of the CCT program. Main institutional issues and constraints are summarized below.

Institutional Issues and Constraints affecting Program Performance

Operational Performance of the CCT Program The CCT program has been fully operational across Turkey since 2004. However, the results of the Quantitative Assessment—based on a large household survey that was nationally representative of the CCT program—suggest that, about 44 percent of CCT- education beneficiaries and 63 percent of health beneficiaries joined the program in 2005. CCT payment delays have been quite common and there have been large fluctuations in payment levels. On average at the time of the household survey, CCT education and health

vi beneficiaries, respectively, received about 64 percent and 57 percent of the total amount entitled. Nevertheless, the survey data also suggest that the regularity of CCT payments has been increasing over time. Since CCT payments are conditional on school attendance and regular visits to health clinics by the beneficiaries, furnishing this information by the relevant institutions to the local foundations and entry of these data to the system are needed to verify the maintenance of beneficiary status for the disbursement of payments. Apparently, logistical and administrative difficulties—which probably caused payment delays mainly in the early stage of the program—are being gradually removed over time. Findings of the quantitative and the two qualitative studies show that beneficiaries know very little about the CCT program. A very large proportion of the CCT beneficiaries and nonbeneficiary-applicants have little or no knowledge of the application and selection criteria, nor of the program conditions. This knowledge is essential in ensuring that the program functions as intended. Communication has not been adequate to promote this understanding. Successful communication is expected to be especially challenging in a social program targeting the poorest six percent of the population, where levels of education and literacy are low. Beneficiaries are not only poor but also women of a generation where education was even less common for girls than it is now. They also have little experience dealing with formal state institutions, and CCT programs are procedurally complex. Some degree of misunderstanding is thus expected, and many program staff have made an effort to communicate program procedures. However, the research finds that this effort has not been sufficient for the successful transmission of necessary information, which in light of these challenges requires particularly intensive effort. The First Qualitative Assessment found that staff shortages, and insecurity of the staff with respect to program continuity challenge the expansion of program coverage as well as the attainment of full program impacts for current beneficiaries. The CCT staff not only has a heavy workload but also is often asked to assist with other equally pressing workload peaks faced by non-CCT programs carried out by the Foundations. Reducing the workload in high CCT concentration areas, regaining the momentum in capacity building, increasing staff specialization in the Foundations, and creating mechanisms for greater decentralization of both the selection and evaluation processes would enhance CCT targeting and impacts. According to SRMP headquarters staff, in order to reduce this workload of the SYDVs and accelerate the implementation process, two steps were taken since the First Qualitative Study: (1) 5,894 person/month of staff employment services were provided to them; and (2) 282 social workers were employed for the provinces/sub-provinces with a population over 50,000 with SRMP financing. Employment of some of these social workers was financed by the own resources of SYDVs.

Sources of Program Information Household survey data indicate that social networks and educational institutions played key roles in helping poor families access the CCT program. Schoolteachers were the most common source of information on education benefits. During the Second Qualitative and Anthropological Study, schoolteachers, neighbors and relatives, the muhtars and the Foundation offices were cited by beneficiaries as the main sources of program-related information. Beneficiaries do not have a single channel for information about the CCT program, and find it difficult to get reliable information.

vii Knowledge of Benefit Packages Most beneficiaries do not understand the differences by sex or type of benefit within their packages. Beneficiaries on the whole understand that the money is given to poor families, and many also recognize that one part of the package is designed to be education support. Very few people know about the health component. Understanding of Conditionalities The Quantitative Assessment finds that the majority of the beneficiaries do not know the conditions they are required to meet in order to maintain their eligibility. About 90 percent of the education-beneficiaries and 87 percent of the health-beneficiaries claimed that no one informed them of the program rules. These findings are supported by both qualitative studies. Opinion about CCT and Program Conditions The Second Qualitative and Anthropological Study finds that beneficiaries and key informants are enthusiastic about the benefits of the CCT program, emphasizing that the transfers help them to feed their families and send their children to school. Key informants are much more uniformly positive about the conditional nature of the benefits than the beneficiaries are. There is a strong provincial dimension to this finding: almost all the ‗negative‘ responses to conditionality came from Van. Community Liaison Among beneficiaries, opinions about a liaison show great variation across the three provinces. There is a general agreement that it should be a woman. Objections raised were based on gender (inappropriate activities for women) and insufficient social capital. Key informants were also ambivalent about the liaison suggestion. The majority of respondents agreed that the (paid) liaison should be a woman. Other important qualities included: literacy, honesty, religion (Muslim), independence, speaking ability, propriety, language (Kurdish in relevant areas) and knowledge of institutions. Opinions about whether the liaison should be a local person were divided. Some beneficiaries said their husbands or mothers-in-law would not allow them to attend meetings, or else their household responsibilities would make such meetings impossible. These respondents said that they would prefer to meet with the promoter in their own homes. Beneficiaries wanted the liaison to help with the resolution of problems and doubts: where was their money? How much should they receive? When could they collect it? Key informants suggested that a promoter should educate beneficiaries about program goals and the conditional nature of the program and health and nutrition issues. Both key informants and beneficiaries were keen on the idea of inviting health and education personnel to liaison meetings.

Targeting Issues

Targeting Effectiveness The Quantitative Assessment concludes that the CCT program is well targeted to the poorest. The income distribution of CCT education and health beneficiaries is highly progressive. High percentages of all beneficiaries belong to poorer income groups, particularly the health beneficiaries. Half of all health-beneficiary households and over one- third of all education-beneficiary households are among the poorest 10 percent of all

viii households in the income distribution at the national level. Indeed, no education beneficiaries and health beneficiaries belong to the richer 30 and 40 percent of all households, respectively. In any form of targeting, there are problems of exclusion (i.e., leaving out those who are needy) and inclusion (i.e., providing benefits to those who do not need them). The evaluation results suggest that, while the CCT program effectively reaches the poorest, a considerable number of nonbeneficiary-applicant households are also among the poorest but excluded from the program. The First Qualitative Study concludes that the CCT appears to have succeeded in targeting the poorest segments of the Turkish society. At the regional level, this is reflected in the fact that the least developed regions and provinces receive the bulk of the benefits. The comparison of beneficiaries with non-beneficiary applicants systematically shows the former to have greater disadvantage in terms of assets and well-being. In addition, the comparisons of these two groups conducted in February and September 2005 also shows that the poverty focus of the program has sharpened as it continued to mature. Stakeholder discussions and interviews with beneficiaries reveal that a low level of leakages (that is, program benefits accruing to non-needy population) exists and, to a large extent, this results from issues pertaining to the application form; continued improvements made in the application form appear to have reduced leakages over time. The Targeting Process at the Community Level The Second Qualitative and Anthropological Study finds that, in all communities prospective beneficiaries are assisted in the preparation of their application materials by some combination of Foundation officials, teachers, health personnel and shopkeepers. Applications are then sent to Ankara for final selection. While this process on the whole seems to be working well, some concerns were expressed by some stakeholders, particularly related to over-centralization and the public‘s perception of them as program gatekeepers. Selection Criteria, Local Understanding and Fairness of Selection

Understanding of the selection process Most beneficiaries and nonbeneficiaries identified poverty as the main criterion for selection. Education needs were also cited as important. Most key informants across Second Qualitative and Anthropological Study communities who addressed this issue felt that people did not understand the selection criteria. Fairness: Errors of exclusion and inclusion Although a significant minority of household informants believed that the targeting process had not been completely fair, in no community was there a strong tendency for non- beneficiaries to believe it had been unfair. Several key informants felt that targeting had not been fair. A point that was raised in Samsun and Van is that there are characteristics in the proxy means test that contain a bias and do not always work out fairly, such as those related to social insurance and certain assets. Tension because of selection The majority of respondents say that there are no tensions between households which have been selected for the program and those which have not. Selection appeals Program documentation outlines an appeals process which applies both to non- beneficiaries who wish to appeal a targeting decision and to beneficiaries who have been

ix suspended or expelled from the program. Foundation officials and local authorities say that such a process indeed exists and that they encourage unsuccessful applicants to re-apply for the benefits. Some of the beneficiaries interviewed had been able to join the program because their appeals were successful. Why some do not apply Among key informants, the most commonly cited reason for not applying to the program is lack of knowledge. This is followed by lack of support or inability to handle the application procedures, honor, transportation costs, and a belief that only a limited number of applications will be accepted. Some people did not apply because they felt that they would be financially ineligible. One man mentioned his disability as a reason: he was bedridden and did not hear about the program. Language was also mentioned once as a reason (by a Kurdish interviewee). It was suggested by a small number of interviewees that legal marriage was an obstacle to their applications.

Impact on Education The CCT program‘s foremost objectives are to increase school attendance rates, for the poor in general and for secondary-school girls in particular, and to decrease dropout rates. The Quantitative Study estimated impacts of the CCT program using a regression discontinuity design (RDD). RDD is an impact evaluation method that estimates the effect of participating in the CCT program by comparing average outcomes between beneficiary and nonbeneficiary households whose proxy means test score is near the threshold score used to determine program eligibility. Key results of the RDD estimates are highlighted below. The CCT program raises secondary school enrollment for girls by 10.7 percent. The CCT program raises primary school attendance for girls by 1.3 percentage points. In secondary schools, education transfers from the CCT program raised girls‘ attendance rates by 5.4 percentage points. The CCT program appears to have improved the quality of education (in terms of test scores) for children enrolled in primary school. Given the relatively small size of the impact of the program on primary school attendance, this effect is not likely to be caused by increased time in school. Instead, the education transfers may be helping beneficiary households to make better use of the schooling inputs, by increasing attention on schooling within the family and allowing families to provide children with more time to study or to focus on their school work. RDD estimates however suggest that the CCT program has no positive impact on primary school enrollment rates. The absence of an effect of the program on primary school enrollment is due in part to the already high enrollment rates at that level of education. Further, estimates show no evidence that the CCT program affected the rate of progression from primary school to secondary school. One likely explanation for this result is that, children often have to commute to other communities to attend secondary school due to the low concentration of secondary school relative to primary school, which increases education costs and time to attend school. The demand-side intervention of CCT alone may not be sufficient to overcome this supply-side constraint.

x Factors Affecting Schooling Decisions and Impact of the CCT Parents place a strong priority on educating boys in all study areas, but are much more ambivalent about girls education. However, even where they are supportive of education, there are many different factors influencing their schooling decisions. The CCT helps many households overcome constraining factors. Across the study areas, households were evenly divided between those who say that the CCT affects their education decisions and those who say it does not. However, even in many cases where they say it does not, it appears that it does but that it is not always considered appropriate to acknowledge that children are sent for money. For parents who value education, the CCT is not a causal factor but it helps make the expense manageable. However, many other households said that they could not afford to send their child without the CCT. The financial assistance also helps to better supply and dress the children, making them more willing to go. In some households, including a majority in Van, CCT money could not compete with the socio-cultural factors constraining education particularly in the village, even at the primary school level. In some households, women conveyed that the CCT backs them up when they want to continue their daughters‘ schooling, when their husbands or other relatives are leaning against it. They can argue that the government wants them to send their girls, and that they will lose the money if they do not send them. However, this is not the case with everyone— especially in rural areas. In the household studies, eighteen categories of factors influencing schooling decisions were identified in the Second Qualitative and Anthropological Study, and are outlined below. These mainly involved economic, social, and cultural dimensions, There are important regional differences. The role of education in children‘s future looms large in parents‘ schooling decisions. For boys, education is seen as necessary for getting employment and better employment. However, high unemployment rates make people cynical about the value of education, particularly in rural areas where people work in the fields and see this as honorable. Nevertheless, there are also those in the village with an appreciation of education and an aspiration for a different life. The main role of education in girls‘ futures is seen as freedom from ignorance. Literacy enables them to go out freely and provides them with the ability to interact with people outside their home. Some also see education as opening a door to better living conditions and a better husband. Unfortunately, many parents only see the value of primary school to girls, particularly in the Van study areas. In Diyabakir and Van it was generally seen as inappropriate for girls to work once they become mature. Although relatively rare, there were exceptions—aspirations for girls to have careers Even in Samsun where girls do work, secondary school was not seen as a priority. Gender issues were the second highest-ranking factor in schooling decisions across the study areas. There are several dimensions to this gender category. The first and most prevalent relates to honor, reputation and sexuality—the perceived threats to girls and their families‘ honor posed by boys at school and men on the street, if girls go to school after they have reached maturity.

xi The other main gender dimension is the expectation that girls will get married, and that their primary role in life will be as wife and mother therefore, the value of education is questioned and in some regions seen detrimental. There are also economic dimensions that weigh in favor of marriage vs. education. The CCT program can contribute toward eroding these barriers. However, in some areas it is insufficient to overcome biases. Transportation to school, and the closely related issue of the location or lack of secondary schools in some areas, is another significant constraining factor in all study provinces, particularly in the rural villages as well as in some urban areas. Schools are often far from people‘s homes and there is no transportation, affordable transportation, or transportation considered safe for girls. People would prefer secondary schools in their communities, not only to alleviate transportation problems and costs, but also to enable parents to keep an eye on their children and preserve their honor. Children‘s performance at school was a surprisingly strong criteria influencing whether parents continued sending them. Those who do well in school are seen as more likely to be able to take advantage of their education. Parents whose children perform well tend to be enthusiastic about education, whereas for failing students they do not force them to continue. Children‘s preference for school—whether they like school or not—is another weighty factor in whether they continue. Safety was another significant factor in parents‘ decisions, primarily in Diyarbakir. Some risks include; men and boys, disciplinary problems, drugs, and violence. Parents feel it is the responsibility of the state to provide safety; some teachers said they felt powerless to control these problems and asked for state intervention. The cost of school expenses, and the broader state of poverty, is the most frequently cited factor of all. This is significant because this is the main factor the CCT responds. Burdensome expenses included transportation and books, followed by school supplies, food, and pocket money. There were regional variations in costs. Poverty also results in boys being taken out of school to help earn money. Another factor related to cost is the reliability of the grant—whether it can be counted on to regularly arrive and thus cover school expenses. Complementary Support for Education The First Qualitative Study shows that, CCT support for education largely overlapped with support from other sectors with respect to the specific target population. The national mobilization in 2001/02 within the education sector to meet the goal of increasing minimum education to eight years for all provided the sector with substantial additional financing. Moreover, the military, the municipal sector, the private sector, civil society organizations as well as individual citizens provided support to universal education and much of this support was directed at poor communities and poor families. As a result, CCT became one of the many different types of support for education in the period 2003-05. The SYDVs themselves provide either in kind or cash support to poor students at the outset of each school year. The Ministry of Education (MOE) provides free schoolbooks for

xii grades 1-8. Poor students in boarding schools have all their expenses, including meals met, but cannot benefit from CCT. A large number of students are provided with full scholarships from MOE or from organizations such as the Turkish Education Foundation. Again, those who receive scholarships from state institutions cannot benefit from CCT. NGOs, such as Chose Your Own Sibling match needy students with the wealthier one to provide direct assistance for education. Nationwide organizations such as Mehmetcik (Soldier) Foundation and Deniz Feneri (Lighthouse) also provide similar support.

Impact on Health and Nutrition

Vaccinations The CCT program aims to increase immunization coverage of children from poor families and to promote usage of health facilities. RDD estimates of the Quantitative Assessment reveal that health transfers from the CCT program lead to an increase of 13.6 percent in the full-immunization rate for preschool children. For children below 6 years of age, the rate of full immunization against tuberculosis (BCG); diphtheria, whooping cough, and tetanus (DPT/triple); polio; and measles jumped from 43.8 percent to 57.4 percent as a result of participating in the health component of the CCT program. The Second Qualitative and Anthropological Study found evidence of the (incorrect) belief that CCT benefits are related to vaccination in two study provinces, Van and Diyarbakır. In a small number of cases, it was evident that vaccination was regarded with suspicion, as a potentially harmful practice. Another suspicion was that vaccination caused infertility. Intra-household discrimination, i.e. vaccinating some children and not others, was also not in evidence. The Socio-Cultural Context for the Health Component of CCT Traditional approaches to illness and healing coexist with ‗biomedical‘ responses. The most important factor underlying each decision which a family takes about what kind of medical care to seek is severity. Quality of Service and Opinions about Health Services The majority of informants who spoke about their experience with doctors, clinics and hospitals referred to very negative experiences. Compounding these problems is the issue of shame and body-centered embarrassment. A point which comes up again and again in the interview data is that people do not go to hospitals for check-ups: they go when they are seriously ill. Health Controls: Beliefs, Attendance and CCT Data from household interviews suggests that few if any people know about the check-up (‗health control‘) conditionality. Where people do acknowledge or recognize the presence of a conditionality, they tend to believe that this concerns vaccination and not check-ups. One doctor also noted that, once applications to the program have been made, people make a conscious effort to attend the clinic regularly, in the hope that this behavior will be observed and taken into account, enhancing their prospects of being accepted as program beneficiaries. Nutrition Although increasing food consumption is not an explicit goal of the CCT program, the income from program transfers and increased interactions with health services may have increased food consumption for beneficiary households. In the Quantitative Assessment,

xiii RDD impact estimates for the full sample of beneficiaries found no average impact of the CCT program on per capita calorie consumption. However, for a subsample of beneficiary households who received transfers from the CCT program on a more regular basis, the CCT program increased their per capita calorie consumption by 22.6 percent relative to a comparison group of nonbeneficiary-applicant households. The First Qualitative Assessment finds that beneficiary families‘ allocation of CCT money for food improves nutritional standards and contributes to children‘s school performance. Mothers generally believe that their ability to provide better food to children makes a difference in their school performance. The Second Quantitative Assessment shows that CCT money has allowed beneficiaries to increase amounts of food bought and purchase food items that previously they would not have been able to afford. However, this change did not translate into better perceived household nutrition.

Impact on Pregnancy Over the course of the program, concerns developed, particularly among some health providers and Foundation staff, that the CCT program—particularly, the pregnancy component of the program—might be creating incentives for families to have additional children that they would not otherwise have. There was concern that this program might be undermining some of the progress that the health sector has made to strengthen family planning. The quantitative and qualitative studies both set out to investigate this issue to determine whether or not it was creating these unintended consequences. The RDD impact estimates show no evidence to support these claims. Quite the contrary, the estimates show that receipt of education or health transfers from the program actually reduces the probability of a woman of child-bearing age (16-49) becoming pregnant by about 2-3 percent. It is somewhat surprising that the program appears to discourage pregnancy, though this may reflect in part the effect of the additional income from the program and visits to health clinics on household fertility decisions. The CCT Pregnancy Benefit: Fertility Decisions and Questions on Incentives In all study areas of the Second Qualitative and Anthropological Study, the vast majority of beneficiaries and non-beneficiaries consistently say that no one would get pregnant to get money. Among those who answered the question, 32 people said they did not know anyone who did this; while 3 said that they did. About half of the key informants in Diyarbakir and Van thought that the pregnancy benefit was causing more pregnancies and thought the birth rate had risen. The Second Qualitative and Anthropological Study lists several reasons why it seems unlikely that people would get pregnant in order to receive the CCT. These are outlined below: There are very strong social pressures on women to get pregnant in all three provinces, from their husbands and especially their families. Having a son continues the family lineage. Having many children, especially many sons, is an indicator of power and makes a family less vulnerable. Birth control is widely viewed as a sin. In this environment, getting a cash grant is not needed as an incentive to get pregnant. There are also economic reasons. Many children mean more labor and greater income earning potential. This is especially true in rural areas where livelihoods

xiv are agriculturally based, though families are smaller now that fewer people depend on the land. Despite these pressures, many women discussed the reasons why not to have more children. They were expensive, expenses are high, women become exhausted. In all three provinces, rumors spread that the CCT is much higher than it is, e.g. 120 YTL per month, or 500-1000 YTL in some unspecified time period. This might explain why there was some incentive affect under those circumstances. However, the majority of people said that ‗no one would get pregnant for money,‘ and many found the question to be humorous, absurd, or offensive. The unreliability of the grant in terms of amount, frequency, and duration also make people less likely to depend on it, especially to make a lifetime decision. Although the CCT is unlikely to be at the center of a fertility decision, it may be that with people caught in the middle and deciding—the CCT could tip the balance. But this does not appear to be the main scenario. The larger issue is the more diffuse perception that the state will take care of people‘s children, from birth through adulthood—but this raises a challenge to the entire CCT concept, not the pregnancy benefit, which most people do not know about. However, again we think that there is stronger evidence that fertility decisions will largely be made on considerations other than a cash grant.

CCT Externalities: Household Dynamics and Women’s Status

CCT does not appear to have any impact on work patterns among adults. They generally follow similar work patterns as they did before the implementation of the CCT program. Impact of the CCT on child labor appears modest. Children continue to value work in the same way they did before, and adults continue relying on it not only as a coping strategy but as a way to increase the marketability of their children. Clothes were the most frequently bought item, followed by food, and school supplies. There is an understanding that CCT money should not be used for food purchases, but primarily for school expenses. Control and decision-making of CCT money is not always left to the woman, although some women do have an important role to play in decisions related to nutrition and health. By giving women ownership of the CCT money, the program has opened opportunities for women to participate in the public sphere and make gender-specific roles less restrictive. Benefits from this change reported by women include independence, freedom, and confidence. In general, there were few reported changes in spousal relations. The First Qualitative Study suggests that CCT has a positive impact on the increase of women‘s participation in society. With few exceptions women during interviews recount with pride their participation in the application process and their use of funds disbursed to them, especially to feed the household and educate their children. CCT helps women formally register their marriage and obtain birth certificates for their children. CCT has helped change people‘s mind-set about female education. Parents as well as female students benefiting from the program recognize that without CCT there would be greater reluctance to send girls to school beyond the first eight years level of basic education (grades 1-8).

xv

Overall Views of Stakeholders In the First Qualitative Assessment, stakeholders hold mixed but generally positive views concerning CCT and suggest specific improvements. Beneficiary households argue in favor of nutrition and education impacts. Fathers generally are content with the enhanced integration of their wives into the broader society and supportive of the education of their daughters. Girls are particularly happy about the support, especially if they are over 10 years of age. Educators invite greater inclusion and support CCT despite the additional burden it imposes upon them. In the Second Qualitative and Anthropological Study, beneficiaries and key informants were mostly enthusiastic about the benefits of the CCT program, emphasizing that the transfers help them to feed their families and send their children to school. Some adverse program impacts were also raised, particularly in the First Qualitative Assessment. Official stakeholders and social commentators have expressed concern that the CCT: (i) significantly increases the burden on health and education institutions;; (ii) increases antagonism against local government institutions when issues arise in CCT implementation; (iii) spreads expectations throughout the country that the poor could rely on the state, one way or the other. However, such expectations were not reflected among actual beneficiaries, and where it was, was expressed as appreciation that the state is there to give them a little bit of help. With respect to the issue of whether the program should continue, the First Qualitative Assessment found that there is a general belief that the program is useful and should continue, with improvements in implementation arrangements, to ensure that beneficial outcomes are achieved. Beneficiary families consistently state that they cannot afford to send their girls to schools outside their communities unless CCT support continues; this is especially so for high school. A large number of ―destitute‖ families have no access to regular financial support so that CCT is often a unique and vital factor in their lives. In these cases, CCT support is used mostly for food and basic needs. Most officials and Foundation staff believe that the program has contributed to increased schooling and preventive health care and add that it should not have been initiated if it were to discontinue within a few years, as this would mean local protests against the Foundations. Many stakeholders, including beneficiary and nonbeneficiary applicants, believe the widespread lack of employment opportunities in the least developed regions makes it imperative to continue to have CCT or a CTT-like mechanism to support the poorest. The overall conclusion of the Second Qualitative and Anthropological Study was that the CCT program is an important, valued program that is having an impact, and should be continued. However, the program faces two major sets of challenges: the first is operational—the need for much stronger communications across all levels of program operations and primarily between program staff and beneficiaries, so that beneficiaries understand the basic logic of the program—their obligations and entitlements. The second challenge is social and cultural, and these factors are harder to respond to. This requires at a minimum intersectoral cooperation and complementary approaches. With respect to both challenges, patience is also required—an understanding that social change takes time—but persistence and dedication to resolving problems from central to local levels will be needed for the program to succeed.

xvi 1

1.1 Background

The premise of the conditional cash-transfer (CCT) programs is that families remain in poverty from one generation to the next because poor parents cannot invest adequately in their children. Decades of research have shown that attention to early child health, nutrition, and education significantly increases children‘s chances of climbing out of poverty later in life. Yet increasing the availability and quality of schools and health services often fails to make much difference when the poor cannot afford them. CCT programs deliberately target those most in need. Reaching the poorest and vulnerable groups is the common thread among CCT programs that rely on both geographic and household level targeting. They provide money to poor families contingent on investments in human capital such as sending children to school and bringing them to health centers on a regular basis. By targeting cash transfers to poor households, CCTs seek to alleviate short-term poverty. By linking the transfers to investments in human capital, they address long-term poverty.

1.2 International Experience on CCT Programs

Because countries allocate substantial resources to CCT programs, the determination of the impacts of these programs has received attention in a large number of countries. The first generation of CCT evaluations was aimed at assessing program impact and operational performance by examining the adequacy of CCT programs' administrative arrangements. They reviewed the efficiency of processes, the extent to which CCT programs reach poor areas and poor households, and the size of expected and unanticipated impacts. The perceptions of the beneficiaries and other stakeholders, and the cost effectiveness of program delivery mechanisms have also been studied. In education, the evaluations included an assessment of changes in school enrollment and attendance rates. In some cases, they also analyzed changes in promotion and repetition rates. In health and nutrition, the evaluations included a wide range of quality of health care indicators. Changes in consumption levels and patterns were central to many CCT evaluations. Evaluations of CCT programs also focused on the relative importance of supply and demand factors in increasing human capital as well as program impacts on maternal and child mortality rates. CCT programs have become an important poverty-reduction tool primarily in Latin America and the Caribbean where they were originally developed, but also elsewhere. Most CCT programs have two components: an education component and a health and nutrition component. The education component consists of a cash grant targeted to households with primary school-age children. In countries with higher educational attainment such as Mexico, Colombia and Jamaica, this component also aims at secondary school-age adolescents. Households receive cash support (and in some cases in-kind support in the form of books and school supplies) if their children are enrolled in school and attend classes regularly. In Mexico, Honduras, Nicaragua, Colombia, and Jamaica, health-related activities were fully integrated with other components of the respective CCT programs, while in Brazil, a

1 This final evaluation report is a synthesis of three studies: The first qualitative assessment (Kudat et al. 2006), completed in early 2006; a quantitative assessment (Ahmed et al 2006), completed in late 2006; and the second qualitative assessment (Adato et al. 2007), completed in early 2007. These text from these sources, written by multiple authors, have been used verbatim in much of this report, and we thus request that the original sources are cited. These original reports also contain much more information and we thus request readers to go to them for additional information.

1 separate program (Bolsa Alimentação) was set up to administer the health- and nutrition- related payments. The health components of all of the CCT programs are based on the assumption that the achievement of optimal health status is constrained not only by low income but also by low demand for preventive health services. Thus, in all of these programs, regular visits to health centers have been a condition for continuing to receive payments. The most popular type of intervention includes a combination of education, health, and nutrition objectives. This category includes programs such as Mexico‘s Programa de Educación, Salud y Alimentación (PROGRESA)2; Colombia‘s Familias en Acción program (FA); Honduras‘ Programa de Asignación Familiar (PRAF); Jamaica‘s Program of Advancement through Health and Education (PATH); Nicaragua‘s the Red de Protección Social (RPS); Bolivia‘s Beca Futuro; Ecuador‘s Bono de Desarrollo Humano; Chile‘s Subsidio Unitario Familiar; and Brazil‘s Bolsa Familia. A second category of programs provide education grants only, including Brazil‘s Programa de Erradicaçao do Trabalho Infantil (PETI), and Agente Joven. A third category, focused on health and nutrition objectives, includes Brazil‘s Bolsa Alimentação and Cartão Alimentação (Rawlings 2004). There is concrete evidence of success from programs in Brazil, Colombia, Mexico, and Nicaragua in increasing enrollment rates, improving preventive health care, and raising household consumption (Behrman and Hoddinott 2000; Behrman, Sengupta, and Todd 2000; Gertler 2000; Hoddinott, Skoufias, and Washburn 2000; Maluccio and Flores 2005; Morley and Coady 2003; Morris 2005; Schultz 2000a-c; Skoufias 2005; Yap, Sedlacek, and Orazem 2001). In tandem with increased school enrollment, CCTs are also effective in reducing child labor (Maluccio and Flores 2005; Parker and Skoufias 2000).

1.3 Evaluating the CCT Program in Turkey

The Conditional Cash Transfer program in Turkey is a targeted social assistance transfer program that provides support to the poorest 6 percent of the Turkish population. CCT is a sub-component of the Social Risk Mitigation Project (SRMP). The specific objectives of the CCT in Turkey is to reach 1.1 million beneficiaries, increase school attendance rates for the poor, decrease dropout rates, increase immunization coverage and usage of health facilities. In line with global trends and the objectives of the SRMP, several qualitative and quantitative evaluation exercises of the Project, including CCT component, have already been completed. As elsewhere in the world, these assessments show that cash transfers were important in responding to the needs of the poor and made important contributions to human capital accumulation. Yet, much is unknown about the perceptions of the stakeholders, the efficiency of targeting, the magnitude and nature of exclusions, level of satisfaction of beneficiaries, and changes in the quality of health and education services in response to the program, among others. The information generated through this assessment would strengthen the empirical basis upon which the Government of Turkey can make informed policy choices to refine and strengthen the CCT program.

2 Launched in 1997, PROGRESA in Mexico (changed its name in 2002 to Oportunidades), the first large scale CCT program in the world, is one of the better known conditional cash-transfer programs.

2 1.3.1 Objectives of the Impact Evaluation

The key objectives of the evaluation are to (1) assess the effectiveness of the program targeting and its coverage; (2) get an unbiased estimate of the impacts of CCT on an agreed set of indicators, and (3) trace through the pathways by which the Social Solidarity Fund (SYDTF) achieves its various impacts. Achieving these objectives required the completion of several activities, including new field data collection, secondary analyses of existing data and desk reviews. The evaluation has been broken down in three distinct components: (a) one large scale quantitative survey complemented by a small panel (longitudinal) survey; (b) the statistical analysis of the results of the surveys and of other sources of data, such as the MIS database of the Project Coordination Unit (PCU) for CCT and the Household Budget Survey (HBS) of the Turkish Statistical Institute; and (c) two qualitative assessments.

1.3.2 Constraints Encountered During the Evaluation3

The evaluation study encountered several methodological, technical, and physical constraints. These constraints are listed below. Methodological constraints: To measure program impact, it is necessary to compare outcomes for beneficiaries to what those outcomes would have been had the program not been implemented. For this, it is necessary to construct a counterfactual measure of what might have happened without the program. The most powerful way to construct a valid counterfactual is to randomly select beneficiaries from a pool of equally eligible candidates. For the CCT evaluation, however, a randomized approach was not feasible. Construction of the counterfactual was further complicated by the fact that the program had already been implemented before the evaluation, and it is national. Thus, there is no immediately obvious comparison group in areas without the program. As a result, for impact assessment the evaluation had to employ somewhat weaker, nonrandomized approaches. In addition to the use of the large household survey (nationally representative of the CCT program) for quantitative assessment of program impact, the assessment also used a two-round panel survey of CCT beneficiary households. However, there was a gap of only 7 months between the first and the second round of the panel survey. This elapsed time is not sufficient for measuring the impact of the CCT program on a number of outcomes. The delay in carrying out the first round of the panel survey shortened the elapsed time between the first and the second survey rounds. The reasons for the delay are explained below. Technical constraints: The availability of a well-organized MIS database hindered the First Qualitative assessment, which included analyses of the MIS data. The evaluation study used the MIS database for the sampling of the quantitative household survey. However, in the beginning of the survey, the survey team could not locate many names and addresses of selected sample households. This

3 In evaluating program implementation, we found several constraints on program implementation that limit potential impacts of the program. These constraints are reported in Section 4.

3 problem was reported to the SRMP. The SRMP staff contacted the Foundations in sample locations to assist the survey teams in finding the sample households. Staff of the local Foundations checked all names/addresses of sample households and provided excellent support to the survey team in identifying the households. Without this support it would not have been possible to reach many respondent households. However, this extensive process resulted in postponing the commencement of the quantitative household survey until the third week of December 2005. Physical constraints: The household survey was also delayed due winter conditions in that affected the speed of the field survey work. During the household survey, a break was given from 8th to 15th January for the Bairam. This also contributed to the delay in completing the survey.

1.3.3 The Final Evaluation Report

This final evaluation report is a synthesis of three studies: The First Qualitative Assessment (Kudat et al. 2006), completed in early 2006; the Quantitative Assessment (Ahmed et al 2006), completed in late 2006; and the Second Qualitative and Anthropological Study (Adato et al. 2007), completed in early 2007. As noted above, we request readers to go to these original sources for more information, and to cite them directly. Following this introduction, Section 2 provides an overview of the CCT program in Turkey. Section 3 discusses the analytical methodology and the data used in the empirical work. Section 4 provides a discussion of institutional issues and constraints affecting program performance. Section 5 portrays the profile of survey households. Section 6 shows the synergies between CCT and other complementary programs. Section 7 assesses the targeting performance of the CCT program. Sections 8 and 9 present, respectively, the impacts of CCT on educational outcomes, and health and nutrition. Section 10 analyzes whether there is any effect of the program on pregnancy. Section 11 discusses CCT externalities, household dynamics and women‘s status. Section 12 provides conclusions and recommendations.

4

2.1 Social Risk Mitigation Project and CCT

The Government of Turkey provides a number of social assistance programs aimed at the delivery of cash and in-kind benefits and social services to individuals and families, who are poor and to individuals with specific needs. The Social Solidarity Fund (SYDTF) and the Social Services and Child Protection Organization (SHCEK) were established as the two main government institutions coping with social risk mitigation, besides municipalities, nongovernmental organizations, such as development foundations, and professional associations. The SYDTF has been one of the main ―social risk mitigation‖ institutions in Turkey since its establishment in 1986. The SYDTF assists vulnerable people or households in urban and rural areas through its affiliated Social Solidarity Foundations (SYDVs) located in each province and sub-province by supplying food as well as heating, education and health support to the very poor and to vulnerable households. SYDTF has supported employability training and income generating subprojects. There are 931 SYDVs in Turkey receiving a regular monthly budget from the SYDTF according to an index based on population, human development, assets, expenditures, and income. Although SYDTF has been assisting poor households in different ways for nearly twenty years, Turkey has not had a well-targeted and designed Social Safety Net (SSN) program for the vulnerable households. The earthquake of 1999 and economic crisis of 2001 led to the realization of the need to reform the SSN by: Reducing fragmentation and duplication of services; Improving targeting mechanisms for the distribution of social benefits; and Increasing service coverage. The Social Risk Mitigation Project (SRMP) was designed to: (i) reduce, over the period 2002-2006, the impact of the 2001 economic crisis on the poorest households; and (ii) improve their capacity to cope with similar risks in the future. These objectives were to be achieved through: (i) an adjustment portion, providing immediate support to the poorest affected by the crisis; and (ii) an investment portion, consisting of three components: 1. Institutional Development: building up the capacity of state institutions providing basic social services and social assistance to the poor; 2. Conditional Cash Transfer (CCT): implementing a social assistance system targeted to the poorest six percent of the population conditional on improved use of basic health and education services; and 3. Local Initiatives: increasing the income generating and employment opportunities of the poor.

The project targets 1.1 million beneficiaries and allocated $360 million to finance the introduction through the SYDTF/SYDVs of an expanded social safety-net targeted to the poorest 6 percent of families (World Bank 2001). In order to ensure that the conditions of CCT were met by the beneficiaries, protocols were signed between the General Directorate of Social Assistance and Solidarity (SYDGM),

5 the Ministry of Health (MOH) and the Ministry of Education (MOE) so that the attendance records of CCT beneficiaries would be regularly reported by the local schools and health centers directly to the Foundations.

2.2 CCT Targeting Mechanism

The CCT program in Turkey uses administrative targeting to identify households with target group members (i.e. children aged 0-6 years, school-age children 6-17 years, and child- bearing age women) who belong to the poorest group in the society (bottom 6 percent of the population in the income distribution). Program administrators use the ―proxy means testing‖ method to identify the poorest. This approach relies on indicators that are highly correlated with household income (or total consumption expenditure), yet are easy to collect, observe, and verify. Points are assigned to selected indicators, and eligibility for program benefits is determined on the basis of a total score, as a proxy for household income. Using data from the Household Income and Expenditure Survey conducted in Turkey in 2001, regression equations are estimated to predict per capita household consumption expenditure, which, in effect, assign weights (points) to the individual indicators. These weights are given by the values of the coefficients of the selected indicators. Total scores are calculated for the respective households by multiplying the values of the individual indicators with their coefficient values, and then adding them up. This way, the proxy means testing method uses the information on all the selected indicators for identifying the needy and the non-needy households.

2.3 Beneficiary Targeting and Monitoring of Eligibility Status

Initially, public information campaigns for the SRMP were held in pilot provinces/sub- provinces. These campaigns are to continue through the lifetime of the CCT project. In this framework, in order to announce the SYDGM‘s activities and to receive public support, people are to be informed about the ongoing activities through the use of information booklets, posters, periodicals, exhibitions, videos, etc. Close cooperation with teachers, health staff, and muhtars (local community leaders) is to be ensured. Posters and brochures for public information are prepared and printed by SYDGM. These materials in addition to the other information and promotion materials which may be developed by the Foundations are put up in the public areas. Following the promotion campaigns at local, regional or national scale, SYDVs receive the applications. The applications are received and recorded every day of the year. CCT‘s target population includes families who cannot send their children to school or who cannot pay regular health visits for their children and pregnant women who cannot have regular health check-ups or give birth in hospitals. Families having no social insurance, expecting babies, having children at 0-6 age and/or school-age children are eligible to apply for CCT benefits. A prescribed application form should be filled and submitted in order to be considered for CCT. These forms are to be delivered through the Foundations, schools, health institutions, and muhtars. The applications for children and students should be submitted by their mothers. In the absence of mother, the child‘s father or a member of the family who is over 18 may apply. In the absence of both parents, the application on behalf of the child can be submitted by his/her legal guardian. The applicants should not be selected to receive CCT benefits if the families have social securities from SSK, Bağ-Kur or Emekli Sandigi (ES).

6 The applications are received at three locations: At SYDV offices: SYDV staff should provide necessary support to those applicants who are illiterate or who need help in filling in the forms. At houses: For those people who cannot reach the SYDV offices or mobile centers due to health conditions or other valid reasons, applications may be accepted at their houses through the help of SYDV staff or other temporarily assigned staff from other public institutions. Other places: In order to avoid the exclusion of people living in areas from where it is difficult to reach SYDV centers, applications may be accepted at different places (schools, sports facilities, village rooms, health centers, etc.). After application forms are received, they are recorded in the Application Form Registration Notebook at the Foundations according to their submission dates. The received applications are numbered, and are filed together with the accompanied documents. Foundations are obliged to present all related documents (application, follow-up, petition of appeal) during the inspections. The data from application forms received from the applicants are entered into computers with custom-designed software. Eligibility for the CCT program is determined by a proxy means scoring formula derived from a model that predicts per capita household consumption expenditure as a function of household demographics, location, housing, and assets variables. Applicants were required to provide data on these variables in the application form. When the data from the applications are entered, the proxy means score of the applicant is displayed. If the calculated score of the applicant is below the predetermined cut-off score, the applicant is eligible to be selected as a CCT beneficiary.4 Verification of the data declared by the applicant is done SYDV staff by randomly selecting at least 70 percent of the eligible applicants. Data verification should be done in cooperation with relevant persons (muhtar, teachers, health care staff, and security staff). This cooperation should be noted in the data verification. In case the information provided by the applicant is not accurate, the applicant is permanently excluded from the beneficiary list. If an applicant is already an active registered beneficiary of other SYDV assistance programs such as food, heating or clothing support, the SYDV may not pay a house visit for data verification. After data verification a final beneficiary list is determined and delivered to foundation offices, schools, muhtars and health clinics. Following the determination of beneficiary and data verification, the eligibility status of the beneficiaries is verified once a year. There is an appeal mechanism for non-eligible applicants who may want the SYDV to reconsider them for the CCT support, and for the beneficiaries whose CCT payment is pending or terminated. Beneficiary status should be sustained under certain conditions. Since the supports provided by the CCT program should depend on positive behavioral changes by beneficiaries, the maintenance of beneficiary status is based on the following conditions: For the children who receive the education support:

o Attend 80 percent of the total school days o Not to repeat the same grade more than once.

4 Based on the available budget for the CCT program, the proxy means cut-off score (COS) that determined the level of poverty that would qualify for the program was set at 10,249.

7

For the children who receive the health support:

o Children at 0 to 6th months require regular check-ups every month. o Children of 7th to 18th months require regular check-ups every two months. o Children of 19th to 72nd months require regular check-ups every six months.

For the women who receive the pregnancy support:

o Regular health check-ups are required every month until the birth. o Birth shall be given in hospital. o Post-birth check-ups are required following the birth.

2.4 Payments

A protocol agreement was signed between the SYDGM and the General Directorate of Ziraat Bank for making the CCT payments through the branches of Ziraat Bank. An agreement was also signed with the General Directorate of Postal and Telegraph Organization to make payments of the beneficiaries living in districts where no Ziraat Bank branch exist or where the large number of beneficiaries makes it difficult to make payments through one branch of Ziraat Bank. Receipts of payments are often facilitated through the distribution of ATM cards. Beneficiaries may receive their ATM cards from the concerned bank branches in their own sub-provinces after signing an ATM contract form. During this process, either SYDV staff or bank staff are supposed to help the beneficiaries who are not literate. Education support should be provided for nine months, with bi-monthly payments in September, November, January, March, and May; and health support for 12 months, with bi- monthly payments in September, November, January, March, May and July. Pregnancy support should be provided from the second month of the pregnancy to the second month of post-pregnancy. As of January 2005, monthly payment amounts for the beneficiaries of the education support are: 18.00 YTL for boys attending primary school 28.00 YTL for boys attending secondary school 22.00 YTL for girls attending primary school 39.00 YTL for girls attending secondary school As of January 2005, monthly payment amount for the beneficiaries of the health support is 17.00 YTL per child. As of January 2005, monthly payment amounts for the beneficiaries of the pregnancy support are: 17.00 YTL per pregnant woman 55.00 YTL after the birth is given in a hospital.

8 This section summarizes the methodological approaches used to evaluate the CCT program in Turkey. The order in which the studies were carried out were: (1) First Qualitative Assessment in Fall 2005; (2) Quantitative Assessment in Winter-Spring 2006; (3) Second Qualitative and Anthropological Study in Summer-Fall 2006; and (4) follow-up panel survey in Fall-Winter 2006. Below, quantitative methods are described first, followed by descriptions of the two qualitative studies.

3.1 Quantitative Assessment of Program Impact (Ahmed et al. 2006)

To measure program impact, it is necessary to compare outcomes for beneficiaries to what those outcomes would have been had the program not been implemented. For this, it is necessary to construct a counterfactual measure of what might have happened without the program. The most powerful way to construct a valid counterfactual is to randomly select beneficiaries from a pool of equally eligible candidates. This has been the approach taken by IFPRI in the previous CCT impact assessments it undertook. For the CCT in Turkey, however, a randomized approach was not feasible. Construction of the counterfactual was further complicated by the fact that the program had already been implemented before the evaluation, and it is national. Thus, there is no immediately obvious comparison group in areas without the program. As a result, for impact assessment the evaluation had to employ somewhat weaker, nonrandomized approaches. These methods typically rely on assumptions that are difficult, but not impossible, to verify (Burtless 1995). A key feature of the design of the CCT Program that affects measurement of program impact is that households must apply to the program. After a household has applied for the program, the rules for inclusion are well defined, based on proxy means tests estimated from a poverty assessments carried out with nationally representative samples. Those that score below a certain cut-off based on a proxy means prediction of their per capita expenditures are admitted into the program and those that score above it are not. Estimating impacts of the Turkey CCT Program relies on the use of quasi-experimental methods that requires construction of a statistical comparison group. Within this framework, there are four distinct groups to consider when identifying the counterfactual, as shown in Table 3.1. The first group comprises those who applied and were eligible (AE), the second, those who applied but were ineligible (AI). The third and fourth groups are those who did not apply, but if they had they would have been eligible (NE), and those who did not apply and were not eligible (NI).

Table 3.1 — Application and eligibility Eligible: Ineligible: Below proxy means cut-off Above proxy means cut-off Applied to program AE AI Did not apply to program NE NI

Because application is voluntary, simply contrasting groups AE and NE is not an appropriate strategy to assess program effects. This is because there may be unobserved heterogeneity between the two groups. The sort of individuals and households that apply are likely to be different (in observed and more importantly unobserved ways) from those that do

9 not apply, and these differences may be associated with the outcomes being evaluated. Nor would it be appropriate to simply compare mean outcomes between the two groups who applied: AE and AI. If the rules of the proxy means test are uniformly applied, the difference between these two groups is clear—the first group was under the cut-off and the second one over it. The latter group, then, is likely to be better off, to the extent that proxy means test does a good job distinguishing levels of economic well-being. This is problematic since a simple comparison would tend to understate the effect of the program if outcomes under study are associated with economic wellbeing as measured by the proxy means score. The methods employed here develop statistical comparison groups within AE and AI. Regression discontinuity techniques do this by comparing households from AE and AI with proxy means scores just below and just above the cut-off score. A method of covariate matching does this by matching eligible applicants who received the program to a group of eligible applicants who did not receive the program during the period under study because their eligibility status had not been verified. As suggested in the Terms of Reference (TOR) for this evaluation, two primary statistical techniques are used to construct a comparison group: regression discontinuity design and a form of matching. These methods develop statistical comparison groups within AE and AI. Both techniques rely on the two key design aspects of the program: the proxy means test targeting model with its eligibility cut-off and the application process.5 In this use of regression discontinuity design (RDD), program impacts are measured as the difference in average outcomes between two sets program applicants: those who scored ―just‖ below the proxy means cut-off point and those who scored ―just‖ above it. Those scoring just above the line will not be beneficiaries, but should be very similar to those who were eligible for two reasons. First, proxy means models do not differentiate well between households with very similar levels of welfare, so that households just above and just below an arbitrary cut-off point may not be substantively different. For this reason, program eligibility for households in the neighborhood of the cut-off score may be considered quasi-random. Second, both types of households applied to the program, eliminating a common source of selectivity bias and suggesting similar unobserved characteristics between the two groups. Because the proxy means score was adhered to very closely in determining program eligibility, the RDD approach provides the strongest method for identify program impacts and is the primary method employed in this report. Details of the RDD and matching methodologies are provided in Ahmed et al. (2006).

3.1.1 The Data

The required data for the quantitative evaluation of the CCT program come from two surveys, as suggested in the TOR for this evaluation: 1. A large, cross-sectional household survey of 2,905 households. This survey was launched in December 2005 and completed in April 2006. The survey included beneficiaries of the education and child health components, and nonbeneficiary- applicants of the CCT program. 2. A two-round panel survey of 750 households selected from eligible households who recently (September 2005 and after) entered the program. The first round of the panel survey was carried out for these 750 households simultaneously with the large household survey. After seven months, a post-intervention panel survey was

5 Households are eligible for the CCT program if their proxy means-tested poverty scores are less than the cut-off point score of 10,249.

10 conducted for the same households to see the impact of the treatment on selected outcomes. The sampling process included random selection of 26 provinces and 52 districts using the probability proportional to size (PPS) method of sampling, based on province- and district-level total number of CCT beneficiary households. Appendix Table A.1 provides the list of selected provinces and districts, and numbers of eligible and non-eligible applicants in those provinces and districts. Appendix Table A.2 shows the distribution of sample households among the selected provinces and districts for the cross-sectional household survey, disaggregated by treatment (education and health beneficiaries) and control (nonbeneficiaries) households. Appendix Table A.3 presents the distribution of sample households for the panel survey. It is worth noting here that the panel survey results have limited use for CCT impact evaluation because of the following constraints: 1. The impact is measured by the change in outcome indicators between the first and the second survey rounds. However, this difference would include all changes in outcome indicators over time, regardless of what causes them. For example, if there were simultaneous non-CCT-related changes over time, then these changes cannot be isolated from the changes caused by the CCT program as there was no comparison group in the panel survey. 2. There was a gap of only 7 months between the first and the second round of the panel survey. This elapsed is not sufficient for measuring the impact of the CCT program on a number of outcomes. The delay in carrying out the first round of the panel survey shortened the elapsed time between the first and the second survey rounds. The reasons for the delay are explained in Section 1.3.2 above.

3.2 First Qualitative Assessment (Kudat et al. 2006)

As is typical of qualitative research, this study looked for paradoxes, contradictions, and fresh perspectives on program impacts on different categories of beneficiaries and those excluded. Qualitative methods enabled us to explore issues more suited to open-ended than closed questions, and those less obvious and unanticipated. It had the benefit of capturing informants‘ perceptions and knowledge in their own words, and enabled us to probe responses, and gather respondents proposed solutions to problems. It also enabled us to consider the significance of local social context. The social analysis these methods enable complements the quantitative focus on levels of program participation and changes in human capital indicators, with an understanding of e.g., the mechanisms through which changes do or do not take place and reasons why; program impacts on gender relations or social relationships between community members; attitudes toward health and education that provide indications of the likely sustainability of program impacts; the way in which people understand program structure and objectives and the effects this has on their practices. By bringing together a number of different types of information, it has been possible to triangulate and show convergence of results, identify complementary findings from these different sources and to examine overlapping and different facets of CCT impacts. The first qualitative assessment contributed to the design of the quantitative research instruments and both qualitative assessments complemented the results of quantitative research by adding breadth and scope to the issues. Methods used in the First Qualitative Assessment:

11 This first attempt at impact evaluation is based on four sets of information: (a) Management Information Systems (MIS) data containing socioeconomic information on all applicants and beneficiaries, as well as on project financing; (b) socioeconomic data obtained from different sources at the provincial level; (c) review of relevant reports and literature; and (d) fieldwork. The fieldwork covered 15 of the 81 provinces in which the program is under implementation. In these provinces, a total of 87 settlements were visited, including province and sub-province (district) centers, urban districts (mahalles), and villages. In some settlements, a number of in-depth (IDI) stakeholder interviews were held with affected households, administrators, education and health service providers. In others, an attempt was made to cover a broader range of stakeholders through IDIs and group discussions. Focus groups were held in four provinces and included province centers, district centers, villages and one town with mothers, primary school children, secondary school children, fathers and education service providers. In total, 557 stakeholders were consulted through semi- structured questionnaires, rapid community assessments, and focus groups. Observational methods were employed in six Foundation offices to estimate the workload and to take an inventory of the issues most frequently experienced by applicants and beneficiaries. Several hundred short discussions were held with beneficiary mothers and children during their visits to the Foundation offices. Appendix Table A.4 presents the geographical coverage and methodology used, and Appendix Table A.5 shows the locations and types of participants of focus group meetings. Appendix Figures B.1 and B.2 show, respectively, a list of key stakeholders and the study model for the First Qualitative Study. While most of the report is based on qualitative interviews conducted through fieldwork, it nevertheless provides critical quantitative information through analysis of Management Information System (MIS) data and analyses of secondary statistical and qualitative information.

3.3 Second Qualitative and Anthropological Study (Adato et al. 2007)

The Second Qualitative and Anthropological Study involved in-depth community and household-level studies, using sociological and anthropological methods, supplemented by key informant interviews and some complementary focus groups. Six localities were studied across three provinces: Diyarbakir, Van, and Samsun. Rural and urban sites were included. The main methods used in this study included: In-depth key informant interviews and focus groups with Foundation staff, health and education service providers, other government officials at district and local level, religious leaders, and others in all localities covered; A set of household case studies using anthropological methods carried out in beneficiary and non-beneficiary households. These case studied involved repeated visits to households and intensive formal and informal interviews with parents, children, and other relatives and neighbors, and participant observation. This research involved in-depth community and household case studies using a range of research methods, carried out in six communities by a research team of sociologist and anthropologists. Three Turkish field researchers lived in program communities over an extended period of time, using ethnographic approaches and qualitative research methods to conduct community and household case studies. Ethnography involves the immersion of the researcher in the everyday life of the people or group being studied, providing detailed descriptions and interpretations, with a focus on the interactions between different aspects of

12 the social system under study. It employs a number of different research methods in combination, including participant observation, in-depth interviews, and informal conversations. Research questions and instruments are found in Appendix C.

3.3.1 Research Components and Methods

The research components of the Second Qualitative and Anthropological Study included the following: Anthropological community and household case studies: Many of the issues undertaken in the Second Qualitative and Anthropological Study were especially suitable for study using anthropological methods; for example, the complex set of factors affecting decisions around schooling and health, the nature of program-related changes in women‘s status in their households; and fertility decisions and their relation to program benefits. The ethnographic case study approach is particularly suited to gaining a more nuanced understanding of the program‘s relationship to beneficiaries and nonbeneficiaries from their point of view. An important element of this case study work is residential fieldwork: researchers live in the study communities for extended periods (in this case, several intervals of several weeks at a time) while they carry out their research. This allows the researcher to establish superior rapport and confidence with informants. This in turn translates into more reliable, candid and deeper data. Topics which might otherwise be difficult to approach become accessible. Multiple visits to study households allow the capture of data at different points in time, and interviewing multiple family members offers a range of perspectives on the program, along both age and gender axes. Carefully noted observations of daily life within households and in the community provide direct information about program effects.

In-depth Household Semi-structured Interviews: In addition to the case studies, during the mid-fieldwork interval, a number of priority issues and some new topics of interest to the study emerged from several sources: discussions with the Client, emerging findings from the quantitative study, and findings from the first phase of fieldwork. It was decided to explore these topics a) with the case study households, and b) with a selection of ‗new households‘ with in-depth semi-structured interviews.

Participant-observation: Observation of activities at the household and community level (participatory and nonparticipatory) was also a key method in the research, allowing the observation of practices, behaviours, and interactions that confirm or contradict what people say, or uncovering issues that people do not mention at all.

Key informant interviews: Key informants with specialized knowledge—including Foundation staff, health, education and other service providers, imams, muhtars, and other government officials were interviewed. Focus Groups: Focus groups were used in the Second Qualitative and Anthropological Study as supplemental to the primary approach of individual interviews, involving group interviews were with some service providers and Foundation staff, either as an alternative to or in addition to individual key informant interviews. At the household level, some informal focus group discussions also took place as part of the household case studies.

3.3.2 Research Design

Site selection: Six sites were selected across three provinces, based on the following criteria: (1) they were included in the survey; (2) they were in provinces with high levels of

13 poverty and those identified by the Client as high priority areas: Eastern Anatolia; Southeastern Anatolia; and Black Sea; (3) provinces were not included in the First Qualitative Assessment, with the exception of Samsun; (4) localities captured geographical and ethnic diversity; (5) localities with a relatively large number of CCT program beneficiaries included in the quantitative survey to reflect high concentrations of poverty; (6) rural and urban diversity. The following six study localities were selected: In Diyarbakir: 5 Nisan and Fatih Mahallesi; In Samsun: Ilyaskoy and Yazilar; in Van: Beyüzümü and YolaĢan. The site selection criteria are found in Appendix Table A.6. Household selection: In each of these communities, beneficiary and non-beneficiary households were selected for case studies and semi-structured interviewing. The 2005-6 survey data was used to select these households, stratifying on the basis of high and low performance on certain health and education indicators. The purpose of this stratification was to gain an understanding of the conditions, practices, events, and perceptions characterizing households with different outcomes on key variables of interest to the research. In all households, we selected those with at least one girl, but wherever possible we selected households with girls and boys, and as many children as possible, particularly those of secondary school age. We selected households with different ethnicities, though in Diyarbakir and Van a high proportion of households were Kurdish. In total 87 households were included in our sample. Of these, we had 41 full household case studies, and 46 households in which we conducted semi-structured interviews with one or more household members. Within these household studies we interviewed 138 adults and 52 children. Additionally, 33 key informants were interviewed, individually or in small focus-groups.

3.3.3 Data Management and Analysis

Data was recorded in notes and tapes, transcribed and translated, and coded in NVivo Qualitative Data Analysis software. Descriptive, pattern and signal coding were used, followed by basic, assay, and Boolean searches. ―Meta matrices‖ were constructed, manually produced matrices/tables that assemble descriptive data into a standardized format where thematic categories can be compared. Although various types of table and matrices were employed in the data analysis for this study, the technique we used most extensively was the case-ordered meta-matrix. Cases (organized by categories-‗sets‘- such as household, community, beneficiary/non-beneficiary status and health/education A simple form of this table was created by testing for the presence of selected descriptive codes among coded household interviews and then exporting the results to Microsoft Excel. Other tables were used based on presence or absence of conditions to generate quantitative data. In the last stage of qualitative analysis, data analysis is dependent on the analytical skills of the research team, triangulating different sources and taking into account all products of the analytical tools.

14 The cross-sectional household survey collected information from the CCT beneficiaries and nonbeneficiary-applicants about the performance of the CCT program (Ahmed et al. 2006). Appendix Tables A.7-A.9 present, respectively, the views of education and health beneficiaries, all beneficiaries about certain common aspects of the program, and nonbeneficiary-applicants. Although the information provided in these tables is self- explanatory, some of the main findings are highlighted below. The CCT program has been fully operational across Turkey since 2004. However, the large household survey results suggest that about 44 percent of CCT-education beneficiaries and 63 percent of health beneficiaries joined the program in 2005. During the survey, the beneficiaries of the CCT program reported that CCT payment delays were common and there have been large fluctuations in payment levels. On average at the time of the survey, CCT education and health beneficiaries, respectively, received about 64 percent and 57 percent of the total amount entitled. Nevertheless, the survey data also suggest that the regularity of CCT payments has been increasing over time. Since CCT payments are conditional on school attendance and regular visits to health clinics by the beneficiaries, furnishing this information by the relevant institutions to the local foundations and entry of these data to the system are needed to verify the maintenance of beneficiary status for the disbursement of payments. Apparently, logistical and administrative difficulties—which probably caused payment delays mainly in the early stage of the program—are being gradually removed over time. The survey indicates that the beneficiaries knew very little about the CCT program. A very large proportion of the CCT beneficiaries and nonbeneficiary-applicants had little or no knowledge of the application and selection criteria. This knowledge is essential in ensuring that the program functions as intended. Social networks and educational institutions played key roles in helping poor families access the CCT program. Schoolteachers were the most common source of information on education benefits. Despite important improvements in a short period of time, there is room for further capacity building to expand CCT coverage, to avoid exclusions and to enhance impacts. The First Qualitative Assessment found that staff shortages, insecurity of the staff with respect to program continuity, and the inefficiencies that result from accessing the web-based system challenge the expansion of program coverage as well as the attainment of full program impacts for current beneficiaries. The CCT staff not only has a heavy workload but also is often asked to assist with other equally pressing workload peaks faced by non-CCT programs carried out by the Foundations. Reducing the workload in high CCT concentration areas, regaining the momentum in capacity building, increasing staff specialization in the Foundations, creating mechanisms for greater decentralization of both the selection and evaluation processes, and expediting solutions to web-based MIS problems would enhance CCT targeting and impacts. The acquisition of software to allow ‗just-in-time‘ reporting of CCT progress and impacts for policy makers also appears as a priority for the next phase of the program. Program impacts are reduced not by program design but by implementation challenges that can be overcome. All three assessments found that payment delays have so far been a regular feature of CCT and, together with other implementation arrangements, cause large

15 fluctuations in payment levels; hence, secondary impacts, such as those pertaining to income and multiplier impacts are thus reduced.

4.1 Sources of Program Information

The household survey found that social networks and educational institutions played key roles in helping poor families access the CCT program. Schoolteachers were the most common source of information on education benefits (Ahmed et al. 2006). The Second Qualitative and Anthropological Study (Adato et al. 2007) found that beneficiaries call upon a variety of sources for information about the program and their benefits. The vast majority of respondents mention teachers/schools, neighbors and relatives, the muhtars and the Foundation offices as sources of information. Other sources of information that are mentioned, but far less frequently, are health centers/personnel, the television, the bank, the post office and shops. Interviews with key informants concur broadly with these results. For example, the health personnel at the Ergani Health Centre said “To be honest, we do not know much about the CCT either. Generally we share what we know with the people. When we feel that we are insufficient, we send people to the Foundation”. The picture which emerges from this data is one in which beneficiaries do not have a single channel for information about the CCT program; because of this, they rely on a variety of different sources. These sources, in turn, are not uniformly informed about the program and are therefore not always in a good position to serve as information channels themselves.

4.2 Knowledge of Benefit Packages

While accepting that these figures are indicative of a tendency and not representative, the Second Qualitative and Anthropological Study found that out of a total of 37 beneficiary households across the three research provinces which spoke about this issue, 30 did not understand the differences by sex, age, or type of benefit (health/education) within their packages. These understandings were somewhat weaker in Diyarbakir and Van than in Samsun. Beneficiaries on the whole understand that the money is given to poor families, and many also recognize that one part of the package is designed to be education support. But very few people know about the health component or about the ways in which education benefits vary depending on sex and age of schoolchildren. Although most beneficiaries are able to cite how much money they have received in recent payments, the majority are unable to say how their benefit packages are calculated (on criteria of number of children, sex of children, school level etc.). Irregular disbursal intervals and frequent breakdowns in the system of informing beneficiaries that their money is available for collection have also resulted in benefit packages that vary depending on how much money is currently ‗owed‘ to the beneficiary. However, the dominant response of beneficiaries to this situation is acceptance without questioning of the entitlement. In some cases, beneficiaries have made enquiries at Foundations, but interview data shows that even this measure has not been very successful because Foundation officials either do not always know themselves how the figures are arrived at, or they are unwilling to say. It is understandable that many beneficiaries are not familiar with the constituent parts of their benefit package—the calculation can be complicated. It may be that there will be limits to their understanding some aspects. However, educating beneficiaries about the structure of their benefits packages would help to make calculation more transparent and remove the air of suspicion which currently surrounds the process.

16 4.3 Understanding of Conditionalities

The Quantitative Assessment (Ahmed et al. 2006) finds that the majority of the beneficiaries do not know the conditions they are required to meet in order to maintain their eligibility. About 90 percent of the education-beneficiaries and 87 percent of the health- beneficiaries claimed that no one informed them of the program rules. Among households in the Second Qualitative and Anthropological Study, more than half said they did not know the conditions. For example, a beneficiary from 5 Nisan noted: I was not aware of the fact that the money is given by obligation (conditionally). I knew that the state gives this money to poor people. I do not have any information about the health aid. I knew that the money is given to afford school expenses of children going to school. I was not aware of the fact that the money is given by obligation (Aysel A.).

Key informants also lament the fact that beneficiaries are not more familiar with the program conditions. As a teacher from the Ergani primary school said: People do not know the fact that this money is delivered with obligation (conditionally). I am sure that if they know that their school success will improve. And children would come to school more often. But the state gives the money. And people think that this money is their right. If we had more information about the program we could help people.

An interesting finding on knowledge of benefits and conditionalities is that beneficiaries are on the whole much more aware of the education component of the program than they are of the health component. Why is this the case? Key informants offered a number of reasons. These include the attitude and knowledge of staff at the health centers vis-à-vis the CCT; the fact that contact with schools that facilitate information transfer is much more frequent than contact with health centers; the common misconception that the health benefit is an ‗immunization aid‘; the less frequent disbursal of health benefits; a better public information campaign for education than for health; the idea that schools and teachers have better accepted and identified with the CCT program, and people‘s generally more attentive attitude to education than to health care. While beneficiaries and key informants are almost all highly enthusiastic about the benefits of the CCT program, emphasizing that the transfers help them to feed their families and send their children to school, key informants are much more uniformly positive about the conditional nature of the benefits than the beneficiaries are. Although all key informants agreed that conditioning the aid on health and educational requirements was a useful step in achieving improvements in human capital indicators, beneficiaries were more divided. Interestingly, there is a strong regional dimension to this finding: while overall, a majority of beneficiary households are supportive and even enthusiastic about the idea of conditionality, almost all the ‗negative‘ responses to conditionality came from Van. Various reasons were offered for these negative opinions: the State has a responsibility to look after everyone; the transfers are too small to demand conditions; the State should not meddle in individuals‘ business.

17 4.4 Strengthen Communications Systems

There is a need to strengthen communications systems so that beneficiaries are better informed of the structure of their benefits package, conditionalities, and when their money arrives at the bank. Successful communication is expected to be especially challenging in a social program targeting the poorest six percent of the population, where levels of education and literacy are low. Beneficiaries are not only poor but also women of a generation where education was even less common for girls than it is now. They also have little experience dealing with formal state institutions, and CCT programs are procedurally complex. Misunderstandings are thus expected, and clearly program staff and others have made an effort to communicate program procedures. However, research findings suggest that it is particularly important to reassess how communications systems can be made more effective in this challenging context. Without effective communication, program impacts will be reduced.

4.5 A Community Liaison

As noted above, CCT programs in many other countries (especially in Latin America) include an institution referred to as a community liaison, or a ‗promoter‘. This person, typically a beneficiary elected to the role by other beneficiaries, plays an important role in program communication. The key areas in which such a promoter works are: 1. Serving as a channel for communication between beneficiaries at the community level and program management structures at a municipal level. 2. In this capacity, answering questions beneficiaries have about the program, initiating enquiries on behalf of the beneficiaries, communicating operational or other problems to program officials at a municipal level. 3. Helping to promote the human capital values of the program among the beneficiaries, encouraging compliance with program requirements. 4. Creating a sense of shared experience and community (‗building social capital‘) among beneficiaries, often through organized community activities such as village clean-up drives. 5. Generally serving as an information and communication resource for both program officials and other beneficiaries. The Second Qualitative and Anthropological Study revealed weaknesses in program beneficiary communications at an early stage of fieldwork. Taking note of the fact that Turkey‘s CCT program does not in its current form include a promoter role, but bearing in mind that the socio-cultural context of Turkey is different than Latin America, we sought the opinions of key stakeholders on the possibility of introducing this institution to the program. In the remaining sections of this chapter we present the results of these research enquiries. Among beneficiaries, opinions about instituting a liaison show great variation across the three provinces. Among the 15 households in Diyarbakir where people responded to the issue, the responses were very positive. In only one household was any doubt about the idea expressed, and in this case it was not a strong opposition. In Samsun, the issue was addressed in 8 households; in three of these, doubts were expressed. In one case, the informant suggested that instituting a promoter role would not help, because women were already so dominated by their husbands. In another, the (male) respondent said that such a system would fail because intra-household relations were so poor in the community. In Van, we have

18 responses from very few households, but of these, none was in any way positive about the idea.6 The beneficiary Nurettin A. said: There is no need for meetings and etc about this benefit. We have lots of work. Women do not go out the house much. What are you going to tell us every month? No one trusts anyone here. There is muhtar and the Foundation. If people want to ask something they would go and ask...There would be tension between people. People don‟t like each other much around here. I don‟t want my daughter or daughters-in- law to go to a meeting. Forget about it. This is another kind of manipulation. Who knows what would be said in these meetings? We know these attempts. You would probably tell girls not to have children, to go to school, to work. You are trying to manipulate us. You have your traditions, we have ours. There is a little amount of money and you are making a big deal about it ( Nurettin A.).

It is interesting to note the reasons for the objections: while some were based on gender relations (men would not be happy about meetings), more were centered on social capital, or the negative nature of it. People cited the poor inter-household social relations in their communities as serious obstacles to implementing such a system. It is not clear whether this is specific to YolaĢan—we did learn that social relations were highly strained in this village. However, there were also tribal hierarchies and hostilities that extend to other settlements where tribal systems are still strong. These would potentially pose difficulties to a community liaison system. This may be a case that points to the value of allowing regional variations in program design. Key informants were also ambivalent about the liaison suggestion. Of the 16 key informants whose opinion was solicited, only 4 responded positively. Several key informants cited the low esteem in which women are held as a major obstacle to such a system; others flagged problems such as the lack of social capital and the dispersed nature of the communities themselves as potential problems. For example, the primary school teacher from 5 Nisan said “I think it is not easy for a woman to inform people. People are ignorant and hard to deal with. Anyway this is a huge region and it is impossible for a woman to inform people.” More positive was the response of the Foundation official from Ilyaskoy, although even in this case, the informant foresaw difficulties: It would be better for us too to have a middleman for communication. We will reach the families that we can‟t reach more easily. The women will have self confidence. But this time the women who are chosen will start to wear away this place every day, I‟m sure. They will feel that they are important; they will think they are something. If the women vote for the representative on their own, they will suffer.

In rural areas of Van it was suggested that teachers could speak to women at schools or in their homes, and men could be informed at the mosque by an imam. In Beyüzümü the Foundation manager suggested a kiosk type of information office at the local level that could provide information as well as distribute and collect forms, though this would be costly.

6 The reason for so few responses is that the fieldworker did not feel comfortable asking the question after the first set of responses from families in YolaĢan who did not want this system and accused her of interfering, and families in Beyüzümü which were uninterested and thought that it was our responsibility to think about these things.

19 Although responses to the concept of instituting a promoter in the CCT program were, at best, ambivalent, informants nonetheless did have suggestions about who might be best qualified to do this job. The majority of respondents agreed that the liaison should be a woman. Other important qualities which were mentioned included literacy, honesty, religion (Muslim), independence, speaking ability, propriety, language (Kurdish in relevant areas) and knowledge of institutions. Opinions about whether the liaison should be a local person were divided: some informants suggested that a local woman would be best because of her knowledge of the community, while others suggested that a local person would only incite jealousy and competition. Responses to questions about payment were more limited, but the prevalent belief was that such a person would have to be paid. Opinion was also divided about whether it would be feasible to hold meetings. While some respondents across the study communities suggested that collective meetings would be the best approach, held in schools or other public places, others said that meetings would not be possible: their husbands or mothers-in-law would not allow them to attend, or else their household responsibilities would make such meetings impossible. These respondents said that they would prefer to meet with the promoter in their own homes. In terms of the information provided in meetings, beneficiaries‘ concerns centered around the resolution of problems and uncertainties: where was their money? How much should they receive? When could they collect it? Key informants took a broader view, suggesting that a promoter should educate beneficiaries about program goals and the conditional nature of the program and health and nutrition issues. We should note that in CCT programs where the liaison position exists, she is responsible for both these aspects of communication: broad program goals and narrower (but also important) day-to-day pragmatic issues and concerns. Both key informants and beneficiaries were keen on the idea of inviting health and education personnel to liaison meetings.

4.6 Summary of Constraints on Program Implementation

In evaluating program implementation, we found the following constraints on program implementation, and combine to explain many of the outcomes reported in this section, and elsewhere in this report: 1. The complexity of program administration that is inherent in a CCT program, and the learning curve that requires the accumulation of experience. 2. Staff shortages at the Foundations. 3. The low level of education and literacy of program beneficiaries, the little experience they have had with state institutions, and the procedural complexity of a CCT program, all of which combine to make it difficult for program and foundation staff to effectively communicate about program structure and procedures. As noted in section 5.1 below, 56 percent of education beneficiaries and 58 percent of health beneficiaries never attended school and 19 percent of househeads of the beneficiary families are illiterate. 4. Cultural biases in beneficiary communities that influence responses to education and health services (see sections 8 and 9). 5. Lack of ‗ownership‘ by the Health and Education departments such that the necessary full commitment to implementation partnerships would be forthcoming. CCTs are inherently intersectoral and require intersectoral cooperation at the highest levels.

20 Most of these constraints can be overcome through dedication of program staff and political commitment to making the program work well—including the ability to learn lessons from experience and evaluation findings. Thus far, we have encountered an open attitude from SRMP officials and staff, and genuine interest in learning and responding in order to improve the program and better serve beneficiaries.

21

Using the large household survey data collected for the CCT evaluation, this section provides profiles of CCT-beneficiary households and compares the profiles with those of the control group, which is selected from households who applied for the program but were not accepted.7 The content in this section is excerpted from the Quantitative Assessment (Ahmed et al. 2006).

5.1 Household Characteristics

Based on the large household survey data (Ahmed et al. 2006), Table 5.1 presents the characteristics of CCT-education-beneficiary households, health-beneficiary households, and nonbeneficiary-applicant households.8 Table 5.1 — Characteristics of survey households Education Health Nonbeneficiary- beneficiaries Beneficiaries applicants Household size (person) 6.6 6.5 5.2 Years of schooling, male household head 3.9 4.2 4.9 Years of schooling, wife of household head 1.8 1.9 2.8 Years of schooling of adult male aged 15 and above 5.2 4.7 5.6 Years of schooling of adult female aged 15 and above 3.0 2.3 3.4 No schooling adult (18 years and above) male (percent) 20.7 21.2 16.7 No schooling adult (18 years and above) female (percent) 55.6 58.1 44.3 Female-headed household (percent) 7.2 3.5 5.9 Per capita monthly expenditure (YTL) 85.9 74.8 104.0 Percent of households with per capita expenditure less than $1 a daya 7.9 12.7 5.5 Percent of households with per capita expenditure less than $2 a daya 55.3 63.2 42.0

Principal occupation of household head Wage laborer 34.1 44.3 35.1 Salaried worker 2.1 2.3 3.3 Self-employed in agriculture 3.9 3.7 2.8 Self-employed in non-agriculture 8.4 7.5 8.3 Seasonal/ Temporary worker 17.5 17.9 16.9 Other 7.6 4.2 8.4 Unemployed 26.3 20.2 25.3 Source: Based on data from ―Evaluating the Conditional Cash Transfer Program in Turkey: 2005-06 Household Survey‖. a1 US dollar =1.3 YTL

7 The household survey icluded a total of 2,905 households, of which program beneficiary households accounted for 54.5 percent. However, during the analysis of the household survey data, it was found that 12.5 percent of the nonbeneficiary- applicant households were eligible for the CCT program (i.e., their proxy means-tested poverty scores are less than the cut- off point score of 10,249), but their selection into the program was pending the verification of their eligibility status. Possibly some of these households became program beneficiaries during the course of the survey. Therefore, these households have been excluded from the analysis in order to avoid potentially biassed results. 8 The household survey included a total of 1,583 beneficiary households, of which education beneficiaries account for 68.5 percent and health beneficaries, 10.5 percent. The remaining 21.0 percent of the households are benefciaries of both education and health components of the CCT program because they have preschool as well as school-age children.

22

While household size is about the same for education beneficiaries (6.6 persons) and health beneficiaries (6.5 persons), nonbeneficiary-applicant households have relatively smaller household size (5.2 persons). Adult educational attainment is low in general and extremely low for women. Exceptionally high proportions of adult female members of households (aged 18 and above) —56 percent for education beneficiaries, 58 percent for health beneficiaries, and 44 percent for nonbeneficiary-applicants—never attended school. Most households in the sample are poor. Per capita consumption expenditures are less than two US dollars a day for 55 percent of the education-beneficiary households, 63 percent of health-beneficiary households, and 42 percent of nonbeneficiary-applicant households. Average consumption expenditures per capita are considerably lower for beneficiary households than those for nonbeneficiary households. For both beneficiary and nonbeneficiary household heads, wage labor is the major occupation, followed by seasonal or temporary work. Heads of one-fifth of health-beneficiary households and about 26 percent of education-beneficiary and nonbeneficiary-applicant households are unemployed. Table 5.2 shows the literacy rate and migration status of heads of sample households. A higher proportion of CCT-beneficiary household-heads are illiterate (19 percent of both education and health beneficiaries) compared to nonbeneficiary-applicant households (12 percent). While the majority of the household-heads are native of the village or town they currently live in, the heads of 28 percent of education-beneficiary households, 23 percent of both health-beneficiary and nonbeneficiary-applicant households, reported that they migrated from other provinces. Table 5.2 — Literacy and migration status of household head Education Health Nonbeneficiary- beneficiaries beneficiaries applicants (percent) Literacy Does the household head know how to read and write? Can read only 1.8 2.1 1.5 Can sign name only 8.8 6.8 6.2 Can read and sign name 4.6 4.6 3.6 Can read and write 65.6 67.5 77.0 Can neither read nor write 19.2 19.0 11.8

Native/migrant Household head is native of the town/village 60.6 64.5 66.6 Household head is not native, migrated from other place in the same province 11.1 12.3 9.6 Household head is not native, migrated from other place in different province 27.5 23.1 23.4 Household head is not native, migrated from other country 0.9 0.0 0.4 Source: Based on data from ―Evaluating the Conditional Cash Transfer Program in Turkey: 2005-06 Household Survey.‖

Table 5.3 presents the age distribution of the survey households by gender and age groups. The patterns of age distribution are quite different among the three groups of households: CCT-education-beneficiary households, health-beneficiary households, and nonbeneficiary-applicant households. As expected, the percentages of school-age population (children aged 6 to 18 years) are higher in education-beneficiary households compared to the

23

other two groups of households. Also as expected, health-beneficiary households have higher proportions of children aged 5 years and below than education-beneficiary and nonbeneficiary-applicant households. Table 5.3—Percentage distribution of population by gender and age group Age in years 60 and 0-5 6-10 11-14 15-18 19-24 25-34 35-44 45-59 above Total (percent of total population) CCT education beneficiary households Male 13.8 20.1 16.7 11.9 6.7 7.4 14.6 7.3 1.4 100.0 Female 12.3 19.0 16.5 11.7 6.4 11.7 14.5 5.7 2.3 100.0 Both sex 13.1 19.5 16.6 11.8 6.6 9.5 14.6 6.5 1.8 100.0 CCT health beneficiary households Male 28.3 19.4 10.9 6.7 4.2 14.7 10.9 3.6 1.2 100.0 Female 27.9 19.1 9.2 7.0 6.6 16.4 9.0 2.9 1.8 100.0 Both sex 28.1 19.3 10.1 6.9 5.4 15.5 9.9 3.3 1.5 100.0 Non-beneficiary households Male 17.2 16.2 12.3 8.9 7.5 15.9 13.4 6.7 2.0 100.0 Female 17.5 16.3 10.4 7.3 10.4 17.8 12.2 5.7 2.3 100.0 Both sex 17.4 16.3 11.4 8.1 8.9 16.8 12.8 6.2 2.2 100.0

Source: Based on data from ―Evaluating the Conditional Cash Transfer Program in Turkey: 2005-06 Household Survey.‖

5.1.1 Regional Variations in Household Characteristics of CCT Beneficiaries9

At the outset, it is important to note that the household survey sample is statistically representative of the CCT program only at the national level, and not at the regional levels. Therefore, the disaggregated analysis of the characteristics of survey households living in each of the seven regions of Turkey is only indicative of the situation in respective regions, and statistically valid comparisons of the findings between regions may not be made from this analysis. Table 5.4 presents the characteristics of education-beneficiary households living in different regions of the country. Household size varies from 5.5 persons in the Black Sea region to 7.1 persons in . Among all seven regions, education- beneficiaries living in Marmara region are the poorest (12 percent of the households have per capita consumption expenditures less than US dollar one a day) and have the lowest enrollment rates in primary (86 percent) and secondary (33 percent) schools. In contrast, those living in the Black Sea region seem to be the wealthiest and have the highest enrollment rates in primary (96 percent) and secondary (70 percent) schools. The rate of illiteracy of household head varies widely—from only 6 percent in the Black Sea region to 21 percent in Marmara region and as high as 32 percent in Eastern Anatolia region. About 70 percent of adult female members of households (aged 18 and above) in Eastern Anatolia region and 62 percent in Marmara region never attended school, while this rate is about 23 percent in the Black Sea region. The unemployment rate of household head ranges from 17 percent in the Black Sea region to 32 percent in Aegean and Southeastern Anatolia regions. While almost three-fourths of the household-heads in the Black Sea region are native of the village or town they currently live in, the heads of 73 percent of households in Aegean region are migrants from other provinces.

9 Findings in this section are in addition to those presented in Ahmed et al. (2006).

24

Table 5.4— Characteristics of CCT education beneficiary households, by regions Regions Southea stern Eastern Central Mediterran Descriptions Anatolia Anatolia Black Sea Anatolia ean Marmara Aegean All Household size (person) 7.0 7.1 5.5 5.6 6.3 6.2 7.0 6.6 Primary-school-age children (6-14 years) who go to primary school (percent of all households with primary-school-age children) 92.3 88.7 95.7 93.2 92.4 86.3 87.2 90.8 Secondary-school-age children (14-17 years) who go to secondary school (percent of all households with secondary- school-age children) 45.2 50.5 69.8 56.6 43.0 33.3 37.9 47.8 Years of schooling, male household head 4.1 3.3 4.9 4.5 4.1 3.7 3.5 3.9 Years of schooling, wife of household head 1.3 1.2 3.9 3.1 2.0 1.5 1.6 1.8 Years of schooling of adult male aged 15 and above 5.4 4.8 5.7 5.4 5.1 4.6 4.9 5.2 Years of schooling of adult female aged 15 and above 2.8 2.2 5.0 4.3 3.1 2.4 2.9 3.0 No schooling adult (18 years and above) male (percent) 18.8 31.9 7.3 7.8 20.2 23.0 20.5 20.7 No schooling adult (18 years and above) female (percent) 60.1 69.8 22.5 29.1 53.8 62.1 58.7 55.6 Household head is illiterate (percent) 16.8 32.0 6.4 13.6 14.6 21.2 15.0 19.2 Household head is native of the town/village currently residing 68.0 69.5 73.4 51.5 50.9 50.6 26.5 60.6 Female-headed household (percent) 7.6 7.1 4.6 7.2 5.8 12.9 6.0 7.2 Per capita monthly expenditure (YTL) 85.2 75.8 111.1 98.2 78.2 78.7 94.8 85.9 Percent of households with per capita expenditure less than $1 a day 8.3 7.7 4.6 6.6 7.0 11.9 10.1 7.9 Percent of households with per capita expenditure less than $2 a day 59.1 63.0 23.9 41.6 64.3 64.3 47.5 55.3

Principal occupation of household head Wage laborer 39.2 39.1 23.9 13.3 38.8 34.1 33.0 34.0 Salaried worker 1.2 2.4 3.7 4.2 1.8 2.4 0.0 2.1 Self-employed in agriculture 1.0 9.2 7.3 1.8 3.5 1.2 1.0 3.9 Self-employed in non-agriculture 5.1 8.6 20.2 4.2 12.9 7.1 9.0 8.4 Seasonal/ Temporary worker 13.2 10.7 23.9 38.6 13.5 23.5 20.0 17.7 Other 8.1 8.6 3.7 8.4 5.9 10.6 5.0 7.6 Unemployed 32.1 21.6 17.4 29.5 23.5 21.2 32.0 26.3 Source: Based on data from ―Evaluating the Conditional Cash Transfer Program in Turkey: 2005-06 Household Survey.‖ 1 US dollar =1.3 YTL.

25

Table 5.5— Characteristics of CCT health beneficiary households, by regions Regions Southeaste rn Eastern Central Mediterra Descriptions Anatolia Anatolia Black Sea Anatolia nean Marmara Aegean All Household size (person) 6.6 7.2 5.7 5.6 5.8 5.3 7.2 6.5 Years of schooling, male household head 5.0 3.5 4.7 5.1 4.5 4.3 3.6 4.2 Years of schooling, wife of household head 1.6 1.3 3.7 3.1 2.7 2.1 1.9 1.9 Years of schooling of adult male aged 15 and above 5.5 4.3 4.8 5.2 4.4 4.4 4.5 4.7 Years of schooling of adult female aged 15 and above 2.6 1.6 4.0 3.6 2.8 2.3 2.3 2.3 No schooling adult (18 years and above) male (percent) 13.0 32.0 11.5 10.0 16.3 11.6 23.9 21.2 No schooling adult (18 years and above) female (percent) 57.7 74.3 29.6 33.8 42.0 46.7 60.4 58.1 Household head is illiterate (percent) 12.0 32.4 9.1 6.3 9.3 12.5 16.2 19.0 Household head is native of the town/village currently residing 73.9 62.8 86.4 66.7 58.1 85.0 17.1 64.5 Female-headed household (percent) 3.2 4.3 0.0 3.2 2.3 5.0 2.7 3.5 Per capita monthly expenditure (YTL) 75.9 72.0 96.7 84.3 74.1 60.8 73.4 74.8 Percent of households with per capita expenditure less than $1 a day 11.8 11.4 9.1 12.9 7.0 22.5 18.9 12.7 Percent of households with per capita expenditure less than $2 a day 62.4 67.4 36.4 50.0 60.5 82.5 64.9 63.2

Principal occupation of household head Wage laborer 50.5 54.3 31.8 11.3 46.5 42.5 40.5 44.3 Salaried worker 2.2 3.8 0.0 3.2 0.0 0.0 0.0 2.3 Self-employed in agriculture 1.1 5.4 4.5 1.6 4.7 5.0 2.7 3.7 Self-employed in non-agriculture 5.4 9.2 13.6 3.2 18.6 2.5 0.0 7.5 Seasonal/ Temporary worker 6.5 10.3 18.2 53.2 2.3 30.0 29.7 17.9 Other 2.2 5.4 4.5 6.5 2.3 2.5 2.7 4.2 Unemployed 32.3 11.4 27.3 21.0 25.6 17.5 24.3 20.2 Source: Based on data from ―Evaluating the Conditional Cash Transfer Program in Turkey: 2005-06 Household Survey.‖ 1 US dollar =1.3 YTL.

26

Table 5.5 shows the characteristics of health-beneficiary households by regions. A comparison of per capita expenditures between Tables 5.4 and 5.5 indicates that health- beneficiary households are poorer than education-beneficiary households in each of the seven regions. As in the case of education-beneficiaries, health-beneficiaries living in Marmara region are the poorest and those living in the Black Sea region are the richest. Only 17 percent of the health-beneficiary household-heads in Aegean region are native of the village or town they currently live in, compared to 86 percent in the Black Sea region. The rate of illiteracy of household head ranges from 6 percent in to 32 percent in Eastern Anatolia region. While 74 percent of adult female members of households in Eastern Anatolia region never attended school, this rate is about 30 percent in the Black Sea region.

5.2 Consumption Patterns

The differences between the consumption patterns of CCT beneficiary and nonbeneficiary-applicant households indicate that nonbeneficiary households are economically better-off than beneficiary households. Tables 5.6 and 5.7, respectively, present food and nonfood budget shares of CCT beneficiary and nonbeneficiary-applicant households. Nonbeneficiary-applicant households spend 21 percent more on food and nonfood consumption than CCT beneficiary households. Although beneficiary households spend almost the same amount of money on food as do nonbeneficiary-applicants; in relative terms, expenditure on food accounts for 46 percent of total expenditures for beneficiaries and 39 percent for nonbeneficiary-applicants. Patterns of food expenditure are similar for the two groups of households. For both groups, bread accounts for over one-fourth of total food expenditures, followed by milk and milk products, and meats (Table 5.6). Table 5.6 — Food budget share CCT Beneficiary Nonbeneficiary- households Applicant households Monthly per capita total expenditure (YTL) 85.6 104.0 Monthly per capita food expenditure (YTL) 39.1 40.5 Share of food in total expenditure (percent) 45.6 39.0

Share of food budget: (percent) Bread 28.8 28.0 Other cereal 9.7 7.5 Pulses 3.5 3.7 Oils 5.6 5.8 Vegetables 6.0 6.4 Fruits 5.8 6.4 Fish 0.3 0.4 Meats 9.5 9.2 Eggs 1.5 1.8 Milk and milk products 10.2 10.6 Spices 1.0 0.9 Sweets 7.3 7.1 Beverages 6.2 7.0 Other prepared food 4.6 5.2 Total 100.0 100.0 Source: Based on data from ―Evaluating the Conditional Cash Transfer Program in Turkey: 2005-06 Household Survey.‖

27

Nonbeneficiary-applicant households spend 37 percent more on nonfood items than beneficiary households do. The patterns of nonfood consumption are also somewhat different between the two groups. For instance, nonbeneficiary households allocate substantially higher proportions of their budget to meet health and communication expenses (Table 5.7).

Table 5.7 — Nonfood budget share Nonbeneficiary- CCT Beneficiary applicant households households Monthly per capita total expenditure (YTL) 85.6 104.0 Monthly per capita nonfood expenditure (YTL) 46.5 63.5 Share of total nonfood expenditure (percent) 54.4 54.5

Budget share of nonfood items: (percent) Clothing and footwear 6.6 5.9 Housing 8.8 10.7 Health 4.8 9.1 Education 12.4 9.0 Transport 5.9 4.7 Personal care and cleaning 7.5 6.7 Communication 5.8 10.2 Tobacco 10.3 11.5 Furniture and appliances 2.4 1.5 Utilities 11.9 10.9 Family events 1.7 1.7 Fuel 20.0 17.6 Other 2.0 0.5 Total 100.0 100.0 Source: Based on data from ―Evaluating the Conditional Cash Transfer Program in Turkey: 2005-06 Household Survey.‖

Table 5.8 presents per capita calorie consumption, calorie shares of food items, and costs of calories by food sources. On average, CCT beneficiary households and nonbeneficiary- applicant households consume almost the same amount of calories—a little over 2,800 kilocalories (kcal) per capita per day. For both beneficiary and nonbeneficiary applicant groups of households, bread accounts for about one-half of total calorie consumption, a much higher figure than bread‘s share of the food budget. This indicates that bread is a relatively inexpensive source of energy. Indeed, bread is the least expensive source of calories (as demonstrated in the bottom section of Table 5.8), closely followed by other cereals. With the exception of beverages, which contribute a negligible amount of calories to the diet, meat is the most expensive source calories, about 25 times as expensive as bread as a source of energy, for both groups of households. Using the household survey data, the relationship between calorie consumption and income is estimated for the entire sample of households in a multivariate framework of analysis. The estimated calorie consumption elasticity with respect to household consumption expenditure (as a proxy for income) is 0.32. This implies that a 10 percent increase in per capita household expenditure increases calorie consumption by 3.2 percent for the survey households.

28

Table 5.8 — Calorie consumption, shares, and costs CCT beneficiary Nonbeneficiary-applicant Item households households Per capita calorie consumption per day (kcal) 2,829 2,819

Calorie shares (percent) Bread 51.3 49.8 Other cereals 16.0 13.1 Pulses 3.6 3.9 edible oils 9.0 10.4 Vegetables 2.2 2.6 Fruits 1.7 2.0 Fish 0.1 0.1 Meats 0.7 0.7 Eggs 0.5 0.7 Milk and milk products 4.2 4.9 Spices 0.1 0.2 Sugar and sweets 6.3 6.6 Beverages 0.4 0.5 Other foods 4.0 4.6 Total 100.0 100.0

Calorie costs by food sources (YTL per 1,000 kcal) Bread 0.25 0.25 Other cereals 0.27 0.26 Pulses 0.43 0.43 Edible oils 0.27 0.26 Vegetables 1.22 1.14 Fruits 1.47 1.45 Fish 2.86 2.70 Meats 6.37 6.27 Eggs 1.34 1.14 Milk and milk products 1.07 0.99 Spices 1.53 0.69 Sugar and sweets 0.51 0.49 Beverages 7.27 7.07 Other foods 0.51 0.52 Source: Based on data from ―Evaluating the Conditional Cash Transfer Program in Turkey: 2005-06 Household Survey.‖

The household survey data suggest that average per capita income transfers from the CCT program account for 6.2 percent of per capita household expenditure of beneficiary households. Thus, with the estimated calorie-income elasticity of 0.32, CCT is expected to increase calorie consumption of its beneficiaries by 2 percent. Since per capita expenditure of nonbeneficiary-applicant households is 21 percent higher than that of the beneficiary households, calorie consumption of nonbeneficiary households should have been 6.7 percent higher. However, beneficiary and nonbeneficiary households consume virtually the same amount of calories (Table 5.8). These results indicate that the beneficiary households consume disproportionately more calories than what CCT income transfers are likely to contribute. The First Qualitative study (Kudat et al. 2006) research conducted to assess the impact of the CCT program supports this finding. Mothers in beneficiary households reported that the CCT money enabled them to buy more and better quality food, which, they believed, improved children‘s nutritional status and their school performance.

29

5.3 Assets and Amenities

The pattern of asset ownership of education-beneficiary households is quite similar to that of nonbeneficiary-applicant households. However, health-beneficiary households have a lower level of asset ownership status than the other two groups of households. Table 5.9 presents the ownership of some selected assets by CCT education- and health- beneficiary households and nonbeneficiary-applicant households. Ownership of a television is most common among all selected assets—96 percent of education-beneficiary households, 93 percent of health-beneficiary households, and 95 percent of nonbeneficiary-applicant households reported that they had television. Ownership of a refrigerator is also widespread among the three household groups—89 percent of education-beneficiary and nonbeneficiary households, and 80 percent of health-beneficiary households own a refrigerator. About one- half of education-beneficiary and nonbeneficiary-applicant households, and one-third of health-beneficiary households have land-line telephones; while 49 percent of education- beneficiary and nonbeneficiary-applicant households and 40 percent of health-beneficiary households have mobile phones. Table 5.9 — Selected household asset ownership Education beneficiary Health beneficiary Nonbeneficiary-applicant Assets households households households (percent of households) Land line phone 51.4 34.1 52.8 Mobile phone 48.9 40.3 48.8 Radio 20.0 11.4 21.5 Television 96.0 92.8 95.0 Satellite connection 7.1 7.2 8.0 VCR 12.0 10.0 11.8 Refrigerator 88.7 80.1 88.7 Dishwasher 1.0 0.2 1.1 Washing machine 57.9 39.8 59.8 Gas stove 71.7 73.5 75.9 Electric oven 15.8 10.8 18.3 Vacuum cleaner 27.5 22.7 33.6 Bike 5.1 2.1 4.7 Motor bike 1.2 1.3 1.3 Water heater 1.2 0.0 1.3

Source: Based on data from ―Evaluating the Conditional Cash Transfer Program in Turkey: 2005-06 Household Survey.‖

Table 5.10 shows some of the dwelling facilities of sample households. Virtually all households have electricity, and 92-96 percent of the households have access to piped drinking water. On the other hand, central heating is almost nonexistent—only 1 percent of nonbeneficiary-applicants and 0.3 percent of education-beneficiaries reported that they had central heating. Coal is the most common fuel for heating, used by 66-68 percent of sample households. For cooking fuel, 85 percent of education-beneficiaries, 77 percent of health- beneficiaries, and 87 percent of nonbeneficiary-applicants use liquefied petroleum gas (LPG).

5.4 Labor Force Participation

Table 5.11 presents the labor force participation and employment status by age-group and gender for the sample households. The labor force consists of everyone above the age of 15 who is employed or unemployed but actively seeking employment. People not counted in

30

the labor force include students, housewives, retired people, disabled people, as well as discouraged workers who are no seeking work.

Table 5.10 — Selected dwelling facilities Education Health Nonbeneficiary- beneficiary beneficiary applicant households households households (percent) In-house bathroom 75.0 69.0 81.4 In-house toilet (flush or no flush) 65.0 50.6 72.2 Central heating 0.3 0.0 0.9 Access to piped drinking water (inside or outside house) 94.7 91.8 95.9 Electricity 99.2 99.2 99.5

Heating fuel used: Dried dung 9.5 14.3 6.7 Firewood 17.4 15.3 18.3 Coal 65.9 66.0 68.0 Natural gas 0.1 0.0 0.1 LPG 0.1 0.2 0.3 Electricity 0.5 0.6 0.7 Other 6.4 3.6 5.9

Cooking fuel used: Dried dung 5.1 8.4 3.6 Firewood 3.7 6.5 4.7 Coal 3.6 5.3 2.9 Natural gas 0.4 0.2 0.6 Gas oil 0.1 0.2 0.2 LPG 84.8 76.6 86.6 Electricity 0.7 0.4 0.2 Other 1.6 2.3 1.2 Source: Based on data from ―Evaluating the Conditional Cash Transfer Program in Turkey: 2005-06 Household Survey.‖

Table 5.11 — Labor force participation, population age 15 years and above CCT beneficiary households Nonbeneficiary-households Male Female All Male Female All (percent)

In the labor force 81.4 9.2 44.5 81.1 6.0 42.5 Wage labor 43.8 49.5 44.4 45.5 45.3 45.5 Salaried 1.2 2.3 1.3 2.0 2.8 2.0 Self employed (non-agriculture) 6.9 3.7 6.6 8.3 3.8 8.0 Farming 1.9 0.0 1.7 1.3 0.0 1.2 Work without pay 0.7 0.9 0.7 0.4 0.0 0.3 Unemployed (looking for job) 45.6 43.5 45.3 42.6 48.1 43.0 Source: Based on data from ―Evaluating the Conditional Cash Transfer Program in Turkey: 2005-06 Household Survey.‖

For all household members aged 15 and above, the labor force participation rates are 44.5 percent and 42.5 percent, respectively, for CCT-beneficiary and nonbeneficiary- applicant households. However, there is a huge difference in labor force participation rates between males and females. Among beneficiary household members aged 15 and above, 81.4 percent of men and only 9.2 percent of women are in the labor force. For nonbeneficiary- applicant households, the rates are 81.1 percent for men and only 6.0 percent for women.

31

The unemployment rates (calculated as those reporting they were unemployed and looking for work, divided by the labor force) are very high—45.3 percent of beneficiary and 43.0 percent of nonbeneficiary-applicant household members in the labor force are unemployed. Wage labor is by far the most important category of employment. Among the employed members (aged 15 and above) of beneficiary and nonbeneficiary-applicant households, 81.2 percent and 79.8 percent are wage laborers, respectively.

5.5 Private Transfers and Remittances

Table 5.12 shows that about 4 percent of education-beneficiary households, 3 percent of health-beneficiary households, and 5 percent of nonbeneficiary-applicant households received private assistance from within Turkey. The incidence of receiving remittances from abroad is negligible—less than 1 percent of the sample households reported receiving remittances from abroad.

Table 5.12 — Private transfers and remittances received Education Health Nonbeneficiary- beneficiary beneficiary applicant households households households Remittance from abroad (percent of all households) 0.9 0.2 0.6 Transfers from within Turkey (percent of all households) 3.6 3.1 4.7 Total amount received – average for those who received (YTL/household/year) 623 200 621 Total amount received – average for those who received and did not receive (YTL/household/year) 28 6 32 Source: Based on data from ―Evaluating the Conditional Cash Transfer Program in Turkey: 2005-06 Household Survey.‖

For those who received private assistance, the average amounts received per household per year were YTL 623 for education-beneficiaries, YTL 200 for health-beneficiaries, and YTL 621 for nonbeneficiary-applicants.

32

A large number of public and private sector programs support human resources development in Turkey and complement CCT with respect to its education and health objectives. Because of the relatively recent implementation of CCT nationwide, many of the poorest families dealt with the 2001 economic crisis with support from a variety of other sources. This section first provides the evidence of complementary education and health support provided by various entities, as reported in the First Qualitative Study (Kudat et al. 2006). The section then presents the incidence of participation of CCT beneficiaries and nonbeneficiary-applicant households in other social assistance programs, based on the large household survey data (Ahmed et al. 2006).

6.1 Complementary Support for Education

The First Qualitative Study (Kudat et al. 2006) states that, CCT support for education largely overlapped with support from other sectors with respect to the specific target population. The national mobilization in 2001/02 within the education sector to meet the goal of increasing minimum education to eight years for all provided the sector with substantial additional financing. Moreover, the military, the municipal sector, the private sector, civil society organizations as well as individual citizens provided support to universal education and much of this support was directed at poor communities and poor families. As a result, CCT became one of the many different types of support for education in the period 2003-05. The SYDVs themselves provide either in kind or cash support to poor students at the outset of each school year. The Ministry of Education (MOE) provides free schoolbooks for grades 1-8. Poor students in boarding schools have all their expenses, including meals met, but cannot benefit from CCT. A large number of students are provided with full scholarships from MOE or from organizations such as the Turkish Education Foundation. Again, those who receive scholarships from state institutions cannot benefit from CCT. NGOs, such as Chose Your Own Sibling match needy students with the wealthier one to provide direct assistance for education. Nation-wide organizations such as Mehmetcik (Soldier) Foundation and Deniz Feneri (Lighthouse) also provide similar support. Specific information includes the following: In 2003, MOE and UNICEF launched a campaign10 to send 600,000 girls to school.11 By October 2005, over 150,000 girls were registered in school as a result of this campaign.12 SYDVs have been providing one-time support to poor school children.13 This support has been given either as cash or in-kind.14 In addition to their routine annual education support, Foundations also consider individual applications from the poor for various types of support, such as financing travel to competitive university entrance examinations, etc.

10 Haydi Kizler Okula Kampanyasi (All Girls to School Campaign). 11 In September 6, 2006 (Radikal Newspaper, p.3), MOE/UNICEF initiated a bicycle competition for each competitor to donate 100YTL to send girls to school. 12 Radikal Newspaper. Kampanyalarin En Guzeli (The Most Beautiful of all Campaigns). 13 October 2005: 9. 13 In many case, if a village is poor, the education support is given to all students rather than singling out the most needy. 14 In most cases, children are given uniforms, shoes or other clothing items. The Foundations also serve as recipients of charity for schooling and pool citizen contributions or NGO contributions together in their pre-school year assistance.

33

Turkish Education Foundation (TEV) and Turkish Education Association (TED) provide scholarships to poor and bright students. Individual initiatives established campaigns. For example, ―choose your sister/brother‖ initiative over internet reached 1,487 schools in 69 provinces. 28,000 students found their sisters/brothers and provided them with books, stationary, school uniforms, etc., and started to send letters to each other.15 The governors and municipalities in each province have funds to assist vulnerable families. For instance, governors, with support from Foundations, place poor high school children in boarding facilities. When such facilities do not exist, they rent flats and place these children in private housing. They provide financial support to poor children to attend private university preparation courses. They also provide scholarships. For the past 5 years, The Association in Support of Modern Living (Cagdas Yasami Destekleme Dernegi- CYDD) and Turkcell collaborate on Kardelenler Project in 41 provinces have offered scholarships to 10,000 girls. Milliyet, a major newspaper, initiated a recent campaign16 and collected about $18 million in 5 months to provide 4,012 girls scholarships. Modern Boarding High School Girls Project received support from Schneider Electric Company to facilitate poor girls to continue their education. ―I have a Daughter in Anatolia who will become a Teacher‖ project has provided nearly 500 scholarships for girls. Mercedes Benz initiated a project, ―Each of our Daughters is a Start,‖ which provided 200 scholarships so far and Schneider Electric has given 50 scholarships to girls in technical high schools. Ericsson‘s Girls of the Knowledge Society project has provided 60 scholarships. All these recent private sector initiatives are expected to continue and grow. Municipal Governments have been important contributors to schooling of the poor. Indeed, national and local newspapers report regularly on support to young girls with a commitment to attend school, especially high school, but who lack access to boarding.17 Chambers of Industry and Trade, Chambers of Small Traders and Artisans, which have their provincial and district organization all over Turkey, are also outstanding organizations in providing education support.

The citizen contribution to education has been even more remarkable. Throughout the country, wealthier citizens have built schools and boarding facilities. Charity for poor students has accelerated.18 All of these efforts, many of which started prior to CCT implementation, complemented and enhanced the cash incentives provided by CCT.

15 www.kardesinisec.com 16 Baba Beni Okula Gonder Kampanyası (―Father sent me to School‖ Campaign). 17 Gaziantep Municipality offered to meet boarding expenses of village girls eager to attend high school in the city. It reported that large numbers of citizens came forward for help. (Radikal. 9 October 2005, p. 5). 18 For instance, a new citizen initiative called ―education one girl,‖ is arranging families to sponsor the education of one girl by providing 50 YTL a month.

34

6.2 Complementary Health Support

The First Qualitative Study (Kudat et al. 2006) suggests that, in the health sector, the single most important poverty program concerned the issuance of green cards to the poor for free treatment as well as preventive health advice on such special occasions as TB week, Cancer week, etc. In addition, special campaigns are launched, as needed, on such topics as dental care. Nurses and midwives also provide advice on hygiene and related topics. The restructuring of the sector and investments in health infrastructure have allowed the poor to seek treatment somewhat more easily. In addition, the state provides free immunization. Mobile health teams visit villages and vicinities all over the country and immunize children. However, rural women especially are in need to access health units and to benefit from preventive health services as well as clinic controls. Therefore, CCT has become the single most important initiative for preventive health care dealing with the poor. Provincial administrators, the Foundations, and the health sector providers pointed out that health centers regularly provided immunization and other types of preventive health support prior to and during CCT implementation to all citizens, regardless of their poverty status. Nevertheless, they recognized the added contributions of CCT in attracting the poor to health centers on a more regular basis.

6.3 Participation in Other Social Assistance Programs

The Quantitative Assessment (Ahmed et al. 2006) based on the large household survey data shows that, in addition to CCTs, many respondents reported that they received assistance from the SYDVs and other agencies of the government. Some households also received assistance from nongovernmental organizations (NGOs). Table 6.1 shows the type of assistance received from various government programs (excluding CCT) and the cash value of assistance. Distribution of coal or firewood by local SYDVs is widespread. Among the CCT beneficiary households, 48 percent of education beneficiaries and 50 percent of health beneficiaries received coal or firewood from SYDVs, and the average value of coal or firewood for those who received them was YTL 150 per year. About 39 percent of nonbeneficiary-applicant households also received coal or firewood from SYDVs. Over one-fourth of beneficiary households and one-fifth of nonbeneficiary- applicant households received health assistance. The average value of in-kind health assistance is the highest among all types of assistance received. Besides SYDVs, sample households also received assistance (textbooks, school supplies, free school bus) from other agencies of the Government of Turkey. About 18-19 percent of CCT-beneficiaries and 6 percent of nonbeneficiary-applicants received free textbooks for their children from other government agencies. Very few households in the survey sample (CCT-beneficiaries and nonbeneficiary- applicants) received assistance from NGOs. Only 4 percent of beneficiary households and 3 percent of nonbeneficiary-applicant households reported that they received food assistance from NGOs. Other forms of support are negligible (Table 6.2).

35

Table 6.1 — Assistance received from government programs (excluding the CCT program) Education beneficiary households Health beneficiary households Nonbeneficiary-applicant households Cash Percent of Value of in- Percent of Value of in- Percent of assist Value of in- household Cash kind household Cash kind household ance kind Received assistance assistance Received assistance assistance Received (YTL assistance Benefits assistance (YTL/year) (YTL/year) assistance (YTL/year) (YTL/year) assistance /year) (YTL/year) Assistance from local Social Solidarity Foundations (SYDVs) Coal/firewood 48.3 3 150 50.1 2 150 38.7 3 128 House repair 0.1 0 110 0.0 0 0 0.1 0 0 Food for household 8.6 10 82 8.5 0 98 6.6 5 61 Clothing and footwear 2.0 0 52 1.9 0 71 0.9 6 45 Medicine 4.3 4 130 6.7 0 142 4.2 4 122 Education 4.1 153 84 0.8 0 88 1.5 9 37 Health 26.2 3 352 27.9 13 292 21.1 3 347 Other 0.9 8 153 0.8 0 13 0.7 50 206

Assistance from other government agencies School supplies 2.1 8 64 1.5 0 124 0.9 0 43 Textbooks 17.6 6 99 19.3 1 95 6.3 1 76 Free school bus 0.9 19 32 1.0 0 52 0.3 0 233 Free food at school 0.7 - 30 0.8 0 0 0.3 0 358 Other 6.0 17 190 2.7 35 175 4.2 3 208 Source: Based on data from ―Evaluating the Conditional Cash Transfer Program in Turkey: 2005-06 Household Survey.‖

36

Table 6.2 — Benefits received from nongovernmental organizations (NGOs) Education Health All Nonbeneficiary beneficiary beneficiary beneficiary -applicant Type of support households households households households (percent of all households) Education support 1.0 0.0 0.9 0.2 Health support 0.3 0.0 0.3 0.0 Support for heating 0.7 0.4 0.7 1.1 Food 3.9 3.4 3.7 3.1 Clothing 1.8 1.0 1.6 0.9 Other support 0.1 0.0 0.1 0.1 Source: Based on data from ―Evaluating the Conditional Cash Transfer Program in Turkey: 2005-06 Household Survey.‖

37

As mentioned in Section 2, the CCT program in Turkey uses administrative targeting to identify households with target group members who belong to the poorest 6 percent of the Turkish population. Program administrators use the ―proxy means testing‖ method to identify the poorest, which relies on indicators that are highly correlated with household income. This section assesses the targeting performance of the CCT program using quantitative and qualitative approaches.

7.1 Targeting Performance: Distribution of CCT beneficiaries and Rejected Applicants Across Income Groups

Although the large quantitative survey for CCT evaluation collected household consumption expenditure data for the sample households, these data are insufficient to show the pattern of distribution of CCT beneficiaries across income groups in the society since the sampling frame did not include all households at the village or county levels. Therefore, the Quantitative Assessment adopted a method of matching the characteristics of the households in the CCT evaluation survey sample with the characteristics of households from a nationally representative survey to carry out an indirect assessment of CCT targeting performance at the national level. For this, we used the data set of the 2003 Household Budget Survey (HBS) conducted by the Turkish Statistical Institute. For details on the method of assessment and the results, see Ahmed et al. (2006). The distribution of CCT beneficiary (education and health) and control households across per capita expenditure groups are presented in Table 7.1. Figure 7.1 illustrates the distribution for CCT education and health beneficiaries. The patterns indicate that the income distribution of CCT education and health beneficiaries among income groups is highly progressive. High percentages of all beneficiaries belong to poorer income groups,

Figure 7.1—Distribution of CCT education and health-beneficiary households, by HBS expenditure deciles

60.0 50 50.0

40.0 35

30.0 27 27 20 20.0 14 10 Percentage ofPercentage all households 10.0 6 5 2 2 1 0 0 0 0 0 0.0 0 0 0 1 2 3 4 5 6 7 8 9 10 (Poorest) (Richest) Per capita expenditure decile derived from HICES

Education beneficiaries Health beneficiaries

38

particularly the health beneficiaries. Half of all health-beneficiary households and over one- third (35.3 percent) of all education-beneficiary households are among the poorest 10 percent of all households in the income distribution at the national level. Indeed, no education beneficiaries and health beneficiaries belong to the richer 30 and 40 percent of all households, respectively. These findings suggest that the CCT program is well targeted to the poorest. The First Qualitative Study (Kudat et al. 2006) also concluded that the CCT appears to have succeeded in targeting the poorest segments of the Turkish society. At the regional level, this is reflected in the fact that the least developed regions and provinces receive the bulk of the benefits. The comparison of beneficiaries with non-beneficiary applicants systematically shows the former to have greater disadvantage in terms of assets and well- being. In addition, the comparisons of these two groups conducted in February and September 2005 also shows that the poverty focus of the program has sharpened as it continued to mature. Stakeholder discussions and interviews with beneficiaries reveal that a low level of leakages (that is, program benefits accruing to non-needy population) exists and, to a large extent, this results from issues pertaining to the application form; continued improvements made in the application form appear to have reduced leakages over time.

7.1.1 Errors of Inclusion and Exclusion

In any form of targeting, there are problems of exclusion (i.e., leaving out those who are needy) and inclusion (i.e., providing benefits to those who do not need them). The most recent poverty estimates in Turkey suggest that, in 2002, 27 percent of Turkish population was poor (World Bank 2005). If we consider the bottom 30 percent of all households in the income distribution to be poor, then the results presented in Table 7.1 suggest that 83 percent of the education beneficiaries and 92 percent of the health beneficiaries are poor, and, hence, are correctly selected as program beneficiaries. On the other hand, the pattern of distribution of the rejected applicant households across income distribution indicates that 59 percent of them are poor but excluded from the program. Table 7.1 — Distribution of CCT beneficiary and non-beneficiary applicant households, by expenditure decile Per capita Education Health Non-beneficiary expenditure decile beneficiaries beneficiaries applicants (percent of all households) 1 35.3 50.4 13.0 2 27.1 27.1 21.4 3 20.3 14.0 24.4 4 10.4 6.3 20.5 5 4.9 1.7 11.8 6 1.8 0.6 5.8 7 0.3 0.0 1.8 8 0.0 0.0 0.9 9 0.0 0.0 0.4 10 0.0 0.0 0.0 All 100.0 100.0 100.0 Source: Based on data from ―Evaluating the Conditional Cash Transfer Program in Turkey: 2005-06 Household Survey.‖

However, the CCT program targets only the poorest 6 percent of the Turkish population. Although precise estimates of the errors of inclusion and exclusion on this basis cannot be made from the decile distributions of households shown in Table 7.1, the patterns of

39

distribution of CCT education and health beneficiaries indicate large errors of inclusion in the program relative to the program‘s goal to benefit only the poorest 6 percent of the population. To state in other words, those who are included in the program are poor as compared with applicants not included in the program; there are however many households in the country who are as poor as those currently benefiting from it but are not yet included. In contrast, the pattern for nonbeneficiary-applicants shows a fairly small error of exclusion of households who applied for the program.

7.2 The Targeting Process at the Community Level

The initial stages of the targeting process—applications and verifications—were studied in detail in the six case study communities in the Second Qualitative and Anthropological Study (Adato et al. 2007). These results are presented below: In 5 Nisan, the school worked in cooperation with the Foundation, managing the first stage of the process: this involved telling prospective applicants about the possibility of joining the program, distributing application forms, assisting applicants in filling out the forms, then collecting them and sending them to Ankara for analysis and selection. In this community, this was a successful and cooperative process. However, we should also note that school officials are not completely comfortable with their (perceived) role as gatekeepers. Although the targeting decisions are in fact not made locally, the fact that the school plays a role in the first stages of the process can make school personnel vulnerable to criticism and complaint. As part of this evaluation, stakeholders were asked questions about the verification process. Verification checks are supposed to be carried out in at least 70 percent of eligible applicant households, in cooperation with local officials and personnel (such as muhtars, teachers, and medical personnel). In this community, Foundation officials noted that although where possible, checks were carried out and targeting decisions revised accordingly, there were simply too many beneficiary families to manage verification checks on all households, given current staffing levels. Indeed, the sheer number of applications in 5 Nisan posed a serious challenge to available technological and human resources throughout the targeting process. In Ergani, as in 5 Nisan, the Foundation, the school and the health centre worked together to assist prospective beneficiaries in preparing and submitting their application materials. In Ergani, the Foundation serves as a particularly strong gatekeeper, in effect carrying out a preliminary verification process even before the applications are sent to Ankara by discouraging applications from families known by the Foundation officials to be less poor. This approach brought its own problems: one official alleged that he had been assaulted and his house‘s windows broken by angry people. Verifications were carried out by Foundation officials in Ergani. The Foundation official interviewed for this study said “Our impressions are important for the selection process,” but then made the interesting observation that “Inspections are carried out after the beneficiaries are chosen and this causes people to distrust the state”. Beneficiaries from Ilyaskoy say that they were assisted by the school and the health centre in preparing their applications (the former is mentioned much more frequently in the data). Some key informants in Ilyaskoy expressed dissatisfaction with the targeting process, suggesting that it was overcentralized and lacking in proper verification.

40

In Yazilar, schools, the health centre and shops worked together with the Foundation in assisting prospective beneficiaries with the preparation and submission of their applications. One shop however was criticized by the Foundation for selling the application forms (these are supposed to be distributed free). According to the Foundation, teachers do not normally like to handle the paperwork during the summer. The verification process is taken seriously in Yazilar; even the teacher is involved. The Foundation official uses photographic evidence to check application data. However, he notes that the verification process is very time-consuming and it is difficult to fit it in with everything else: I am CCT personnel, but believe me that I can only give my 1-hour to this program, because I have to deal with other projects of the Foundation. I would like to focus only on CCT, but I also work as accountant. For a better selection, families must be closely followed after they start to receive money...Of course, if I were responsible from only CCT program, I would investigate these 560 families in two months. 560 is not a big number, but I have other duties.

In Beyüzümü, shops and schools were involved in distributing forms and assisting applicants to fill them out. Again, we note that at least one shop was selling the forms; however, one beneficiary pointed out that he preferred to pay a small amount than travel to the city centre Foundation office: “Once I gave 1 lira for a form but I am not sure what it was for. People are selling forms. It is OK. It is easier than going to the city centre to get them...” (Cercis S.). Foundation officials are critical of the process in Beyüzümü, suggesting (as in Ilyaskoy) that it is overcentralized and that verification should be done at a local level first, before accepting applications. Officials also point out that the program started late in Van and that the data management systems were slow. Several beneficiaries made the same point, suggesting that there were so many beneficiaries that the system had become gridlocked. Beneficiaries also note that verification is ongoing in Beyüzümü, and that the program is in some cases being withdrawn from households where verification reveals a problem. In YolaĢan, the muhtar announced the start of the application process from the mosque. The process does not seem to be very transparent in this community; even the health personnel do not know much about it. Teachers are involved in the verification process, but this carries its own dangers. As the Director of National Education said: “When you ask me about CCT I need to say that my teachers have many problems about that program. The headman and the teachers verify the poor people in the villages. The villagers get angry with them. They threaten the teachers. They once beat a teacher in the bazaar.”

7.3 Selection Criteria, Local Understandings, and Perceptions of Fairness

The Quantitative Assessment (Ahmed et al. 2006) found that beneficiaries know surprisingly little about the CCT program. An overwhelming proportion of the CCT beneficiaries and nonbeneficiary-applicants have little or no knowledge of the application and selection criteria. This knowledge is essential in ensuring some minimum levels of social accountability in the program. When asked about their understanding of the targeting process, most beneficiaries and non-beneficiaries in the Second Qualitative and Anthropological Study (Adato et al. 2007) correctly identified poverty as the main criterion for selection; of these, some identified

41

poverty as ‗green card.‘ Education needs were also cited as important. Corruption and contacts were mentioned by a very small number of informants, while a similarly minority believed that entitlement was universal. However, of the 12 key informants across the study communities who addressed this issue, only three felt that people understood the selection criteria. Interestingly, 3 key informants, all from the same community, stated that they themselves did not know how beneficiaries were selected. Many stakeholders had suggestions for improving the targeting process. Virtually all of these centered on improving the verification process through more household visits. One informant suggested that greater teacher involvement would also be a good thing. Key informants, with their more specialized knowledge, offered a much wider range of suggestions for improving the targeting process, and also the system of allocation and entitlement. It is interesting to note that many refer to the idea of decentralizing the process, placing more emphasis on the role of Foundations and local authorities because of their greater knowledge of local conditions; in fact, targeting in CCT programs throughout the world is typically quite a centralized process precisely to reduce the possibility of local level corruption and patronage. However, mixed systems of centralized targeting with local level review can be used with success (e.g. the Bolsa programs in Brazil).

7.3.1 Fairness: Errors of Exclusion and Inclusion

Although a significant minority of household informants believed that the targeting process had not been completely fair, in no community was there a strong tendency for non- beneficiaries to believe it had been unfair. Interestingly, of the 21 key informants who spoke to this issue, only 6 believed that the targeting process had been fair. Among both beneficiaries and non-beneficiaries across all communities, the data shows a slight tendency towards the belief that errors of exclusion have occurred rather than errors of inclusion, but this could also be a sociocultural/emotional effect of tending to pay more attention to people being excluded wrongly than to people being incorrectly included. A point that was raised in Samsun and Van is that there are characteristics in the proxy means test that contain a bias and do not always work out fairly, such as those related to social insurance and certain assets. For example, seasonal construction workers might have social insurance for a period when they apply, but are otherwise unemployed. In some areas all may have a TV. regardless of poverty, but those households may have no one who finds work. Rural villagers might not have plumbing but their costs of living are less than some in cities who end up poorer.

7.3.2 Tension Because of Selection

The majority of respondents say that there are no tensions between households which have been selected for the program and those which have not. Instances where respondents did identify tensions tended to be among the beneficiary population rather than the non- beneficiaries. Where tensions exist, they are expressed not only across the beneficiary/non- beneficiary divide, but also over the amount of money which people receive. The dominant emotions which non-beneficiaries mention experiencing as a result of their exclusion are sadness, sorrow, and lack of hope, occasionally mixed with resentment. Hayri O. from Ergani says that she feels anger and offence: If I had the money, I would spend it on my children. I could look after my house and my children better. I see that my friends receiving that money make investments with it. I get angry, then. I feel offended when my friends have money in their pocket and I do not.

42

Rather than turning on the selected beneficiaries, non-beneficiaries prefer to locate the blame for the decision externally: with the Foundations or other officials (from schools, health centres or local governments); this is a point which was also raised by Foundation officials who felt that they were regarded as gatekeepers.

7.3.3 Selection Appeals

Information about the appeals process is limited. Program documentation outlines an appeals process which applies both to non-beneficiaries who wish to appeal a targeting decision and to beneficiaries who have been suspended or expelled from the program. In this process, the person applies to the Foundation office, where they present their case. The Board of Trustees then reviews the events/situation and makes a decision, informing the applicant in writing (SRMP 2005). Foundation officials and local authorities say that such a process indeed exists and that they encourage unsuccessful applicants to re-apply (―petition‖) for the benefits. Two of the beneficiaries we interviewed had been able to join the program because their appeals were successful.

7.3.4 Why Some Do Not Apply

Although most beneficiaries and non-beneficiaries as well as a number of key informants suggest that everyone applied for the program, some others assert that there are households which did not apply and offer various explanations for this. Among key informants, the most commonly cited reason for not applying to the program is lack of knowledge. This is followed by lack of support or inability to handle the application procedures, honor, transportation costs, and a belief that only a limited number of applications will be accepted. To a large degree, data from interviews with people who did not apply to the program mirrors these explanations, although other reasons are cited too. Some people did not apply because they felt that they would be financially ineligible. One man mentioned his disability as a reason: he was bedridden and did not hear about the program. Language was also mentioned once as a reason (by a Kurdish interviewee). It was suggested by a small number of interviewees that legal marriage (or rather the lack thereof) was an obstacle to their applications.

7.4 Comments on the Proxy Means Test Scoring Formula

The CCT program uses the proxy means test—an indicator-based method of targeting— for selecting program beneficiaries from a pool of applicant households. The First Qualitative Assessment (Kudat et al. 2006) found that transparency in the identification of the poor has been achieved by the development of the proxy means test. However, all three Assessments identified some problems with the criteria used in the proxy means test, though problems of a different nature. Both Qualitative Assessments found that many stakeholders question the appropriateness of the criteria for eligibility and reliance on declarations of the applicants to make key decisions on inclusions/exclusions. While some of these stakeholder concerns appear to be misplaced based on a preliminary analyses of the large database created to serve as Management Information System (MIS), the difficulties involved in getting access to these data hinder a comprehensive effort in this direction. Using the household survey data on proxy means test indicators for CCT beneficiaries and nonbeneficiary-applicants, the results of a multivariate analysis of program participation reveal that the majority of the proxy means test indicators are not statistically significant determinants of program participation (Ahmed et al. 2006). At the same time, the

43

Quantitative Assessment concludes that the CCT program is well-targeted to the poorest. One plausible explanation of the discrepancy between the two sets of findings of the Quantitative Assessment could be that the information on indicators collected and verified during the process of verification contributed to the accuracy of targeting. The Second Qualitative and Anthropological Study points to the proxy means test scoring formula leading to errors of exclusion. For example, criteria do not take into account circumstances such as those of temporary workers, those with certain consumer goods but no income, and certain rural/urban differences. In sum, the findings of this study (quantitative and qualitative) suggest the need for improvements of the predicting power of the proxy means test algorithm.

44

The CCT program‘s foremost objectives are to increase school attendance rates, for the poor in general and for secondary-school girls in particular, and to decrease dropout rates. This section first presents quantitative findings of the impact of the CCT program on educational attainment. These findings are then explained with qualitative assessments. Since 1970, Turkey has sought to increase the duration of primary or basic education from 5 to 8 years. The Eight-Year Compulsory Basic Education Law was finally passed in 1997. The current structure of the Turkish education system was established in 1997-1998 school year, with 8-year compulsory primary education for children aged 6 to 14. Upon completion of the 8-year primary school cycle, students may enroll in secondary school, which is four years in duration (World Bank 2005a; World Bank 2005b).

8.1 Educational Attainment: Descriptive Analysis of Quantitative Household Survey Data

The large household survey (2,905 households) for this evaluation was specifically designed to collect the needed information to assess the effectiveness of the program in achieving educational outcomes and other objectives. Based on the household survey data, the rates of school enrollment, attendance, completion, and progression are estimated and costs of education are calculated for the education-beneficiaries and the comparison group of nonbeneficiary-applicants. The analysis also provides a comparative assessment of the quality of education between the education-beneficiaries and the comparison group. The findings of descriptive analyses of survey data are presented in this section. Section 8.2 below provides the results of the assessment of impact of the CCT program on educational attainment. Table 8.1 presents two types of enrollment rates for primary and secondary education: the net enrollment rate and the gross enrollment rate.19 The results suggest that net enrollment rates, generally considered better indicators of educational attainment than gross rates, are high for children at the primary (grades 1-8) education level, but quite low at the secondary (grades 9-11) level. For the entire sample of education-beneficiaries, about 91 percent of all children (boys and girls) aged 6-14 are enrolled in primary school. In contrast, only 48 percent of all children aged 14-17 are enrolled in secondary school. While boys overtake girls in terms of net enrollment at both levels of education, the gender gap is much larger in secondary school enrollment. The overall enrollment rates are higher for children from nonbeneficiary-applicant households at primary and secondary levels of education. Household profiles indicate that the characteristics of nonbeneficiary-applicants—some of which might influence the enrollment rates—are quite different from those of CCT-beneficiaries. Therefore, it is important to match the beneficiary households with nonbeneficiary-applicant households as a comparison group for a sound assessment of impact on school enrollment. This is done in this study and the results are provided in Section 8.2 below.

19 The net enrollment rate (NER) is the ratio of enrollment by children of the official targeted age (e.g., aged 6-14 for 8-year primary school) in a given level of schooling (e.g., primary) to the total number of children of the official targeted age. The NER excludes under-age and over-age children. The gross enrollment rate (GER) is the ratio of total enrollment for a given level of schooling to the total number of children of the official age. GER can be greater than 100 percent and is heavily influenced by the extent of under-age and over-age of enrolled children.

45

Table 8.1 — School enrollment rates Net enrollment Gross enrollment Boy Girl All Boy Girl All (percent) Education beneficiaries Primary education (grades 1-8) 93.4 88.2 90.8 98.5 91.7 95.2 Secondary education (grades 9-11) 57.9 38.2 47.8 67.6 40.7 53.8

Nonbeneficiary-applicants Primary education (grades 1-8) 94.8 91.4 93.2 99.8 94.7 97.4 Secondary education (grades 9-11) 64.9 46.3 56.3 74.1 50.3 63.2 Source: Based on data from ―Evaluating the Conditional Cash Transfer Program in Turkey: 2005-06 Household Survey‖. Note: Primary net enrollment rate = all primary school-going children aged 6–14/all children aged 6–14. Primary gross enrollment rate = all primary school going children/all children aged 6–14. Secondary net enrollment rate = all secondary school-going children aged 14-17/all children aged 14-17. Secondary gross enrollment rate = all secondary school going children/all children aged 14-17.

Enrollment rates are only one indicator of educational attainment, and can mask other problems in education. One vital indicator is the rate of school completion, or conversely, dropouts. For children from education-beneficiary households, 80 percent of boys and 60 percent of girls who entered primary school completed it, making the primary school dropout rates 20 percent and 40 percent, respectively, for boys and girls. For education-beneficiaries, completion rates are low for secondary education—65 percent of boys and 52 percent of girls who entered secondary school dropped out before completing grade 11. At the primary education level, completion rates are 4.2 percentage points higher for children from nonbeneficiary-applicant households than for those from education-beneficiary households. However, at the secondary education level, education-beneficiary girls have 7.8 percentage points higher completion rate than the rate achieved by nonbeneficiary girls. The rates of progression from primary (grade 8) to secondary (grade 9) level of education are relatively lower for girls than for boys from both education-beneficiary and nonbeneficiary-applicant households. In 2005, about one-third of all girls from beneficiary households stopped going to school after completing primary education. One of the conditions of the CCT program is that the children who receive education support should attend at least 80 percent of the total school days. In order to assess the fulfillment of this conditionality, the household survey collected information on students‘ school attendance. Table 8.2 shows the percentages of primary and secondary school students who attended school in all school-days in November 2005. For both primary and secondary levels of education, full-attendance rates are higher for students from education-beneficiary households than those from nonbeneficiary-applicant households. The difference between the two groups is the highest for secondary school girls—full-attendance rate is 12 percentage points higher for secondary school girls from education-beneficiary households than those from nonbeneficiary-applicant households. The household survey questionnaire included a module to collect data for measuring academic performance of students from education-beneficiary and nonbeneficiary-applicant households. Annual examination results in 2005 for 5th, 8th, and 10th grades were recorded from students‘ report cards. Table 8.3 presents the results. In general, students from CCT- education-beneficiary households tend to outperform students from nonbeneficiary-applicant

46

households. These results indicate that the CCT program might have contributed to better quality of education.

Table 8.2 — Students who never missed school in November 2005 CCT education beneficiaries Nonbeneficiary applicants (percent of students) Primary school student (grades 1 to 8) Boys 86.3 84.5 Girls 86.5 85.0 Boys and girls 86.4 84.7

Secondary school student (grades 9 to 11) Boys 80.0 76.7 Girls 86.0 74.0 Boys and girls 82.4 75.7

Source: Based on data from ―Evaluating the Conditional Cash Transfer Program in Turkey: 2005-06 Household Survey.‖

Table 8.3 — Academic performance of students in 2005 Education beneficiaries Nonbeneficiary-applicants Performance indicators Boys Girls All Boys Girls All 5th grade exam results (percent of students) Very Good/good 85.1 89.7 87.2 77.1 91.1 82.8 Medium/passed 12.2 9.1 10.8 22.9 8.9 17.2 Not successful 2.7 1.1 2.0 0.0 0.0 0.0 8th grade exam results Very Good/good 75.3 87.1 80.2 64.0 78.9 69.9 Medium/passed 22.6 11.9 18.2 34.9 17.5 28.0 Not successful 2.1 1.0 1.6 1.2 3.5 2.1 10th grade exam results 4.00 - 5.00 (high score) 41.5 37.5 40.0 27.3 50.0 36.4 3.00 - 3.99 (medium score) 50.8 50.0 50.5 57.6 50.0 54.5 0.00 - 2.99 (low score) 7.7 12.5 9.5 15.2 0.0 9.1

Source: Based on data from ―Evaluating the Conditional Cash Transfer Program in Turkey: 2005-06 Household Survey.‖

The CCT program is supposed to provide annual payments of YTL 216 per boy and YTL 264 per girl in primary school. These payments, if received fully by the beneficiaries, are adequate to cover the combined annual direct and indirect education costs the households incur on average for their primary-school-going children. Education costs are much higher at the secondary level than at the primary level of education—YTL 357.1 per boy and YTL 323.7 per girl. The prescribed CCT annual payments for secondary school students (YTL 336 per boy and YTL 468 per girl) are enough to cover the annual direct and indirect costs of education for secondary school girls, but not for boys. The evaluation results indicate that most education beneficiaries did not receive the full amount of the prescribed CCT payments. On average at the time of the survey, education beneficiaries received about 64 percent of the total CCT amount entitled.

8.1.1 Analysis of the Panel Survey Data on Enrollment

As mentioned in Section 3.1.1, the evaluation included a two-round panel survey of 750 CCT-beneficiary households who recently entered the program. The first round of the panel

47

survey was carried out for these households simultaneously with the large household survey. After seven months, a second round of the panel survey was conducted for the same households to see the impact of the CCT program on selected outcomes. Table 8.4 provides the net and the gross enrollment rates for CCT-education-beneficiary households in Round 1 and Round 2 of the panel survey. The enrollment rates are consistently higher in Round 2 than Round 1 for children (boys and girls) from education- beneficiary households. The difference in net enrollment rates between Round 1 and Round 2 at the secondary (grades 9-11) level of education is larger (5.6 percentage points) than that at the primary (grades 1-8) level (2.0 percentage points). Both of these differences are statistically significant at the 0.01 level of significance.

Table 8.4— School enrollment rates for education beneficiary households, panel survey results Net enrollment Gross enrollment Boy Girl All Boy Girl All Round 1 Basic education (grades 1-8) 90.8 85.1 87.9 95.6 89.0 92.2 Secondary education (grades 9-11) 48.9 29.0 39.2 53.4 30.8 42.4

Round 2 Basic education (grades 1-8) 93.5 86.3 89.9 97.4 89.0 93.1 Secondary education (grades 9-11) 52.0 37.9 44.9 61.0 44.1 52.5 Source: Based on data from the ―Evaluating the Conditional Cash Transfer Program in Turkey: 2006 Household Panel Survey.‖ Note: Basic education net enrollment rate = all basic school-going children ages 6-13/all children ages 6-14. Basic education gross enrollment rate = all basic school going children/all children ages 6-14. Secondary net enrollment rate = all secondary school-going children ages 14-17/all children ages 14-17. Secondary gross enrollment rate = all secondary school going children/all children ages 14-17

The results of the analysis disaggregated by gender suggest that the net enrollment rate for primary-school-age boys is statistically significantly higher in Round 2 than Round 1. However, although net enrollment increased slightly for primary-school-age girls from Round 1 to Round 2, the difference is not statistically significant. In contrast, secondary- school-age girls‘ net enrollment is statistically significantly higher in Round 2 than Round 1, but the increase from Round 1 to Round 2 is not statistically significant for secondary-school- age boys. While enrollment rates for boys are higher than those for girls at both primary and secondary levels of education, the gender gap is much larger at the secondary level. However, it is encouraging to notice that the gender gap in secondary school net enrollment has reduced from Round 1 (19.9 percentage points) to Round 2 (14.1 percentage points). The above patterns of changes in enrollment rates indicate an improvement from Round 1 to Round 2, even over a short period (7 months). However, these results should be interpreted with caution because the sample of households in the panel survey is small and not representative of the CCT program in Turkey, and because there was no comparison group in the panel survey to control for factors other than the CCT program that could have had an influence on the changes in enrollment.

8.2 Assessing the Impact on Education

The Quantitative Assessment (Ahmed et al. 2006) measured the impacts of the CCT program on educational attainment using a regression discontinuity design (RDD) method

48

applied to the large household survey data. RDD is an impact evaluation method that estimates the effect of participating in the CCT program by comparing average outcomes between beneficiary and nonbeneficiary households whose proxy means test score is near the threshold score used to determine program eligibility. The RDD methodology for this evaluation and results are explained in detail in Ahmed et al. (2006). The results of the RDD estimates are summarized below. The CCT program raises secondary school enrollment for girls by 10.7 percent. RDD estimates show that girls aged 14-17 whose families received the education transfers were significantly more likely to be enrolled in secondary school than those belonging to nonbeneficiary applicant families, with a difference in the probability of being enrolled of 10.7 percent. Estimates also indicate that, in rural areas, the CCT program caused a 16.7 percent increase in the probability of being enrolled in secondary school among children aged 14-17. For boys in rural areas, the estimated impact is even larger. Beneficiary boys had a 22.8 percent higher secondary school enrollment rate than nonbeneficiary boys as a result of being in the program. The CCT program raises primary school attendance for girls by 1.3 percentage points. This effect is driven by a significant 2.2 percentage point increase in primary school net attendance rates for girls in urban households that received the education transfer. Girls in rural households who received the education transfer had no significant difference in attendance rates, conditional on enrollment in primary school, than girls from nonbeneficiary applicant families. Overall, the education transfers increased attendance rates in primary schools for both boys and girls by 0.8 percentage points. In secondary schools, education transfers from the CCT program raised girls‘ attendance rates by 5.4 percentage points. As in primary schools, this effect was lead by impacts in urban areas, where girls in families receiving the education transfers had a 6.2 percent larger net attendance rate in secondary school than girls in nonbeneficiary applicants families. The program showed no effect on secondary school attendance for boys. The CCT program appears to have improved test scores for children enrolled in primary school. RDD estimates show that grade 5 students whose families received the education transfers were 20 percent more likely to receive the top score (―very good‖) on exams than grade 5 students from nonbeneficiary applicant households. This result was somewhat sensitive to model specification. Given the relatively small size of the impact of the program on primary school attendance, this effect is not likely to be caused by increased time in school. Instead, the education transfers may be helping beneficiary households to make better use of the schooling inputs, by increasing attention on schooling within the family and allowing families to provide children with more time to study or to focus on their school work. RDD estimates however suggest that the CCT program has no positive impact on primary school enrollment rates. The share of children aged 6-14 from education beneficiary households enrolled in primary school grew more slowly—by 3 percentage points—from 2003 (before CCT) to 2005 (after CCT) than those from the comparison group of nonbeneficiary-applicant households. Although the rate of primary school enrollment grew a healthy 5.9 percentage points for

49

beneficiaries from 2003 to 2005, the enrollment rate jumped 8.9 percentage points for a comparison group of nonbeneficiary-applicants over the same period. We speculate that the surprising increase in enrollment rates among nonbeneficiaries may be due in part to spillover effects from the program arising from the erroneous belief that, by enrolling their children in school, nonbeneficiaries may improve their chances of receiving the program in the future. Alternative impact estimates, on a subsample of beneficiary households who received transfers from the CCT program on a more regular basis, showed no statistically significant difference with the comparison group in growth of primary school enrollment from 2003 to 2005. This evidence suggests that even if the program maintains regular delivery of transfers in the future, significant positive impacts of the program on primary school enrollment are unlikely. The absence of an effect of the program on primary school enrollment is due in part to the already high enrollment rates at that level of education. Further, RDD estimates show no evidence that the CCT program affected the rate of progression from primary school to secondary school. One likely explanation for this result is that, children often have to commute to other communities to attend secondary school due to the low concentration of secondary school relative to primary school, which increases education costs and time to attend school. The demand-side intervention of CCT alone may not be sufficient to overcome this supply-side constraint. The First Qualitative Assessment (Kudat et al. 2006) found that the CCT increases awareness of female education and reinforces the hopes of the poor that the future of their children may be positively different if they are educated. CCT has helped change people‘s mindset about female education. Parents as well as female students benefiting from the program recognize that without CCT there would be greater reluctance to send girls to school beyond the first eight years level of basic education (grades 1-8). The Second Qualitative and Anthropological Study (Adato et al. 2007) set out to analyze through in-depth household studies how and why the CCT was or was not affecting decisions around children‘s schooling, and what would be needed to increase its impact, through program design changes or complementary service or program investments that respond to constraints on schooling (these constraints are analyzed in Section 8.3). When asked about the CCT in particular and its influence on decisions to send children to school, across all households in the Second Qualitative and Anthropological Study responses were split roughly equally between those who said that the money did and did not have an effect on their decisions. This study revealed explanations for this divided finding. In analyzing the interview data, it became clear that there are multiple influences and decisions are not straightforward. Poverty, schooling expenses, and money are important factors of influence, but they comprise just one sphere of influence—previous sections of this report discussed the non-economic spheres. For those who clearly valued education, many stated that they would send their children with or without the money: ―The program did not result in any change. I was intending to send my children to school. The program has merely assisted me with affording the expenditures‖ (Seher A.) The nature of these responses seem to suggest that while it is probably true many value education and would send their children regardless—or at least want to send them—it also looks like people do not feel comfortable saying that they send their children to school for money. But even where they said it was not their main motive,

50

they did emphasize that the grant helped them to be able to afford schooling. It is also interesting that some children commented that the grant was making the difference, even where their parents said differently. For households that were not sending children to school, some said that their expenses were too high, they could not afford school even with the grant. And in some households, particularly in Van as will be seen below, money could not compete with sociocultural beliefs weighing against education. Some of the case study households acknowledged directly that the CCT played a central role in enabling their girls to continue their schooling: through financial assistance, and by enabling the mothers to effectively respond when their husbands suggested that the girls do not continue through secondary school (Box 8.1). Although the women do not acknowledge it in these terms, they are effectively empowered to take a stand with respect to a decision that they otherwise would not be able to take without the government offering financial support and the message that girls should go to school. In another case, a beneficiary explains how her husband registered the girls at school, but then told them to stay at home, and the money was cut off: “I was happy then, he needed to send them regularly. I told him that the government sees everything. He came here and asked for the money and they told him girls must go to school every day. He could not say anything then.”

Box 8.1—The role of the CCT in enabling girls schooling: Financial support and moral persuasion in Diyarbakir (Adato et al. 2007). Sükran and her husband have 5 daughters and one son. The CCT benefit plays a key role, both in enabling the household to meet its expenditures, and in helping the mother to convince the father to let the girls continue. Sükran, the mother, said that they would send their children to school whether or not the program existed, but that the program made a great difference—it enabled them to meet their children‘s expenditures, and she “paid more attention to her children‟s continuity due to the program” Although the father says that he would continue sending his daughters to school as long as they were interested in going, his wife tells a different story—that he wanted to take them out after the 8th grade and that she has had to argue and persuade him. He did not want to send their eldest daughter to the school because she “was quite grownup.‖ She put pressure upon her husband, and wanted her to attend until the end of the eighth class so that she would not be left as ignorant. She said that her husband incessantly said that their daughter would get married sooner or later. But when she told him that the CCT aid was given for students who attended secondary school, he said he would consider it. Sükran wanted to send her daughters if they were successful and intelligent, and she wanted her children to have occupations. But they “would not be able to send them to school as they were in unfavorable conditions economically…. they could send their children to school, and that she could respond when her husband argued against sending them to school, thanks to the aid. She said that she said that it was the state who lets them go to school, who meets their expenditures, and that she could respond to her husband more easily. He used to say that he did not have any money. „So how can I send them?‟ he used to say, but he cannot say something against it any more.” The father also said that they were able to send their children to school thanks to the aid that the state gives, but that it would be easier if the state also provided books.

The program provides not just financial incentives but serves a consciousness raising function. In Cercis S‘s household in Beyüzümü, the father said: ―I will send her to school for two reasons. First now government is helping. They are giving school money. And second I am curious about it. Lets see what will happen if she goes to school.” This comment is interesting because it implies that in addition to sending money, the government is also sending a message that children should go to school. This has a weight of its own, apart from the financial incentive. Some people do not respond well—saying respectfully that the state should mind its own business. But others are curious, or imply that if the state thinks this is important that it must be important.

51

As noted in earlier discussions, there are households for whom the CCT does not influence their decisions because their reason for not sending their daughters to school are not primarily financial. In another Beyüzümü household the mother said that they are not sending their daughters to school, because her husband “won‟t send them even if government gives money. Student money did not change anything about that because it is not appropriate to send girls to school according to our customs.” (Gazi A.) Although parents were more reserved about saying that money influenced their decision, children may not have the same inhibitions and some directly asked us or indirectly tried to send a message to help them go to school by continuing the benefit . In Beyüzümü, one girl told us that: “my father would not send us to school if he does not get money. He may not send us to secondary school so please tell them to give more money to my father and other fathers. (Mihri A.) In another, a girl told us that her father sent her and her sister to school but ―I know that‟s only to get the money. They give him money to send me school. My teacher told me. He sent my sister to school also with me for that money. She was older than me but we were in the same class. If there were no money he would not send me, my sister or others to school. My father will send me to secondary school if you give him more money and if no one asks him to marry me.”(Beriyan T., Beyüzümü) Her mother confirmed that her husband did not want to send their daughter to school but the benefit changed his mind. He is allowing her to go to school as long as the money continues; however, in his case it is a strictly financial decision and marriage can be too; in this case he will take her out of school when she finds a husband. An important finding of the research is the importance of the CCT to primary education in places like YolaĢan, where many of the factors discussed throughout this chapter mitigate against girls going to primary school, and the benefit does make a difference. This points to the importance of taking into account regional variations in program design: in regions with high primary school enrollment, it may be advisable to provide the CCT for secondary school only. However, in villages such as YolaĢan and even in Beyüzümü where values are not entirely different, continuing the primary school benefit may be critical to providing a better future to girls—at least a basic primary education and enabling the possibility of further education down the road. At the same time, it is evident that the CCT program alone will not be sufficient to overcome many of the obstacles to schooling described in this chapter, particularly secondary school and particularly for girls. Other efforts to respond to logistical concerns such as transportation and school location, safety at schools, children‘s performance, and attitudes and values will be a crucial part of the strategy that the CCT program can not tackle on its own. In Van they used the campaign ―Let‘s Send Our Daughters To School‖ (Haydi Kizlar Okula), going door to door to raise people‘s consciousness, that was reported to have had a good effect. This approach, combined with providing school books and the availability of primary schools, made a significant difference in primary school attendance. But there remain tremendous obstacles to secondary school attendance—both material and non-material—that need an integrated response.

8.3 Factors Explaining Education Decisions and Impact of the CCT: Overview of Findings of the Second Qualitative and Anthropological Study (Adato et al. 2007)

Despite the program impacts described above, enrollment rates are still low, particularly at the secondary school level, and particularly in some regions. The Second Qualitative and Anthropological Study set out to understand the nature of the constraints to children‘s

52

education, and to girl‘s education in particular. Why are enrollment rates as low as they are, particularly at the secondary school level, as indicated in the survey results? Frist, in order to get an overview of attitudes toward education across our study households, the Second Qualitative and Anthropological Study examined whether parents prioritize education of boys and girls in comparison with other factors that mitigate against continued schooling. In all three study areas, there is strong support indicated for boys education, with approximately 88 percent of those answering the question indicating support for. For girls the picture is more mixed. In Diyarbakir and Van among beneficiaries there is a roughly even split in these attitudes, with a leaning towards lower priority given to girls education relative to other factors. In Samsun girls education is strongly prioritized, at least in terms of parents‘ expressions of their priorities. However, this assessment does not take into account whether parents were able to act on these priorities or whether children actually stayed in school, based on a number of factors explored below. The Second Qualitative and Anthropological Study explored the intricacies of parents attitudes toward education, their expectations for their children‘s futures, and how much importance they gave to their children‘s education relative to other priorities and why. The household case studies also enabled us to explore the attitudes of children themselves, how their own experiences and preferences affect schooling outcomes. The family case studies investigated why each child stayed in school as long as he or she did, and why he or she left school. Key informants added important insights into the context of schooling. We uncovered a complex web of social and economic factors that influence schooling decisions, with regional differences. In the Second Qualitative and Anthropological Study household studies, 18 different categories of factors emerged that affected education decisions, the most prevalent 12 of which are shown in Figure 8.1. It is important to note that the main factors are economic— Figure 8.1— Factors affecting education decisions by regions

120

100

80

60

40

20 Percentallof households

0

Other

Money from work School performance Help needed at home

Gender roles and expectations Safety and influences at school Cost of school expenses/poverty Children's feelings about school Role of education in chidlren's future Transportation and location of schools Factors affecting decisions

Diyarbakir Samsun Van

53

poverty and costs of school expenses; and social and cultural—gender issues, school performance, and children‘s attitudes toward school. Two salient issues—the role of education in future welfare and transportation issues/location of schools—have strong economic and social dimensions. There are also important regional differences.

8.3.1 The Perceived Role of Education in Future Welfare

Parents and children make education decisions based on short term costs and long term benefits. For boys, education is important for getting employment, and better quality, more dignified employment. A more respected man is said to be a better husband and father. At the same time, high rates of unemployment has made people cynical about the value of education. This came out most strongly in the study communities in Van and Diyarbakir, where unemployment is highest, but expressed in Samsun as well, where many do not continue with secondary school because they think that having an occupation or learning a craft is more valuable than having a secondary education, or they are tempted by the material possessions that can be acquired in the short term from working. In the village in YolaĢan in Van, attitudes toward education are similarly shaped by the perception that jobs are hard to find in the city, but the value of education is even less for rural life and agricultural labor. If there are no jobs, then education does not contribute toward obtaining food. This helps to explain why the CCT may not be having as strong an impact as it could in Van, and points to the importance economic development interventions that complementary the approach of the CCT. The main role of education in girls‘ future is articulated as freedom from ignorance. Schooling enables them to read and write, and for Kurdish-speaking women, to learn to speak Turkish. Education helps them to „move around,‟ outside the home, to talk with others, their husbands, with doctors and nurses, and government officials. The recognition of the importance of being able to talk with officials is raised in the context of the CCT program, where for some women it is their first experience of these kinds of interactions. For some households, particularly in the Van study communities, these benefits can be achieved through primary education and they do think further education is needed. Further education is still associated with work opportunities, and in Diyarbakir and Van, it is not acceptable for grown girls to work. Furthermore, educated girls are also presumed to have difficulties fitting in with their communities and households. In the urban study areas, we found girls with aspirations to go to university and have careers, and parents who supported these aspirations, even as they had adjusted slowly. but they revealed that this can only occur if the girl leaves her town so people in her community will not know. Nurses and teachers are their role models. If the CCT increases girls and parents exposure to teachers and nurses, this might help to make these role models more visible, potentially strengthening and spreading these aspirations. In Iyaskoy the situation is very different, and women can work. However, general secondary school (as opposed to a vocational school) is not seen as that relevant to obtaining work.

8.3.2 Gender issues: Marriage, Honor, Appearance, Sexuality, and Roles and Expectations of Women

If the different types of gender issues that are cited as barriers to girls schooling are taken together, gender is the second most highly cited factor across the households in the Second Qualitative and Anthropological Study, cited in 36 percent of households in Samsun, 71 percent in Diyarbakir, and 100 percent of households in Van. This aggregated category represents several concerns and assumptions. The first and most prevalent concern was

54

related to honor, reputation, and girls sexuality—risks and threats faced by girls and their families if girls go to school, especially secondary school. The objection revolves around concerns that girls will be attractive to men and thus are at risk, and thus once they are mature, girls should not be interacting in the public sphere outside of the watch of their parents. If something happens, it will damage the family‘s honor, and many people say that „honor is more important than money.‟ The next most prevalent explanation is the strong cultural belief across the study areas is that girls‘ main need is to get married, and the social identity and expectations of young women revolve almost entirely around marriage and raising children. The role of secondary education therefore is questionable. In some areas girls marry very young, and once they are married they do not go to school. Furthermore, because girls will go to live with their husbands when they are married, the benefits of education are seen to accrue to the husband and his family rather to her own parents—it is thus an investment with little return. The overall finding of this section is that the CCT, because its direct and indirect objectives include increasing girls schooling and improving women‘s status, is a critical intervention that has the potential to make an enormous contribution to girls and women‘s conditions in the short and long term. However, because of the sheer strength of the gender biases it confronts, it will not be enough on its own. In some dimensions of the problem, redesign of program features such as increasing the amount of the benefit for girls‘ secondary school attendance can help. In other dimensions, complementary approaches will be needed, pertaining to transportation and location of schools (see section 8.3.3), as well as other programs aimed at changing attitudes. To some extent, it will need to be accepted that change will take time, and in some areas may not be achieved in the foreseeable future. Various respondents said that an 8th grade education is sufficient for girls; in fact, the struggle sometimes is to keep them in school through the 8th grade: She said that her husband said that they should not send their eldest daughter to the school as she was quite grownup. She said that she put pressure upon her husband, and wanted her to attend until the end of the eighth class. They would not be left as ignorant if they study until the end of eighth class. She said that the girls were already grownup after the eighth class. She said that her husband incessantly said that their daughter would get married sooner or later and they would be left with their son and themselves and they could get on better together, and he said that they would consider it when she said that the aid was given for high school students (Sukran A.).

The CCT is seen here as potentially making a difference in enabling the girl to go to secondary school. Education to the 8th grade is frequently cited as sufficient, probably because completion of this grade was formerly mandated by the state. It is thus conceivable that as secondary school becomes normalized and seen as an expectation if not a state requirement, the idea that ‗8th grade is sufficient‘ may begin to subside. However, this is likely to take many years, and it is not certain whether this would happen. It is also the case that the 8th grade also on average coincides with more obvious physical maturity for girls and this is another inhibiting factor. Extreme poverty is also a reason that girls are married early—if their parents have trouble supporting them economically, a husband can take on this role. Another reason for early marriage is parents‟ concern that the girls will get involved with boys. Marriage relieves

55

the family of this responsibility. Girls are also a form of „capital,‟ where if someone pays the marriage price for her she will be taken out of school and forced to marry. The amount of the CCT is small to compete with this. The household studies in Van revealed this powerful role of marriage, where parents conveyed that school does not impart the skills that girls need for marriage, but also that people look down upon educated girls and that education makes them less desirable as a marriage partner. On the other hand, it is also the case that attitudes have changed over time, though very slowly, at least in the city: “At old times girls should get married at the age of 14-15 so there was no need for school but now they stay at home for a longer time so if there is a good school with no rumors about boys, fathers send their daughters.” (Cercis S)

The picture derived from this research is of a process of change but a very slow one, that has different starting points and different paces in different regions of the country. Traditions such as age of marriage, the appropriateness of sending girls to school and the level of education they can respectfully attain, and what they are allowed to do with that education, are influenced by many factors: the media, role models in communities, parents own experiences of life with and without education, and social programs. The CCT program can contribute to this process of change, by offering monetary incentives that will in some cases delay the age of marriage and thus enable the continuation of schooling, and through its indirect message that girls‘ education is important and valued by the state. Honor emerges repeatedly as a theme in relation to girls and their schooling. Parents express an anxiety about girls going to school and being vulnerable to attention from boys at school or men in town, or about girls becoming interested in boys, and the implications for the honor of the family if people were to observe these interactions (see Box 8.2). Several

Box 8.2—Honor and Transportation in Yolaşan (from Adato et al. 2007) Kardelen T‘s household has two boys who are attending school in town, and one girl who is not being sent to primary school. There is no primary school on their side of the village, it is too difficult to walk, and there is no transportation system. The father explained that: I have said many times to the teacher to send car to my house to take her otherwise how can I take her to school every day, to the other side of the village? My house is on this part of the village but the school is on the other side. Let boys to walk, it doesn‟t matter but girls can not walk there by themselves. Anyway I am sending my sons to the school in town. There is the free bus for them but girl can not go to the town, no. What‟s she got to do in the town? If a girl is going to school, the school should be in the village. What can I do if the school is on the other side? Even there is the free bus, a girl can not go out of the village. There are bad people all around. Let‟s say someone fools her, abducts her, who will clean my name? Are you or is the teacher going to clean my name? A girl could only go to school if there is a school in her village, near her home. She should be in front of her mother and father all the time. That‟s very simple. It would be better if the teacher is a woman. A boy could find a car even by hitchhiking and go to his school. There are the free buses for them. He is a man anyway. I won‟t let my daughter to streets for that money. Don‟t look at me, I would not send her for any money. Her mother confirmed that this problem was common in the village: People don‘t send their daughters to town but girls are going to the school in the village. So maybe it would be good to have school in the village but we live in the other part. My children can not walk that way to school in winter. Or they would get sick. It is good for my boys to be taken by bus but I would be afraid if my daughter would go with that bus. We know the driver, he is from our village but my husband still says it is not safe for the girl. It may be because of the other men in the town.

56

informants refer to the importance of educating parents to reduce these concerns. But parents also express that the schools need to do a better job of protecting girls and guaranteeing to the parents that their daughters will be safe. These findings on gender issues have different types of implications for the CCT program. On the one hand, they demonstrate the enormity of the challenges that this program faces. But they also suggest that the CCT may be playing a role of greater importance than is visible from the current enrollment increase revealed by the full quantitative survey, resulting from the increased visibility of girls in school—and the message sent by the state about the importance of girls‘ education.

8.3.3 Transportation and the Location of Schools

Transportation and the location of schools was identified as a constraining factor to schooling by almost half of the Second Qualitative and Anthropological Study households, with the greatest concentration in the rural study communities in Yazilar and Yolasan. The issue involves the distance of schools from people‘s homes, and concerns about transportation—cost (in some places higher than the grant), availability (government transportation services do not reach everywhere) and safety (mainly, threats from boys and men). Closely related to the issue of transportation, and included in these numbers is the issue of location of secondary schools—people‘s desire to have secondary schools in their communities. This would not only alleviate transportation problems and costs, but also make parents more comfortable, as they are better able to keep an eye on their children, or in some areas avoid the embarrassment of being publicly perceived to let their daughters travel around. Some households were sending their children to school but had complaints about distance and transportation: they wished that the state would provide it. In Yazilar, several parents said that they could not afford the transportation fees, and told the story of the school service explained above. Reflecting several views expressed, one beneficiary “If there was school service, all children would go to school.” (Sükran Y.) Though the CCT is supposed to help with these costs, sometimes is not enough, and also it is not seen as reliable (given past delays). In Van, one foundation official thought a system involving free transportation would be more effective than the CCT at the level of secondary/high school. The father of one household in YolaĢan whose daughter stopped schooling after 5th grade also indicates the role of social expectations and conformity: Only if all families start to send their daughters to secondary school, I will send my daughters to secondary school. Change should be done all together. If there are still respectful families who do not send girls to town, then it is not good to send them. If all girls in the village were going to school, then we could rent a minibus for them and some one we know could take them to school. (Ahmet A.)

This may suggest that even if the CCT has not yet succeeded in achieving high enrollment rates; it may be playing an important role in increasing the visibility of girls schooling, and providing more examples and role models that may make other families more comfortable and in the long run lead to greater change. Thus far, in a village such as YolaĢan, the program had not made great strides, but it is the subject of conversation, and at least seems to be increasing the likelihood that girls will go to primary school, a necessary prerequisite for secondary school. The case of YolaĢan also suggests that it is not advisable to cancel the benefit for primary school education across the country as a whole, but rather to

57

enable variation in program design across regions to be responsive to different economic and cultural characteristics.

8.3.4 Children’s Performance in School

Children‘s school performance, widely referred to as whether they are ‗successful at school,‘ was a surprisingly strong criteria for whether parents wanted to continue sending them.. This is also related to another surprisingly common factor, children‘s preference for school, because children who do well in school tend to like it and those who do not perform well tend not to like it. Still, this criterion is different in that it is often the parent‘s decision than the child‘s to remove the child on this basis. The salience of this factor suggests that if a complementary effort is put into improving school quality and performance of students, it could go a long way toward encouraging parents to send their children. Reasons for considering performance relates to considerations of children‘s future prospects. Those who do well in school are seen to be more likely to be able to take advantage of their education, obtaining better employment later. They may perform poorly because they are not trying hard, they are not that intelligent, or have emotional, physical, or behavioral problems. For those who do not do well, they are seen as better off working, in the case of boys, or getting married, in the case of girls. Since school costs money and it is a sacrifice for the family to pay it, it is also perceived as a waste of money if the child is not going to take advantage of the opportunity. In this light, the CCT is a responsive intervention because it enables the household to take a risk in sending the child to school, alleviating the concern about ‗wasting‘ scarce household money. This can give failing children more opportunities to improve, or allow them to receive some education, even if they do not perform that well. On the other hand, not all parents agree: Orhan L discussed his poorly performing children and concluded that: “ it would be meaningless to send them no matter if they delivered the aid or not if their lack of success continued.” The picture drawn here is of a need to complement the CCT with other approaches that aim to tackle some of the underlying non-economic barriers to school enrollment. This may call for social services or extra support for students during or after school. But it also suggests that much is dependent on the role and attitudes of parents. It may be that the effect of the CCT will be felt in the next generation, as educated students grow up to be educated and more supportive parents.

8.3.5 Children’s Own Views of School

One of the more unexpected finding of the Second Qualitative and Anthropological Study was the extent to which children‘s own will influences decisions and outcomes with respect to schooling. The conventional assumption in many parts of the world is that it is parents who are primarily responsible for decisions on children‘s schooling, whether they will enroll, how long they will continue. Certainly, CCT programs are designed to serve as incentives, primarily to parents who receive cash for household expenses. We found however, that children play a significant role in these decisions, and for a number of reasons, they often choose not to go to school. For some children a dislike of school comes from bad or humiliating experiences, often related to different dimensions of poverty. Some children start school late, are repeating grades, or have re-enrolled (some to get the CCT) and are older than the normal age for grade, which embarrasses them. Another humiliation for children relates to a more direct symptom of poverty: the inability to buy books, or decent clothing (one benefit of the CCT was found to be new pride for children as they come to school with better clothing and supplies). In Van and Diyarbakir, we found some cases where children did

58

not know Turkish and therefore struggled at school. But, as we have seen elsewhere in this chapter, there are children who love school and want to be educated.

8.3.6 Safety and Influences at School

Parents‘ concerns about safety in the schools, and bad influences on their children is another factor explaining schooling decisions, though most prominently in the Diyarbakir study areas. The issue serves to illustrate—as do the other factors but here more starkly—that different regions of the country, and probably even different districts and localities, will have different reasons why parents persist in resisting sending their children to secondary school. These factors should be taken into consideration, then in a more comprehensive approach to increasing schooling, complementary interventions to the CCT program must be found.

8.3.7 The Cost of Schooling and Need for Assistance

Economic constraints on schooling—expressed by study households as the cost of school expenses and poverty—were the most frequently cited factors across the sample, cited by 80 percent of households in the Second Qualitative and Anthropological Study. This is an important finding because, in the face of all the social factors discussed thus far upon which the CCT program will have a more limited impact, the cost of schooling is where the program has its direct impact. Parents frequently specified the items for which they must pay if their children go to school, and complain about their expenses: including books, pens, cloths, food, and pocket money. In general, they estimate that their expenses are considerably more than the amount paid by the CCT (twice the amount was often suggested as necessary, and as much as three times in Samsun), particularly for secondary school students. The quantitative survey data found that the cost of the grant was sufficient except for boys‘ expenses at secondary level, and it is likely that people ask for more than what they need. But people were quite consistent in their statement that ‗the money is too little‘ and key informants in all study areas also confirmed that very poor families have trouble making ends meet even with the grant. Costs also are said to vary by region, and with them the adequacy of the level of the CCT. Another reason for the perceived inadequacy of the grant is that the state does not cover the cost of secondary school books as they do for primary school. In the case of the Fahriye T‘s household in Ergani, the CCT is directly responsible for a re-evaluation of a plan to take a girl out of school upon completion of primary school: I did not want my daughter to go to school. But now that there is this aid money, I will let her go to school; maybe she can acquire a profession too. I did not want to send her to school so that she could help me at house…I thought that my daughter was a grown-up, and that it was not appropriate for her to go to school. But I will let her go, maybe she can become a nurse.

In Yazilar one household said that they could not afford secondary school, and when the researcher pointed out that they would lose their grant if she did not go, they said that because the money arrived irregularly, they could not count on it. Households in Yazilar complain of the cost of transportation to secondary school, or the cost of the dormitory. There are also expectations of proper clothing for children to have if they are going to attend school.

59

This section assesses the impact of the program on health and nutrition, and explores the factors explaining these effects.

9.1 Vaccination

The CCT program aims to increase immunization coverage of children from poor families and to promote usage of health facilities. The results of descriptive analysis based on data from the large household survey suggest that the immunization rates are higher for children less than 6 years of age from health-beneficiary households than those for the nonbeneficiary-applicant households (Table 9.1). Although the vaccination coverage was almost universal, the doses were not completed for a considerable share of children from both groups of households. For example, 98 percent of children from health-beneficiary households and 95 percent from nonbeneficiary-applicant households were vaccinated against diphtheria, whooping cough, and tetanus (DPT), but the dose was completed for 84 percent and 82 percent of children, respectfully, from health-beneficiary households and nonbeneficiary-applicant households.

Table 9.1 — Immunization of children of age 6 and below Vaccination Health beneficiaries Nonbeneficiary-applicants (percent of all preschool children) Vaccinated against tuberculosis (BCG) 96.0 95.3 Vaccinated against diphtheria, whooping cough, and tetanus (DPT/triple) 98.1 95.2 DPT dose completed 83.9 81.1 Vaccinated against polio 94.2 92.9 Polio dose completed 86.5 81.1 Vaccinated against measles 95.9 93.8 Fully immunized 79.6 74.8 Source: Based on data from ―Evaluating the Conditional Cash Transfer Program in Turkey: 2005-06 Household Survey.‖

The above result from descriptive analysis is supported by the rigorous impact analysis using the regression discontinuity design method. RDD estimates reveal that health transfers from the CCT program lead to an increase of 13.6 percent in the full-immunization rate for preschool children. For children below 6 years of age, the rate of full immunization against tuberculosis (BCG); diphtheria, whooping cough, and tetanus (DPT/triple); polio; and measles jumped from 43.8 percent to 57.4 percent as a result of participating in the health component of the CCT program.

9.1.1 Analysis of the Panel Survey Data on Immunization

Table 9.2 presents the immunization rates for children aged less than 6 years from CCT- health-beneficiary households in Round 1 and Round 2 of the panel survey. In the first round of survey, 77.9 percent of the children were fully immunized. The rate increased to 81.3 percent in the second round of survey. This difference between Round 1 and Round 2 of the panel survey is statistically significant.

60

Table 9.2—Immunization of children of age 6 and below, panel survey results Vaccination Round 1 Round 2 (percent of all preschool children) Vaccinated against tuberculosis (BCG) 95.6 94.9 Vaccinated against diphtheria, whooping cough, and tetanus (DPT/triple) 95.3 95.5 DPT dose completed 86.3 87.2 Vaccinated against polio 88.9 94.6 Polio dose completed 88.6 86.2 Vaccinated against measles 94.5 93.6 Fully immunized 77.9 81.3 Source: Based on data from the ―Evaluating the Conditional Cash Transfer Program in Turkey: 2006 Household Panel Survey.‖

9.1.2 Second Qualitative and Anthropological Study Findings on Vaccinations

In the Second Qualitative Study (Adato et al. 2007), the issue of vaccination was explored as a part of an overall analysis of people‘s attitudes to health and preventative medicine. We were also interested in discovering whether any kind of relationship existed between the CCT program and patterns of vaccination among beneficiaries and non- beneficiaries. Vaccination is achieved in two ways. In some cases, children are brought to clinics for vaccination. They are also vaccinated in an outreach program, where nurses go house-to- house vaccinating children. This latter approach to vaccination appears to have been more successful—as one beneficiary from Beyüzümü said “I want the nurses to come home and do the vaccinations. Sometimes they come and sometimes they don‟t and I did not take my children to the health clinic when the vaccination time came so I could not get money. I came here and asked for my money but they told me that I did not take my children to the health clinic” (Mihri A). Indeed, given the ambivalent relationship with hospitals and clinics described above, it is easy to understand that people prefer to have nurses visit their homes on vaccination drives. This quote is telling for another reason too: it illustrates the belief that the CCT benefit is related to vaccination. This beneficiary thought that she was supposed to receive money for vaccinating her children; in fact, clinic staff told her (correctly) that she needed to bring them to health controls in order to receive benefits. We found evidence of this belief that CCT benefits are related to vaccination in two study provinces, Van and Diyarbakir. In both of these provinces, beneficiaries made reference to ‗vaccination money‘. Moreover, this was borne out by the observations of heath staff. According to health centre personnel in YolaĢan, “People don‟t understand about the health controls but they think that the condition is vaccination...‖. This observation invites us to reflect on the ‗collateral‘ benefits of the CCT program: while one interpretation would suggest that people should not have incorrect understandings about the program, another suggests that ‗beneficial‘ behavior, even when founded upon a false belief, is a positive outcome. Vaccination was often described as a woman‘s responsibility: although serious illness required men‘s intervention to take sick family members to the hospital, vaccination (insofar as it was seen as a responsibility) was cast as a mother‘s job. In a few cases, it was evident that vaccination was regarded with suspicion, as a potentially harmful practice—a belief resulting from people‘s observation of the fever which can be a side effect of some live vaccinations. This may explain why the survey found that a high proportion of people did not

61

complete their vaccination doses.20 Another suspicion was that vaccination caused infertility. Indeed, health centre personnel explained that some people, particularly in Southeastern Anatolia, believed that the government was trying to systematically reduce the population by administering ‗sterility drugs‘ disguised as vaccines. However, we found little evidence of these beliefs in household interviews—rather we found examples where people said that had heard this but since learned it was not true. It seems likely that people‘s less-than-enthusiastic view of vaccination has more to do with the generally ambivalent conceptualization of hospital health care than with any real suspicion of negative effects caused by vaccines. We should note though that in cases where vaccination was viewed negatively, it is likely that health controls were affected in the same way, because of the fact that people associate the health components of CCT primarily with vaccination and not with health controls. We found little measurable difference in attitudes towards vaccination between the beneficiary and non-beneficiary population. This result matches the quantitative survey results which show total immunization rates of 70.7 percent and 70.4 percent, respectively among beneficiaries and non-beneficiaries across the three study provinces. Intra-household discrimination, i.e. vaccinating some children and not others, was also not in evidence: only one informant, a beneficiary from Diyarbakir, suggested that she favored her male children in health practices.21

9.2 Illness

Since diarrhea is an important cause of child morbidity, its incidence among children is an important indicator of linking health and developmental outcomes. The results of Quantitative Assessment based on the large household survey (Ahmed et al. 2006) show that about 36 to 40 percent of children aged 6 and below from sample households had diarrhea in the summer of 2005. About 90 percent of the households reported that they had consulted medical professionals for treatment of diarrhea for their children. Medical consultation for illness is jointly determined by demand for and supply of health services and knowledge about medical practices at household level. Over 90 percent of CCT health-beneficiary and nonbeneficiary-applicant households sought medical consultation for illness, which is indicative of universal access to health services (Ahmed et al. 2006).

9.2.1 Health-Related Practices Used by Households and the Socio-Cultural Context for the Health Component of CCT

The Second Qualitative and Anthropological Study (Adato et al. 2007) shows that, households, both beneficiary and nonbeneficiary, employ a range of traditional healing practices at home, and visiting traditional healers within the community. This situation appears to be slowly changing; it was common among informants to refer to traditional healing as a practice rooted in the past. As is often the case, such ―traditional‖ approaches to

20 In the national survey, it was found that vaccination coverage was almost universal, but that the doses were not completed for a considerable share of children from treatment and control households. For example, 98 percent of children from health- beneficiary households and 95 percent from nonbeneficiary-applicant households were vaccinated against diphtheria, whooping cough, and tetanus (DPT), but the dose was completed for 84 percent and 82 percent of children, respectfully, from health-beneficiary households and nonbeneficiary-applicant households (Ahmed et al. 2006). 21 That said, we should note that the greater shame attached to girls‘ and women‘s bodies can itself result in a de facto sex discrimination within households, in cases where families are reluctant to seek care for girls‘ health problems because of shame. This issue relates to healthcare seeking in general and is not of course restricted to vaccination practices.

62

illness coexist pragmatically with ―biomedical‖ responses, the latter being more widely applied in cases of ―serious‖ illness. Thus while families tend to treat minor ailments at home, in cases of graver health problems, they turn to the hospital or to the services of a doctor. Although one might expect rural communities, being both more ―traditional‖ and also usually more remote from hospitals, to show greater dependence on traditional healing, there does not appear to be a strong urban-rural distinction at work here: the coexistence of traditional and biomedical responses does not show much variation across the study sample on either the beneficiary/non-beneficiary or the rural/urban axis. In the urban Van community Beyüzümü our respondents were more somewhat more inclined than in other study areas to reject traditional healing in favor of biomedical solutions (interestingly, Beyüzümü does not have a local clinic and residents must leave the community to attend medical services in other nearby communities). Interviews with informants across the study communities indicated that there are always a range of factors at work, underlying each decision that a family takes about what kind of medical care to seek. The most important and salient of these is severity of illness, although even this has shifted over time. In the past, people went to the hospital when they were severely ill, presumably beyond the powers of traditional remedies and healing techniques. Now, “...everybody has green card and they go to hospital even for a headache...” (Nilgun K.): according to informants in all study communities, hospital attendance has become much more common and although people still express ambivalence about their experiences at hospitals, they are nonetheless willing to go, at least, in part, because the green card allows them to access free medical care. While the green card has certainly opened up the possibility of more frequent hospital care-seeking, many people are still disinclined to attend the hospital unless they or their children are seriously ill: (She) said that she had a green card and benefited from it when she got sick... She said that they did not go to the doctor very frequently but only when they get seriously sick... She also said that her children took her to the hospital as she was illiterate...she said that nobody in their region went to the hospital unless she/he is seriously sick, and that they already hardly bear the distress of hospital when they are seriously sick...She said that they generally used analgesics at home. She said that they made mint and lemon tea in events of common cold (Azize S.).

Note that this quotation also tells us other things about informants‘ health care-seeking decisions: the informant notes that she is illiterate, and that she feels ―distress‖ in the hospital, and indeed that this distress is a widespread phenomenon in her region. Why should this be the case? In the following section, we discuss this issue and also address people‘s experiences with clinics, hospitals, and doctors more generally.

9.2.2 Opinions about Health Services

Beneficiaries and non-beneficiaries, in all communities, regard doctors with a degree of suspicion and sometimes outright hostility. They spoke about their experience with doctors, clinics, and hospitals in ways which referred to very negative experiences, in which they were not given proper explanations and were attended by hurried doctors in crowded conditions. Some informants accused doctors of demanding cash from green card holders, while other respondents complained that they felt discriminated against because they spoke only Kurdish and no Turkish.

63

There appear to be a range of possibly related explanations for this sense of antagonism. One concerns what people perceive as the hostile attitude of hospital staff. As noted, in the Kurdish-speaking eastern areas of our study (Van and Diyarbakir), language can also be a problem, both in terms of comprehensibility and in terms of discrimination and inferiority. Compounding these problems is the issue of shame and body-centered embarrassment. Many women said that it was improper or uncomfortable for women or girls to go to male doctors. For example, one beneficiary mother from 5 Nisan refused to take her daughter, who was suffering from irregular menstruation, to see a doctor because she felt it would be too embarrassing. Another beneficiary from Ergani said that she had had two operations, both carried out by male surgeons. In each case, she felt obliged to conceal the sex of the surgeon from her husband for fear of his negative reaction. Finally—and this is a point that is particularly relevant to understanding the sociocultural context in which the CCT program is operating—we should note that in spite of the green card program, there is a marked antipathy or ambivalence toward the concept of preventative health care.22 A point that comes up again and again in the interview data is that people do not go to hospitals for check-ups: they go when they are seriously ill. With the possible exception of vaccination, health care is regarded as essentially reactive rather than proactive.

9.2.3 Health Controls: Beliefs, Attendance, and CCT

Data from household interviews suggests that very few people know about the check-up (‗health control‘) conditionality. Indeed, as we note below, where people do acknowledge or recognize the presence of a conditionality, they tend to believe that this concerns vaccination and not check-ups. When asked whether they took their children to regular controls, virtually all beneficiaries responded as described above: that they took their children to the clinic/doctor/hospital when they were sick and not otherwise. Responses from non- beneficiary interviewees were similar. One doctor from Beyüzümü suggested that the people who brought their children in for check-ups would do so anyway, regardless of the program, because they are “health conscious.‖ This same doctor also noted that, once applications to the program have been made, people make a conscious effort to attend the clinic regularly, in the hope that this behavior will be observed and taken into account, enhancing their prospects of being accepted as program beneficiaries. In addition to the generalized ambivalence about clinic- or hospital-based health care, key informants suggest a range of reasons why people do not attend check-ups more regularly. These included migration to other communities, distance and cost of transportation, complexity (because check-ups must be performed at different periods depending on ages of children) and a shortage of medical personnel.

9.3 Nutrition

Although increasing food consumption is not an explicit goal of the CCT program, the income from program transfers and increased interactions with health services may have increased food consumption for beneficiary households. From the Quantitative Assessment (Ahmed et al. 2006), estimates of the income elasticity of food (calorie) consumption show that the average impact of the income transfer alone is likely to be small, a 2 percent increase

22 Reluctance to engage fully with preventative health care was also identified in the first (2006) qualitative study of Turkey‘s CCT (Kudat 2006: 54).

64

in per capita calorie consumption. Indeed, RDD impact estimates for the full sample of beneficiaries found no average impact of the CCT program on per capita calorie consumption. However, for a subsample of beneficiary households who received transfers from the CCT program on a more regular basis, the CCT program increased their per capita calorie consumption by 22.6 percent relative to a comparison group of nonbeneficiary households. The First Qualitative Assessment (Kudat et al. 2006) found that income impacts are growing as the program matures and the program makes an important difference in the wellbeing of the extremely vulnerable and destitute families. The allocation of CCT for food improves nutritional standards and contributes to school performance. Mothers generally believe that their ability to provide better food to children makes a difference in their school performance. Beneficiary mothers were consistent in mentioning the importance of improved food for children as being directly relevant for their school performance. While the quantitative evaluation attempted to quantify the relationship between household income and calorie consumption, the Second Qualitative and Anthropological Study (Adato et al. 2007) assessed household dietary practices to discern whether households perceive improved food security and nutrition quality in their diets since receiving the CCT. In comparing the diets between beneficiary and non-beneficiary households there does not appear to be any evident differences. Vegetables are the most common food type consumed followed by grains and diary products. For very poor families, fruits are said to be consumed with less frequency mainly because of unavailability or high prices, and chicken and beef are seldom consumed: ―The rich buy a bounty of meat; they buy sausages, and pastrami. And the poor buy whatever is cheap. They cannot buy meat or fruit‖ (Seher A, 5 Nisan). Urban and rural households recall consuming similar foods, except for milk and other diary products, such as yoghurt, which are consumed in larger quantities in rural households and in households that maintain connections with rural villages. Products that can be bought in bulk, such as flour, sugar and pasta, are consumed more regularly but purchased less frequently in order to take advantage of lower prices from the wholesaler: ―Our children eat whatever we eat. I go to shopping twice a week. Mostly I buy vegetables from the market place. I can scarcely buy fruit. I buy once in two months flour, margarine, sugar, pasta from the wholesaler‖ (beneficiary mother, Ergani). Generally, males in the household believe that ensuring good nutrition and overall health are women‘s responsibility. Women are well aware of men‘s lack of interest in food and nutrition, and agree that it is them who have to ensure children receive the best food possible. Such strong non-intersecting responsibilities regarding nutrition can have important implications on the wellbeing of a child depending on who is the ultimate person responsible for food consumption decisions. It does appear that a large percent of women remain in control of CCT money in Samsun and Diyarbakir, and less frequently in Van (see chapter 11). In households where this is the case, and if money is spent on food coupled with the fact that women are more aware of nutrition, then the CCT program must to some extent be effecting positive changes in the nutrition situation of children and families. While the objective of the CCT program was not to improve child nutrition per se, it could be assumed that the extra money provided by the program would allow households to purchase additional food for their children, inevitably leading to improvements in child nutrition. Generally, households indicated that CCT money allowed them to increase amounts of food bought and purchase food items that previously they would not have been able to afford. However, when households were asked about their perceptions on changes in their diets, most households reported eating similar food and having similar ―typical diets‖,

65

regardless of whether they received CCT money or not. In other words, increases in amounts and types of foods eaten did not translate into better perceived household nutrition. One probable explanation is that inconsistency in receipt of CCT money does not allow households to experience a dramatic change in their diets. Better and more food purchases occur but in bursts, and understandably this is not always perceived as an improvement: ―My mother spends her money for kitchen expenses as well. But this changes every month: sometimes she buys better food, and sometimes she buys only our basic needs. Some months, our income is higher due to child money, so she can buy better things.” (Ilyaskoy)

66

In cooperation with the Ministry of Health, the CCT program is designed to create incentives for pregnant women to obtain adequate prenatal care, and to reduce pregnancy risks through pregnancy check-ups, vaccinations, protection from pregnancy anemia, and birth and post-birth care in hospitals. As of January 2005 the program pays 17 TYL per month to pregnant women, and a one-time payment of 55 TYL if birth is given in the hospital. The condition for receiving this support is that pregnant women attend monthly health check ups from the start to the end of their pregnancy, birth given in the hospital (for the one-time payment) and post-natal check-ups. If identity information is given to the Foundation after birth, the baby is registered as a health beneficiary. Over the course of the program, concerns developed, particularly among some health providers and Foundation staff, that this pregnancy component of the program might be creating incentives for families to have additional children that they would not otherwise have. There was concern that this program might be undermining some of the progress that the health sector has made to strengthen family planning. The quantitative and qualitative studies both set out to investigate this issue to determine whether or not it was creating these unintended consequences. The Quantitative Assessment (Ahmed et. al. 2006) based on the large household survey found that the CCT program has no effect on pregnancy of child-bearing-age women from participating households. The RDD impact estimates show no evidence to support these claims. Quite the contrary, the estimates show that receipt of education or health transfers from the program actually reduces the probability of a woman of child-bearing age (16-49) becoming pregnant by about 2-3 percent. It is somewhat surprising that the program appears to discourage pregnancy, though this may reflect in part the effect of the additional income from the program and visits to health clinics on household fertility decisions. Besides the RDD estimates, the results of a multivariate regression analysis also suggest that participation in either health or education components of the CCT program has no statistically significant effect on pregnancy. The Second Qualitative and Anthropological Study focused intensively on this question and explanations for why the CCT would or would not be increasing pregnancy rates. Findings from this research (Adato et al. 2007) are reported in the remainder of this chapter. Findings were consistent with those of the survey. Among those who answered the question (some did not want to because they found the question offensive), 32 people said they did not know anyone who did this; while 3 said that they did. In the households case studies we explored people‘s attitudes toward fertility and family planning, how their family size came to be what it is, their motivations for having many children or not, and how the CCT benefit might or might not figure into their fertility decisions or the decisions of anyone else they knew. We also conducted interviews and groups discussions with key informants, in particular, health professionals and Foundation staff, about their views on these topics. With respect to the question of the CCT benefit, our findings were consistent with those of the survey: —nearly all beneficiaries and non-beneficiaries overwhelmingly said that they would not become pregnant to obtain a cash benefit, nor did they know anyone who would. In fact, many thought the question to be either strange, humorous, absurd, or offensive. In some households people did not want to talk about this or it was not comfortable to ask— particularly in YolaĢan. Still, some key informants said that they had encountered some people who had done this, and others had not but still believed that this was a problem. There

67

were also reasons why it was believed to have been a problem more in the past than now, due to misunderstandings of the amount of money to be paid. The interview material is quite convincing that people do not get pregnant for the CCT, and we conclude that they do not. However, decisions around fertility are often economic, and because people‘s understanding about different aspects of the aid is very vague, the answer to this question is not entirely straightforward. These issues are explored below. The main finding of the Second Qualitative and Anthropological Study (Adato et al. 2007) is that there are so many social pressures on families to have more children—and economic pressures not to have more children, that the CCT is largely irrelevant to their decisions. These issues were the same in all three study areas, though to varying degrees, with social pressures highest in Van (see box 10.1). With respect to social pressures, large families make a household feel stronger, respected and powerful. To have few children is strange, mothers-in-law in particular are said to pressure couples. Many said that birth control is a sin. There are also softer aspects. One father said that ―Having many children is pleasant. Being a crowded family is nice.” (Bedri E.) In rural areas, children are needed to take care of livestock, gardens and fields—but much less so now than in the past, when these activities were a much more significant part of people‘s livelihoods than they were. The exception was Yolasan, where children were needed to work. But in Diyarbakir and Samsun, people stated strongly that it is better not to have too many children—they say that in the past they did not know better, but have since learned that it is better not to have so many, and most say they do not want more. Having many children is said to be difficult and exhausting, and if poor, a constant struggle. It is also harder to earn a livelihood now, and children‘s expenses are higher than they used to be—in part related to education. Even in Van, where the importance of large families were stated vociferously, some women quietly said that they did not want more children. They spoke of being very poor, of having health problems, of the children having health problems, of being terrified over missed periods. But they also said they did not have much choice. The vast majority of the study households stated strongly that no one would get pregnant because of the CCT program: People would not get pregnant for money. They would get pregnant because they want it. They may have many kids to have more sons. I don‟t know, here you must have a son. People make fun of you if you don‟t. People want sons to have their descendants. Or out of ignorance, they want sons (Esma B.).

In Diyarbakir we did come across three cases where the CCT was suggested to be a pregnancy incentive. In one household, the mother said that she would have another child if she knew she could have the CCT aid regularly. However, she has just two children, and would like to have a third, but can not afford another one, unless her husband finds a job. It thus appears that the CCT could make a difference if there is a strong intention for another child regardless and a small financial constraint. Misunderstanding about the level of the benefit also has an effect. In Diyarbakir and Van, rumors circulated that the amount of the grant was much higher than it was, as much as 1,000 TYL. In Van even health center staff had this misunderstanding. Another reason why the pregnancy benefit is unlikely to increase pregnancy rates is that few people in all three areas were aware of it—they do not differentiate it from the other parts

68

of the CCT. It is all “child money.‖ But there is thus a concern that people assume that the government will provide ―child money‖ throughout the child‘s lifetime. In one sense this could be seen as accurate—if one goes from pregnancy to health to education benefit. However, the aid is seen as unreliable in the short and long term, not well understood in terms of what it is, where it comes from and when, and it is unlikely that people would make life long decisions based on it.

Box 10.1—Samsun women’s reactions to questions about the CCT and pregnancy (Adato et al. 2007) In asking questions to women in Ilyaskoy and Yazilar about whether they or anyone they know would get pregnant in order to get the CCT money, many women laughed. The idea struck the women as humorous, and they teased the fieldworker for having to ask what they appear to have regarded as an absurd question: Allah, were there woman that got pregnant just for this money, really? Ha ha ha! I also didn‟t hear about those women who think like „They will continue to pay for my baby after birth‟, I really didn‟t hear, ha ha ha. Of course, you have to ask these questions, this is your duty! But mothers have to think about their children‟s future as well. (Ilyaskoy, Sengul G.) Someone had been there before asking the same question, and women in Yazilar also made a joke out of it. During the interview a neighbor started to laugh: “Last year, some guys like you came here to write about this money....Before you, another person asked me the same question, he asked: „Are there any woman who get pregnant to receive child money‟. And I said „No, all women got tired of their existing children‟ (Yazilar, Hayriye Y.) She said that new brides must have babies, but it has nothing to do with money.

Most key informants in Samsun and Tekkekoy, in particular the health and education staff, and local government officials, do not think that the CCT is encouraging pregnancy in their area. However, the Foundation in Samsun has been affected by what they have heard on this issue, that there has been “a population explosion in the East,‖ and decided not to provide applications or information on this part of the program. The Foundation in Tekkekoy did not know much about the pregnancy benefit, and had not received any applications for it. The health center professional in Buyuklu said that when they mentioned the pregnancy support to people, there were no overwhelming applications for it, just a few.

Box 10.2—People don’t want children because of that money. People want to have children anyway— Van (Adato et al. 2007) Gazi A lives in Beyüzümü with her husband, six children, their sons‘ wives and their grandchildren. They receive the CCT benefit for their sons; their daughters never went to school. Gazi explains that women must give birth to many children here, ―men want crowded families to show their manhood.‖ Women who do not have children are taken to religious men or given medicines. If this does not work her husband can take a second wife or send the woman home. She said that “there is the new invention called birth control which is a sin, write this down for yourself.” She does not want her daughters or daughters in law to use birth control: Children are good. They are the will of the God. Here we want more children. I want my daughters and my sons to have lots of children. Child is a blessing. If you have lots of children, they would take care of you in the future and your family would be powerful and your name would survive. Child is the honor of the mother. Young brides don‟t want to have many children but they are wrong. What is that got to do with student money?... I don‟t know anyone who get pregnant to get child money…. People know that government money is not permanent. But it is a silly question. God gives the child. Men want lots of children anyway. They want power and they want their name to survive. Her husband also says that they want to have lots of children: ―Why does this bother you?‖ He points out that his sons are looking after his own family and his parents: ―Children mean power and they are carrying your name. God gives the child and his fortune. It is a sin not to want something God gives. I want my sons to have children, to have sons. No, I don‘t think people want children because of that money. People want to have children anyway.

69

Foundation staff explained that when the pregnancy aid first started, the TV channels were announcing it in subtitles, saying “Good news: money will be given to those who give birth!” Naturally, the medical centers panicked.” This is seen as being a reason that the health department did not appropriate the program. The debate on this issue seems to be based on opinions, and it is difficult to establish fact. Even among key informants, however, there was some ambiguity and uncertainty. In fact, among 26 key informants who spoke to this issue across the six study areas, about half thought the benefit caused increased pregnancy and the other half did not. We do not believe that the ‗pregnancy effect‘ is entirely imagined, and it is certainly possible to see how it could have some incentive effects, in circumstances where a family wants another child very much and where a small amount of money is perceived as tipping the decision. But the finding of the Second Qualitative and Anthropological Study at the community and household levels is that this is not the main circumstance under which people make these decisions, and that the CCT aid has a minor if any influence on people‘s fertility decisions. Still, many service providers and Foundation officials remain ambivalent around this issue and some recommend reducing the number of children that would be covered. (Requirements for all children to meet conditions could be maintained. See section 12.)

70 The qualitative research went beyond assessing impacts on education and health, to look for CCT ‗externalities‘—unintended consequences, potentially positive or negative effects, on work patterns, and intrahousehold dynamics.

11.1 Adult Work Patterns

A concern among policymakers and stakeholders about cash transfer programs is that they might create a disincentive for people to participate in the labor market. Given work insecurities and fluctuating income, the thought is that households will become reliant on steady CCT money. In the large household survey for the Quantitative Assessment (Ahmed et al. 2006), beneficiary families were asked whether the opportunities made available from the CCT program would discourage them to look for a job or tempt them to work less. A full 100 percent of the beneficiary households said that CCT would not dissuade them from keeping a salaried job or from doing their job regularly. The Second Qualitative and Anthropological Study (Adato et al. 2007) focused on this issue of work disincentive to shed light on whether or not this is in fact occurring. When urban and rural beneficiary households were asked about CCT‘s impact on joblessness, the most common response was that the money received is both unreliable and insufficient to have a significant impact on the labor market participation rate. For example, a mother beneficiary from Ilyaskoy explained: “No, it is impossible that people give up looking for a job after starting to receive child money. I didn‟t hear because the amount received from the Foundation is too little. It helps, but I don‟t think so that people will trust that money and give up looking for a job because we are not even certain about when to receive child money.”

In addition, perspectives on the importance of working —the self-worth it brings to a bread-winner to personally provide for his families and the respect this brings to the family— are important factors to ensure an able person remains in the labor force. These sentiments were expressed by both beneficiaries and non-beneficiaries. Work ethic is also perpetuated by expectations from community leaders, who promote a sense of responsibility among men to earn a living for their family through their own means. Furthermore, discussions with household members on their current work situation and work-seeking habits suggested that little has changed among beneficiary breadwinners. In the majority of beneficiary households there are men who either have a job or are searching for one, and the latter talk about the stresses of not having a job and clearly want one (see Box 11.1).

72

Box 11.1—CCT does not hinder work patterns (Adato et. al. 2007) Nadire T, a beneficiary from Ergani, talked about her husband‘s relentless search for employment: ―My husband is very diligent and endeavoring, and he works when he finds a job, he does any job he can find. There is no employment in Ergani so he goes somewhere else outside the town…. My husband never avoids a job, he seeks a job all the time, and whatever the conditions are he works when finding a job.‖ Neriman Y, a beneficiary from Yazilar, believes the money is not sufficient to significantly impact work patterns: ―Could someone give up looking for a job because of this benefit? There is no such thing. How important is it to receive 60 YTL every two months, who might leave his job just because of that amount?"

The CCT program thus does not appear to have a significant impact on adult participation in labor markets. Generally, men follow similar work patterns as they did before the implementation of the CCT program. Instances where the CCT caused men to consider less work were infrequent. Although key informants expressed concern that the CCT might have detrimental effects on employment or work seeking behaviors, as expressed by a muhtar from Ilyaskoy: “There are also some people not looking for job because of this benefit, they say „we already receive benefit‟. They have not the working discipline. There might also be people not enthusiastically looking for job, but this is just a prediction. And their excuse is ready: „There is no job‟. Some don‟t work, but they smoke Marlboro.”

We did not find evidence of this in our household interviews, and the research findings on how the CCT is viewed suggest that these concerns are, for the most part, unfounded.

11.2 Child Labor Patterns

Child labor is oftentimes a coping mechanism for poor households during time of economic hardships. An important result of CCT program is their safety net component, which can help diminish the impact of economic shocks on poor households and hence reduce reliance on child labor. The First Qualitative Assessment (Kudat et al. 2006) found that the ―opportunity cost‖ of CCT, the degree to which families lose money when their children go to school instead of seek work, is variable, but generally low in rural areas and in underdeveloped regions in which CCT concentrates. Inquiries with school age children with work show that CCT has not affected children‘s participation in the labor market, especially in industrial areas of Western Turkey. In-depth interviews and community discussions pointed out that both girls and boys of poor families find work after they complete the 8th grade and for those CCT contributions constitute merely 8-9 percent of their income from industrial employment. The Second Qualitative and Anthropological Study (Adato et al. 2007) explored perceptions on the value and need for child labor to better understand whether CCT money has impacted households‘ reliance on child labor. Parents‘ reasons for their reliance on child labor were consistent across all six communities. Families who struggle economically stated that they make their children—generally only boys—work as soon as they reach the appropriate age. Orhan A, a father in 5 Nisan noted: ―What could I do as I undertake the burden of the whole household alone? If my son was older he would work and make a contribution to the livelihood of the family.‖ These families use child labor as a coping mechanism to alleviate their economic difficulties, and CCT receipt does not seem to enable them to move away from relying on such a strategy. A second reason mentioned regularly

73

across communities was that work experience is important for a child‘s future. Parents believe that education is not enough to provide children with the necessary skills to find a good job. Joining the labor force at a young age helps children to ―learn about life and have a craft‖ and to ―get accustomed to the conditions of work life‖ (Gazi A., Beyüzümü). Interviews and observations in the field suggest that child labor continues to be a widespread phenomenon. It is still common for children in school to participate in either paid or unpaid labor; girls continue helping with housework outside of school hours, boys and less frequently girls work on family fields during summer holidays. Seher A, who lives in 5 Nisan, sends her children to school but relies on her daughters to help with house chores, while her son is allowed to play: ―My daughters both do housework and attend school. They look after their young siblings. The boy goes to school, goes out, and plays games. My daughters do not go out.‖ In Abdullah C‘s household from Ilyaskoy, girls are not allowed to work but boys are expected to work during the summer, even with the CCT: “I will let my sons work in summer holidays. Child money doesn't make difference; I will send my sons to work even they receive child money. But I won't let my daughters work.” The impact CCT has on child labor appears modest. While recent school enrollment rates have increased in areas receiving CCT money, the Second Qualitative and Anthropological Study suggests that program money has probably resulted in reductions in the frequency and number of hours worked during the school year, rather than the prevalence of children working at some point during a given year. Importantly, household interviews suggest that perceptions of child labor and reliance on it have not changed. Children continue to value work in the same way they did before, and adults continue relying on it not only as a coping strategy but as a way to increase the marketability of their children. As long as child labor is perceived as a net positive for both the household and the child, children will continue to work regardless of the educational consequences.

11.3 Use of the CCT Money

Among study beneficiaries, clothes were the most frequently bought item, followed by food, school supplies, and among some households for paying back debts (primarily with wholesalers and grocers). Health care and household services, such as electricity, water and rent were seldom mentioned. Although the number of households who use the CCT for food expenditures is high, there are many who believe that this money is exclusively intended for children‘s school expenses and should not be used for purchasing food. Enver D a beneficiary father from Ilyaskoy notes: ―I spend the money for children, not for kitchen. They say it should be spent for children, for notebooks, pencils, etc…. No, we don‟t spend it on food, the state says that we are supposed to spend it for our kids.‖ Regardless of what the money is used for, generally program recipients expressed that the money was not sufficient to cover all basic needs (see Box 11.2).

74

Box 11.2 Making ends meet (Adato et. Al. 2007) ―The money coming from the CCT affords scarcely our debt to the grocery shop. I am living in this house, too. I use the money I have to pay our debts… The amount they pay is not enough for us. I do not have a husband, so they should have given me more. I am grateful though because of the money we receive. At least I have an income, although it comprises of a small amount… My children have no father. We have no house. We live with my in-laws. My in-law takes care of us. The state should help people like us more. The expenses of school children should be met. The state should give the children whatever the school is asking from them, including school uniforms. I cannot afford all these. If I buy one of our needs, I cannot afford another. I try to buy as much as I can afford, but I cannot afford most of our needs. My elder daughter looks after a nurse‟s little child. She earns 120 TL a month. This is all our income. What can I do with this amount of money? None of my other children works. Should I spend this money on my children‟s school expenses or should I provide them with food?‖ (Aysel A, beneficiary mother, 5 Nisan).

11.4 Intrahousehold Relations

The CCT program designates mothers as recipients of the money. This feature of the program design has brought into play new cultural and intra-household dynamics. Due to women‘s restrictive role in household decisions and control of money, selecting mothers as recipients can create new and evolving female-male roles. These are evident from the time CCT money is collected at the bank through to purchasing decisions and intra-household relationships. The First Qualitative Study (Kudat et al. 2006) found that the CCT has had a very strong impact on the increase of women‘s participation in society. With few exceptions women during interviews recount with pride their participation in the application process and their use of funds disbursed to them, especially to feed the household and educate their children. CCT helps women formally register their marriage and obtain birth certificates for their children. The Second Qualitative and Anthropological Study (Adato et al. 2007) found some similar effects, but concluded that they were much more moderate, and involved gender dynamics that played out in complicated ways. These findings are presented in the remainder of this section. Generally, the cultural norm is for men to handle the money, and imposing a new role—women as recipients—was disturbing to both men and women in conservative parts of the southeastern Turkey. Some men stated that they disliked the government imposing a social requirement for women, seen as going against their traditions and ways of life. For example, in YolaĢan, one man in a beneficiary household stated: ―There is a problem. What‟s that money got to do with the mother?. I don‟t understand that. It was good before. I was taking their mother‟s id with me to the town and they were giving me the money but there is a new invention. They are saying she must come with me to take the money. Do I have to take my wife out where men gather around? Next step will be for women to go by herself… These are all new inventions to make our work harder. These are to change our life.‖ Women too questioned their new role as recipients, expressing that it is difficult to comply because of not being accustomed to leaving their houses and entering an environment usually considered a man‘s world. However, differences were found across the provinces and these concerns were more common in Van than in Diyarbakir and Samsun. Differences in decision-making and resource distribution among spouses were found to be province-specific as well. Beneficiaries in Van stated that the husband is both the decision-maker and spender. Men have complete control over household resources, and this seems to result in their control over CCT money. In Samsun and Diyarbakir, the situation is

75

slightly different. Even though generally the beneficiaries noted that it is the husband who decides how to spend the household income, in households where women have an important role to play in decision-making related to nutrition and health, women are the ones who control the CCT money. The benefit for them in keeping the CCT payment is that they now manage their own money, which gives them more freedom and choice over purchasing decisions. Participation in the CCT program in and of itself will change the reasons why women leave the house— all women have to go out to collect the money. The Second Qualitative and Anthropological Study examined how women perceive this change. Among women interviewed, around half perceived no changes in their day-to-day social interactions and the other half did. For women who notice little change, although they now need to leave the house more in order to collect the money, these are insignificant in their overall daily activities. Sengul G, a 27 year old mother from Ilyaskoy, has 4 children and one nephew living in her household. Although she goes out to spend the CCT money, her life patterns have not changed as she has to stay at home to take care of her children: ―Has this grant led to some changes in my life? Every two months, I buy whatever my children want. Apart from this, nothing changed. No, it‟s impossible; I don‟t go out more frequently. I have 4 children to look after, I can‟t go out.‖ Among some, the increase in number of outings is seen as an added burden to their lives. They do not like going out and so resist the new opportunities given to them. A beneficiary mother from Yazilar noted: ―I go to the town when I have a business but it makes me sick. I don‟t like to go there. If I have to or if I hear that the money has arrived at Bank, I go to Tekkeköy. Otherwise, I don‟t.‖ Particularly among women living in Van, their difficulty to adapt to this new opportunity for outside social interactions and participation in monetary transactions is compounded by an overall resistance from within their communities. Women who did report changes in their outing patterns expressed a sense of gratification at being able to add more outside interactions to their routine: “I go out more nowadays in order to get the money from school, and in order to spend it in the market place. I use to go out for shopping also formerly, but now I shop more frequently, and it is the school trip that is added to my routine.” (5 Nisan) What is important is not just that they go out more frequently but how these new outings can impact other aspects of their lives. Some women note that they go out more to speak with their neighbors, and others comment that they get together with other beneficiaries to travel together to the collection point. Participating in the program has helped women find commonalities among themselves, enjoying some of the responsibilities together and ultimately translating into a general increase in outside social interactions. Although the research looked into whether the program encouraged women to participate more frequently in group-based activities (as found in other CCT programs), little information was gathered as virtually no group-activities exist in the communities studied. What can be highlighted is the information captured on whether women would participate if such activities were established. The overwhelming response was that women would not be interested or able to join. Generally women stated that their husbands would not allow them to participate, and some believed that social life among women is considered inappropriate and would not like to take part in something that is not socially accepted. An established and structured group-activity may be too unconventional for both men and woman. The way the CCT works is more effective— women begin to participate in activities traditionally considered a male‘s sphere, which in turn brings about more exposure and opportunities to

76

interact with other women. With time these can develop into strong social networks and support for women. The Second Qualitative and Anthropological Study assessed whether the unique stipulations of the CCT program have helped to promote self-confidence and independence among women. Although the response rates were not very high across communities, the majority of women did indicate newly acquired confidence, capabilities, and agency. Firstly, the act of providing the money to women allows them to have and control an income of their own. This ―salary‖, as some women call it, gives them a strong sense of freedom allowing them to no longer depend exclusively on money provided by their husbands. The management of the money is also seen as their ―right‖, a right that was granted to them by the government and cannot be changed. Secondly, the conditionalities attached to the program provide women with a negotiation tool they can use to have a more effective and stronger voice in household-decisions. Thirdly, during the application process and later during collection of CCT money, women are required to interact with institutions. This is something new for many and can be intimidating at times and yet, a number of them expressed feeling more confident after the experience knowing that they are now more able to interact in the social world. Hanife A from Yazilar noted: ―My mother-in-law does not believe that I can go there alone without help. On the other hand, I believe I can. Nobody helped me when I was applying for this money. I took the documents to related offices, I went to government building and to Bank, and I did all these by myself!‖

Box 11.3 CCT impact on women in Van, but constrained (Adato et. Al. 2007) Halime T from YolaĢan cannot leave the house unless she is accompanied by a male relative. But for her, the outings related to the CCT have taken on a new form. Even though her ability to travel on her own has not changed, she believes the CCT has given her more control over outside interactions as her husband or sons are not able to execute the collection of CCT money without her: ―Men generally stay out of the house. We stay in. But now Yusuf needs to take me to town to get money. I like that. Although I am giving money to him, now I can ask him to buy something. But other than that there is no change. What would change? Money is the men‟s business. We give the money to them anyway.‖ CCT has increased the need for her involvement in outside occurrences. Yet, ultimately she perceives that little has changed because money continues to belong in men‘s sphere of influence.

Although the program has provided a sense of independence and confidence to some women, it is not easy to measure the extent of it because of the resistance among individuals and communities to allowing such gender-specific transformations to openly occur. In addition, as long as management of money is seen as an exclusive male responsibility (particularly seen in Van as illustrated in Box 11.3), it is difficult for women to take advantage of all opportunities provided by the program. This is not to say it has not effected positive changes in the lives of women in and outside the house. In fact it has the potential to effectively reverse some of the factors that are limiting its impact— through schooling and by bringing women into what is conventionally seen as a men‘s world. Spousal relations did not reflect the level of change that might be expected from both an increase in household economic resources and the different social dynamics set in motion by selecting women as recipients. Nevertheless a few changes were noted, both positive and negative. Some households reported less conflict because of an increase in household income, while others mentioned disputes over who was entitled to keep the CCT money. This could indicate one of two things: either little has changed because money continues to belong in men‘s sphere of influence or small changes in decision-making dynamics have less

77

of an effect on wider aspects of intra-household relations. Gender roles and relations are quite sharply defined in Turkey and there is a strong socio-cultural investment in them — men are sensitive to even subtle challenges and women are mostly unaware of another perspective or unwilling to engage it explicitly. Nevertheless, the subtle changes we found are significant in this context.

78 The CCT program has been fully operational across Turkey since 2004. However, about 44 percent of CCT-education beneficiaries and 63 percent of health beneficiaries joined the program in 2005. At the macro level, the CTT focused on: (a) Developing a rationalized targeting system based on a proxy means test,; (b) Allowing access to benefits linked to behavior changes to promote human capacity, especially with regard to the role of women and the use of education and health facilities; and (c) Developing a monitoring and evaluation system. These objectives have largely been achieved in the order listed above and represent substantial accomplishments for Turkey. Despite a number of institutional constraints, it appears that the CCT has made a contribution at the level of individual families, especially with regard to: (a) Meeting immediate ad hoc needs of the poor, especially for food, during a period of financial stress; (b) Making important contributions to household income in the least developed regions and among families with large numbers of children; (c) Increasing enrollment rates and allowing children to remain in schools; and (d) Encouraging the poor families to make effective use of existing health facilities. In addition, several other impacts were found in the First Qualitative Study., through interviews and by joint interpretation of existing statistical data on poverty in the : (a) Narrowing the development gap between Turkey‘s richer and poorer regions; (b) Increasing consumption of local foodstuffs; (c) Encouraging more formal ―citizenship‖; and (d) Improving the status of women. The quantitative evaluation provides considerable evidence that the CCT program has had substantial impacts on a number of key education and health outcomes. The program is responsible for large and significant improvements in secondary school enrollment for girls and for children in rural households, increased immunization rates, and increased food consumption for some households. This is an impressive array of impacts for a CCT program of this size. Quantitative results also suggest that improvements in operation of the program and some adjustments in program design could lead to an expansion of impacts in the future. Logistical and administrative difficulties at the start of the CCT program meant that many beneficiary households did not begin receiving transfers until more than a year after the formal start of the program. As a result, the duration of time spent in the program was one year or less for many households in the sample. Overall, the quantitative analysis suggests that the program has had significant positive impacts on school participation at the secondary level, but not at the primary level. This problem arises in part because primary school enrollment is already high. As a result,

79

transfers to many households with primary school-age children were inefficient in that they could not change human capital investment behavior: most of these households would have sent their children to primary school even without the CCT program. This argues for shifting program resources towards supporting additional increases in secondary school enrollment. The extent of this shift in priorities should be managed to reflect the relative social returns to additional improvements in primary versus secondary school participation.

Stakeholder Perspectives on the Impact of CCT

In the First Qualitative Assessment, stakeholders hold mixed but generally positive views concerning CCT and suggest specific improvements. Beneficiary households argue in favor of nutrition and education impacts. Fathers generally are content with the enhanced integration of their wives into the broader society and supportive of the education of their daughters, provided that labor market opportunities for children did not fetch good returns and the schools are reasonably accessible. Girls are particularly happy about the support, especially if they are over 10 years of age. Educators invite greater inclusion and support CCT despite the additional burden it imposes upon them. In the Second Qualitative and Anthropological Study, beneficiaries and key informants were mostly enthusiastic about the benefits of the CCT program, emphasizing that the transfers help them to feed their families and send their children to school. Key informants are much more uniformly positive than the beneficiaries about the conditional nature of the benefits: while all key informants agreed that conditioning the aid on health and educational requirements was a useful step in achieving improvements in human capital indicators, beneficiaries were more divided. However, we should note here that there is a strong regional dimension to this finding: almost all the ‗negative‘ responses to conditionality came from Van. Some adverse program impacts were also raised, particularly in the First Qualitative Assessment (Kudat et al. 2006). Official stakeholders and social commentators have expressed concern that the CCT: (i) significantly increases the burden on health and education institutions;; (ii) increases antagonism against local government institutions when issues arise in CCT implementation; (iii) spreads expectations throughout the country that the poor could rely on the state, one way or the other. However, such expectations were not reflected among actual beneficiaries, and where it was, was expressed as appreciation that the state is there to give them a little bit of help. With respect to the issue of whether the program should continue, the First Qualitative Assessment found that there is a general belief that the program is useful and should continue, with improvements in implementation arrangements, to ensure that short-term outcomes/impacts are achieved. Several reasons are cited in support of this: Beneficiary families consistently state that they cannot afford to send their girls to schools outside their communities unless CCT support continues; this is especially so for high school. Many families also state that they would send their children to school regardless of the availability of CCT support, although with less conviction for their daughters than for their sons, and more so with regard to primary education than for secondary education. A large number of ―destitute‖ families have no access to regular financial support so that CCT is often a unique and vital factor in their lives. In these cases, CCT

80

support is used mostly for food and basic needs, and is a visible factor in their ability to send their children to schools outside the immediate community. Most officials and Foundation staff believe that the program has contributed to increased schooling and preventive health care and add that it should not have been initiated if it were to discontinue within a few years, as this would mean local protests against the Foundations. Other stakeholders also believe that it would be unfair to terminate the program as many among the beneficiaries and applicants are destitute and applications continue to be received in every Foundation office. Many stakeholders and nonbeneficiary applicants believe that a large number of families who are as poor as those benefiting from them are currently excluded. Many stakeholders, including beneficiary and nonbeneficiary applicants, believe the widespread lack of employment opportunities in the least developed regions makes it imperative to continue to have CCT or a CTT-like mechanism to support the poorest. Stakeholders, including the beneficiary families, are nevertheless cognizant of the fact that the state cannot support a large number of families forever. Many official stakeholders suggest changing the composition of state support so that a large amount of resources are allocated for employment creation; applicant families, especially men, also claim that they prefer work to other types of support. The overall conclusion of the Second Qualitative and Anthropological Study (Adato et al. 2007) was that the CCT program an important, valued program that is having an impact, and should be continued. There are wide regional, local, and individual variations in effectiveness and in attitudes toward the program; however, it would be hard to conclude that the net outcome is not positive. The CCT tackles the immediate effects of extreme poverty, encourages investment in human capital for the long terms, and challenges deeply entrenched values—if with varying success rates at different paces. On all of these fronts, for different reasons, it takes time and patience to make a difference. The Second Qualitative and Anthropological Study concluded that the program faces two major sets of challenges: the first is operational—the need for much stronger communications across all levels of program operations and primarily between program staff (Foundations, but they need a new form of assistance) and beneficiaries, so that beneficiaries understand the basic logic of the program—their obligations and entitlements. These improvements are within the capacity of the program—though it needs full buy-in and cooperation from the health and education sectors—and will result in significant program improvements and impacts. The second challenge is social and cultural, and these factors are harder to respond to. This will also require intersectoral cooperation and complementary approaches. With respect to both challenges, patience is also required—an understanding that social change takes time—but persistence and dedication to resolving problems from central to local levels will be needed for the program to succeed.

Recommendations

The CCT program—which has been successful in raising girls‘ enrollment in secondary school, boosting immunization coverage, and increasing household food consumption—could be even more successful. Recommendations for improvement are divided into three areas:

81

(i) meeting and enhancing program objectives; (ii) improving program implementation; and (iii) strengthening program monitoring.

Meeting and Enhancing Program Objectives The objective of improving the ability of the poor to deal with poverty will be enhanced by: 1. Continuing the program for existing beneficiaries until there is more evidence that the considerable progress made so far is self-sustaining. Support for the poor will be needed over the medium to long term, especially in Eastern and Southeastern Anatolia where poverty is widespread and the opportunities to expand employment are limited; 2. Securing additional financing so that all households that falls within the poorest six percent of the population is met, while maintaining benefits for the poorest four percent currently served. The fact that the program has already reached two thirds of the six percent target is a cause for celebration not criticism considering that the nation-wide implementation has a short history of about two years and that a smaller number of beneficiaries were targeted by the SRMP. 3. Ensuring that those currently outside the program who are demonstrably as poor as those benefiting from the program are accepted into the program. Measures such as the expeditious review of the backlog of applications, a rapid review of current status of all beneficiaries and the deletion of mistaken inclusions would help enhance program‘s ability to reach the remaining poorest households given the success of the initial Project and the expectations of the poor families; 4. Considering the inclusion of households connected to a social security system but are nevertheless poor for reasons of family size and/or disability of family members. In this context, considering continuing CCT benefits of households participating in other sub-programs of the Project, especially in temporary employment activities. 5. Continuing to review the appropriateness of application forms and procedures as well as the appeals process to reduce the burden on applicants, beneficiaries, education and health institutions and Foundation staff; 6. Improve the predicting power of the proxy means test model to more accurately identify the poorest. Some of the criteria in the proxy means test should be reconsidered, and some regional and/or rural/urban variation built into the system to account for unanticipated biases. More information should be gathered from Foundations to better understand these biases and determine where there may be needed changes to proxy means criteria. These actions will further improve the targeting effectiveness of the CCT program by reducing the errors of inclusion and exclusion. With respect to education: 7. Continue to provide education support to secondary-school-age girls until there is evidence that the gender gap in education has been eliminated, secondary school enrollment rates have reached the targeted levels, and dropout rates have been reduced. 8. Consider raising CCT payment levels for secondary-school, and in particular, for secondary school-age boys. The secondary school dropout rate is extremely high for boys, even higher than that for girls. In fact, evidences from Turkey and other countries suggest that child labor is much more prevalent among boys, particularly for older

82

boys, than among girls; and hence, the opportunity cost of attending school is higher for boys. The current levels of CCT payments cover the direct and indirect costs of education for secondary school girls, but not for boys. 9. The amount of the grant should be reconsidered, with consideration given to regional variations in cost of living and cost of schooling. 10. Consider limiting education transfers for primary school children to a set number of children per family. The exact number would have to be a decision made by the government, with attention to cultural and political issues. However, although the number of children would be capped, a condition could be added that, in order to be eligible for CCT, all primary-school-age children (aged 6-14) in the family must attend school. This modification of program design will make additional funds available for raising payment levels for secondary school, and eliminate any accusations that the program causes increased pregnancy rates. However, given the extreme difficulty in understanding the program structure and relationship of conditionalities to benefits, this change runs the risk of exacerbating confusion and undermining impacts. If implemented, then the persistent and effective communication is all the more imperative. 11. In areas with very high rates of primary school enrollment (before the program started) consideration should be given to stopping the primary school benefit and increasing the amount for secondary school participation. However, this should be done on a regional/provincial basis—in provinces where there are gaps in primary school enrollment the CCT is important to giving children the primary education needed to continue to secondary, or at least basic education. Also, such as change should involve no longer admitting new families with primary school age children, rather than cutting off current beneficiaries. 12. A review should be done of the location of secondary schools and transportation options, and the program should work with the appropriate government departments to improve these options. In poor regions with low enrollment, and where increasing enrollment is a high priority, a three-part strategy should be considered: (1) Building more secondary schools in rural areas and otherwise closer to where people live, especially in particularly conservative regions where travel for girls is very restricted; (2) In areas where it is not feasible to build a new school (or until such time as it is) use primary school facilities (and possibly some staff) for secondary school in the afternoon; (3) Where secondary schools can not be brought to a locality, and transportation is insufficient, improving transportation systems that are responsive to local conditions. This means that they may vary by region or location. In some areas free or subsidized bus travel will be adequate; others may require a smaller vehicle with a locally known driver. Although the provision of transportation is generally the responsibility of a different branch of government, in areas not covered it might be possible for another (poverty alleviation) program to generate employment through small, private transportation services where communities vet the drivers. These second two approaches could also enable small vehicles to carry boys and girls separately, as a local decision that the state does not specifically promote (we recognize that separate buses for boys and girls would not be desirable, but our research does indicate that an informal solution such as this could overcome some households‘ reluctance. (4) An increase in dormitories in areas where families are willing to send their girls to dormitories—this will only work in some regions and not others.

83

13. The Education department should invest more to increase school safety, security, and quality, particularly in poor communities experience problems in this area. This provides parents and students themselves with incentives for school attendance that may be stronger than money. Consideration should also be given to making social workers available where needed to deal with children‘s problems. 14. The program should coordinate with other programs that promote awareness raising, in the area of girls education, health care, and possibly family planning—though this last area should only be undertaken where it will not alienate people from the other messages. Without changes in attitudes, the cash incentive of the CCT will not be sufficient. Where possible, in tight-knit, conservative communities efforts to enlist support of ‗influential people‘ (in tribal or other terms) e.g. with respect to allowing girls to go to school, may be helpful. With respect to health: 15. The SRMP should collaborate with the health sector to promote education on primary health care, to make families more likely to participate in the services. This would be an opportunity to promote learning and also to dispel rumors that circulate, e.g. about vaccines and fertility, etc. Education on nutrition and hygiene could take place at the same time. This can be done in conjunction with providing specific program information with respect to the health component and purpose of the grant. This might also be a place for a gentle message with respect to control of the grant by women (we are not sure of how else or whether to address the finding of women turning over the money to men; this would again require regional variation). In CCTs in Latin America, participating in health and nutrition education, in the form of attendance at monthly talks given by a health professional, are a condition of the program. This should be considered as an added condition, as it is one of the main ways to ensure longer term improvements in health and nutrition practices. However, in many areas gathering women in groups is not possible, and it may make sense to provide additional information when women attend health controls. Options should be explored with those in Turkey who have developed pedagogical modalities for promoting awareness among women. 16. With respect to the pregnancy benefit, we are hesitant to make recommendations. There is no convincing evidence of need for a change; however, limiting the pregnancy benefit to the first five children (or allow the number to vary regionally, so that larger families are supported where they are they norm). This would reduce hostility to the pregnancy component from those in Foundations who are already deciding not to implement it, and from health staff and family planning advocates who feel their efforts are undermined. Families that are already larger should be grandfathered in and any limitation should only apply to new children. 17. Efforts in other spheres of government should continue or accelerate to maximize investment and job creation, to reduce the need for the grant and to provide additional incentive to continue schooling. This is of course not within the responsibilities of the SRMP, but the program should be aware of the interconnectedness of different development strategies and, as part of government, advocate for these multi-sectoral approaches.

84

Improving Program Implementation

18. Urgent attention should be given to improving systems of communications so that beneficiaries are well informed of the program logic, structure and conditions. Health and education staff and local leaders also need to be better informed about program structure and conditions so that they do not perpetuate misunderstandings. An improvement must thus be made in the design and implementation of systems of communications between the central program office and local Foundations, and between Foundations, muhtars, and beneficiaries, under the current basic communications structure. 19. Additional communications systems should also be considered; for example, the establishment of another level of liaison between Foundations and beneficiaries that can assist the beneficiaries and muhtars with local communications of program information: benefits structure and conditions; basis for beneficiary selection decisions; when the money has arrived; and other issues, as well as be available to assist the Foundation in answering questions. This liaison system has been an essential function in CCT programs in other parts of the world and has been widely successful. The appropriate structure for Turkey would need to be carefully thought through: whether it is volunteer or paid, what type of person would be viable (e.g. teachers), where (e.g. schools, individual homes, kiosk), what type of information, etc. The approach can vary at a regional level. Different systems could be tested on a pilot basis and scaled up if successful. 20. Establishing a permanent program of staff training and exchanges of information and experiences as it is clear that most ideas on program improvement exist already at the staff level and simply needed to be harvested, adopted and disseminated. 21. Strengthening capacity building efforts and increasing staff levels in high CCT concentration areas and where large numbers of actual and potential applicants exist. 22. Starting a pilot program for decentralizing beneficiary identification and evaluation functions. 23. Providing more resources for collaboration with education and health sector staff, including developing joint participatory strategies for improving human capital more broadly among the poor and for reducing supply side constraints in these key sectors. 24. Eliminate CCT payment delays and fluctuations in payment levels. Promote the use of electronic ATM cards for CCT payments, which will enable the beneficiaries to easily withdraw the payments and to check the balance. This system will greatly facilitate timely payment disbursements by Ziraat Bank. 25. Station a representative from the Foundation (or a new community liaison) at a ‗help desk‘ at the payment points to answer beneficiary questions about the amount of the grant received. They could be available for the first three days after funds have arrived. 26. Consider holding periodic meetings at schools, not long after payment delivery but also in cases of delays, to provide the community with information about the program and answer any questions they may have. Strengthening Program Monitoring and Evaluation

27. Improve the monitoring system for the CCT program. Monitoring needs a substantial degree of attention. A good monitoring system can go a long way toward solving many

85

operational problems identified in both quantitative and qualitative evaluations of the CCT program. Specifically, a reliable system is needed for monitoring CCT payment disbursements to beneficiaries. 28. Reorganize the Management Information System (MIS) database and increase resources in this area so that this facilitates implementation. The database should be easily accessible for assessing program performance. Appropriate training of the MIS staff at the headquarters and field offices is needed to improve their technical skills to efficiently operate the MIS. We understand that such training courses already have been undertaken by SRMP staff, as recent SRMP documents suggest. 29. More resources are devoted to program monitoring so that data on key program activities, incentives and behavior are available more rapidly and regularly. Purchasing, installing and custom-tailoring a ‗just-in-time‘ software to generate a set of policy relevant results may be given priority; 30. The data are evaluated more systematically and evaluation results are fed into program management at the local, regional and national levels; 31. A permanent program of social assessment is established at the local level to provide information on important beneficiary variables and on program impact; 32. More resources are used for mass communications, especially on program impact; As there is no evidence of widespread misuse of resources, consideration should be given to lightening the fiduciary framework that is not needed for improving program performance. 33. Consider commissioning a second evaluation of CCT program in 2008. The information generated in the current evaluation in the early stage of the program (2005- 06) would provide an important baseline for the second evaluation.

86

Adato, M., T. Roopnaraine, N. Smith, E. Altinok, N. Çelebioğlu, and S. Cemal. 2007. An Evaluation of the conditional cash transfer program in Turkey: Second Qualitative and Anthropological Study. Project Report. Washington, D.C.: International Food Policy Research Institute. Ahmed, A. U., D. Gilligan, A. Kudat, R. Colasan, H. Talidil, and B. Ozbilgin. 2006. Interim impact evaluation of the conditional cash transfer program in turkey: a quantitative assessment. Project Report. Washington, D.C.: International Food Policy Research Institute. Behrman, J., and J. Hoddinott. 2000. An evaluation of the impact of PROGRESA on preschool child height. Washington, D.C.: International Food Policy Research Institute. Behrman, J., P. Sengupta, and P. Todd. 2000. Progressing through PROGRESA: An impact assessment of a school subsidy experiment. Washington, D.C.: University of Pennsylvania and the International Food Policy Research Institute. Burtless, G. 1995. The case for randomized field trials in economic and policy research. Journal of Economic Perspectives 9 (2): 63-84. Gertler, P. J. 2000. Final report: The impact of PROGRESA on health. Washington, D.C.: International Food Policy Research Institute. Hoddinott, J., E. Skoufias, and R. Washburn. 2000. The impact of PROGRESA on consumption: A final report. Washington, D.C.: International Food Policy Research Institute. Kudat, A., with contributions from H. Tatlidil, B. Ozbilgin, C. Baykal, M. Adato, and A. U. Ahmed. 2006. Evaluating the conditional cash transfer program in Turkey: A qualitative assessment. Project Report. Washington, D.C.: International Food Policy Research Institute. Maluccio, J. A., and R. Flores. 2005. Impact evaluation of a conditional cash transfer program: The Nicaraguan Red de Protección Social. Research Report 141. Washington, D.C.: International Food Policy Research Institute. Morley, S., and D. Coady. 2003. From social assistance to social development: Targeted education subsidies in developing countries. Washington, D.C.: Center for Global Development and International Food Policy Research Institute. Morris, S. 2005. ―Conditional cash transfer programs and health.‖ Food Consumption and Nutrition Division, International Food Policy Research Institute, Washington, D.C. Photocopy. Parker, S., and E. Skoufias. 2000. Final report: The impact of PROGRESA on work, leisure, and time allocation. Washington, D.C.: International Food Policy Research Institute. Rawlings, L. B. 2004. New approach to social assistance: Latin America‘s experience with conditional cash transfer programs. Washington, D.C.: World Bank. Schultz, T. P. 2000a. Impact of PROGRESA on school attendance rates in the sampled population. Washington, D.C.: International Food Policy Research Institute. Schultz, T. P. 2000b. School subsidies for the poor: Evaluating a Mexican strategy for reducing poverty. Washington, D.C.: International Food Policy Research Institute.

87

Schultz, T. P. 2000c. Final report: The impact of PROGRESA on school enrollments. Washington, D.C.: International Food Policy Research Institute. Skoufias, E. 2005. PROGRESA and Its Impacts on the Welfare of Rural Households in Mexico. IFPRI Research Report No. 139. International Food Policy Research Institute, Washington, DC. Social Risk Mitigation Project (SRMP). 2005. Project Operational Manual. Volume 2. Conditional Cash Transfers. World Bank. 2005. Turkey: Joint poverty assessment report. Report No. 29619-TU. Human Development Sector Unit, Europe and Central Asia Region. Washington, D.C.: World Bank. Yap, Yoon-Tien, G. Sedlacek, and P. Orazem. 2001. Limiting child labor through behavior- based income transfers: An experimental evaluation of the PETI program in rural Brazil. Washington, D.C.: World Bank.

88

89

Table A.1 — Selected provinces and districts for the household survey, and the eligibility status of applicants Eligible for CCT Eligible by Eligible upon Total Province District Applicants Ineligible PMT score appeal eligible (number) ĠSTANBUL SULTANBEYLĠ 3,042 1,495 1,547 0 1,547 K.ÇEKMECE 1,508 810 683 15 698 EDĠRNE UZUNKÖPRÜ 1,688 439 723 526 1,249 MERKEZ 1,951 226 710 1,015 1,725 BALIKESĠR MERKEZ 2,091 1,574 492 25 517 ĠVRĠNDĠ 1,599 1,136 463 0 463 ĠZMĠR KONAK 1,723 1,225 498 0 498 ALĠAĞA 327 93 171 63 234 AYDIN MERKEZ 2,429 1,262 1,167 0 1,167 ĠNCĠRLĠOVA 505 246 258 1 259 MANĠSA MERKEZ 3,714 2,186 1,528 0 1,528 TURGUTLU 1,690 807 836 47 883 BURSA OSMANGAZĠ 3,030 2,067 940 23 963 KARACABEY 725 213 512 0 512 SAKARYA MERKEZ 3,238 1,633 1,380 225 1,605 AKYAZI 752 372 357 23 380 ANKARA ALTINDAĞ 6,923 3,806 3,111 6 3,117 ÇANKAYA 2,400 1,396 1,004 0 1,004 KONYA EREĞLĠ 3,365 1,953 1,412 0 1,412 KARATAY 955 467 488 0 488 ANTALYA MERKEZ 2,216 1,081 1,135 0 1,135 SERĠK 933 369 556 8 564 ADANA YÜREĞĠR 8,472 3,147 5,325 0 5,325 CEYHAN 2,113 716 1,397 0 1,397 K.MARAġ MERKEZ 7,737 3,261 4,476 0 4,476 TÜRKOĞLU 3,060 1,644 1,416 0 1,416 NĠĞDE MERKEZ 5,221 2,061 3,108 52 3,160 ÇĠFTLĠK 2,034 287 1,747 0 1,747 SORGUN 2,747 1,992 748 7 755 SARAYKENT 1,011 533 478 0 478 ZONGULDAK EREĞLĠ 1,019 383 636 0 636 ALAPLI 588 252 336 0 336 SĠNOP DURAĞAN 2,114 276 1,834 4 1,838 BOYABAT 549 203 289 57 346 SAMSUN MERKEZ 6,465 3,400 3,048 17 3,065 TEKKEKÖY 1,326 793 533 0 533 ORDU GÖLKÖY 2,221 1,090 1,130 1 1,131 ULUBEY 885 349 536 0 536 ERZURUM MERKEZ 2,334 590 1,744 0 1,744 KÖPRÜKÖY 1,397 843 554 0 554 AĞRI MERKEZ 4,775 1,653 3,122 0 3,122 HAMUR 1,321 291 1,025 5 1,030 ELAZIĞ MERKEZ 13,946 7,435 6,511 0 6,511 KOVANCILAR 2,349 424 1,862 63 1,925 VAN MERKEZ 23,155 10,426 12,729 0 12,729 GÜRPINAR 3,509 452 3,012 45 3,057 GAZĠANTEP ġAHĠNBEY 18,385 12,725 5,660 0 5,660 NĠZĠP 6,247 2,882 3,365 0 3,365 DĠYARBAKIR MERKEZ 13,449 2,247 11,201 1 11,202 ERGANĠ 4,903 1,371 3,532 0 3,532 BATMAN MERKEZ 15,109 3,369 11,727 13 11,740 BEġĠRĠ 2,149 235 1,910 4 1,914 TOTAL 207,394 90,186 114,962 2,246 117,208 Source: Based on data from the MIS database of the SYDGM.

90

Nonbeneficiary- District Province District Beneficiaries applicants total Province total

ĠSTANBUL SULTANBEYLĠ 22 21 43 64 K.ÇEKMECE 11 10 21 EDĠRNE UZUNKÖPRÜ 9 7 16 30 MERKEZ 8 6 14 BALIKESĠR MERKEZ 15 13 28 51 ĠVRĠNDĠ 13 10 23 ĠZMĠR KONAK 14 10 24 31 ALĠAĞA 4 3 7 AYDIN MERKEZ 18 15 33 41 ĠNCĠRLĠOVA 4 4 8 MANĠSA MERKEZ 28 24 52 77 TURGUTLU 14 11 25 BURSA KARACABEY 23 18 41 55 OSMANGAZĠ 7 7 14 SAKARYA MERKEZ 23 20 43 54 AKYAZI 6 5 11 ANKARA ALTINDAĞ 53 43 96 135 ÇANKAYA 21 18 39 KONYA EREĞLĠ 25 19 44 56 KARATAY 7 5 12 ANTALYA MERKEZ 17 14 31 45 SERĠK 8 6 14 ADANA YÜREĞĠR 65 55 120 150 CEYHAN 17 13 30 K.MARAġ MERKEZ 62 47 109 154 TÜRKOĞLU 24 21 45 NĠĞDE MERKEZ 40 33 73 104 ÇĠFTLĠK 18 13 31 YOZGAT SORGUN 20 17 37 52 SARAYKENT 8 7 15 ZONGULDAK EREĞLĠ 8 7 15 23 ALAPLI 5 3 8 SĠNOP DURAĞAN 16 14 30 39 BOYABAT 5 4 9 SAMSUN MERKEZ 49 41 90 109 TEKKEKÖY 11 8 19 ORDU GÖLKÖY 18 14 32 45 ULUBEY 7 6 13 ERZURUM MERKEZ 16 20 36 54 KÖPRÜKÖY 9 9 18 AĞRI MERKEZ 39 38 77 95 HAMUR 9 9 18 ELAZIĞ MERKEZ 106 88 194 224 KOVANCILAR 16 14 30 VAN MERKEZ 173 145 318 367 GÜRPINAR 26 23 49 GAZĠANTEP ġAHĠNBEY 142 116 258 346 NĠZĠP 47 41 88 DĠYARBAKIR MERKEZ 104 86 190 260 ERGANĠ 40 30 70 BATMAN MERKEZ 116 95 211 244 BEġĠRĠ 17 16 33 TOTAL 1,583 1,322 2,905 2,905 Source: Based on data from the MIS database of the SYDGM.

91

Beneficiary Province District Education Health District total Province total (number) ĠSTANBUL SULTANBEYLĠ 2 1 3 59 K.ÇEKMECE 38 18 56 EDĠRNE UZUNKÖPRÜ 5 2 7 10 MERKEZ 2 1 3 BALIKESĠR MERKEZ 1 1 2 5 ĠVRĠNDĠ 2 1 3 ĠZMĠR KONAK 41 20 61 61 ALĠAĞA 0 0 0 AYDIN MERKEZ 4 2 6 7 ĠNCĠRLĠOVA 1 0 1 MANĠSA MERKEZ 9 5 14 38 TURGUTLU 16 8 24 BURSA OSMANGAZĠ 0 0 0 2 KARACABEY 1 1 2 SAKARYA MERKEZ 1 1 2 4 AKYAZI 2 0 2 ANKARA ALTINDAĞ 20 10 30 40 ÇANKAYA 7 3 10 KONYA EREĞLĠ 4 1 5 22 KARATAY 12 5 17 ANTALYA MERKEZ 0 0 0 0 SERĠK 0 0 0 ADANA YÜREĞĠR 12 6 18 18 CEYHAN 0 0 0 K.MARAġ MERKEZ 0 0 0 19 TÜRKOĞLU 13 6 19 NĠĞDE MERKEZ 16 8 24 36 ÇĠFTLĠK 8 4 12 YOZGAT SORGUN 5 3 8 8 SARAYKENT 0 0 0 ZONGULDAK EREĞLĠ 0 0 0 0 ALAPLI 0 0 0 SĠNOP DURAĞAN 3 1 4 4 BOYABAT 0 0 0 SAMSUN MERKEZ 4 2 6 7 TEKKEKÖY 0 0 1 ORDU GÖLKÖY 9 5 14 17 ULUBEY 2 1 3 ERZURUM MERKEZ 11 5 16 17 KÖPRÜKÖY 1 0 1 AĞRI MERKEZ 13 6 19 47 HAMUR 19 9 28 ELAZIĞ MERKEZ 4 2 6 8 KOVANCILAR 1 1 2 VAN MERKEZ 143 71 214 220 GÜRPINAR 4 2 6 GAZĠANTEP ġAHĠNBEY 0 0 0 12 NĠZĠP 8 4 12 DĠYARBAKIR MERKEZ 6 2 8 44 ERGANĠ 24 12 36 BATMAN MERKEZ 17 8 25 39 BEġĠRĠ 9 5 14 TOTAL 503 247 750 750 Source: Based on data from the MIS database of the SYDGM.

92

Table A.4 — Geographical coverage of the assessment and the methodology used in the First Qualitative Study Province District Target Groups Study conducted Methodology

Yüreğir Mothers BA FG ADANA Kozan Service Providers (Education and Health) BA FG Village Primary School Children BA FG Central Key Stakeholders SH IDI AĞRI Community Members SH RCA ADIYAMAN Central Community Members SH RCA ANKARA Altındağ Community Members C IDI+RCA BĠTLĠS Central Key Stakeholders SH IDI BURSA Osmangazi Community Members SH RCA Central Community Members SH IDI URFA Bozova Key Stakeholders SH RCA Central Key Stakeholders C IDI+RCA+A ELAZIĞ Kovancılar Community Members SH IDI Central Key Stakeholders C IDI+RCA ERZURUM IIıca Community Members SH IDI ĠSTANBUL Sultanbeyli Key Stakeholders SH IDI Bornova Service Providers (Education) BA FG ĠZMĠR Bergama Primary School Children BA FG Village Fathers BA FG KONYA Karatay Community Members SH RCA Secondary School Boys and Girls (Day BA FG Central School) ORDU Fatsa Mothers BA FG Village Primary School Children BA FG Central Key Stakeholders C IDI+RCA SAKARYA Akyazı Community Members SH RCA Central Key Stakeholders C IDI+RCA VAN Gürpınar Community Members SH RCA Secondary School Boys and Girls BA FG Central (Boarding School) YOZGAT Sorgun Primary School Children BA FG Village Mothers BA FG A: Anthropological Studies SH: Stakeholder IDI: In-Depth Interview FG: Focus Group Assessment C: Comprehensive BA: Beneficiary RCA: Rapid Community (SH+CF) Assessment Assessment

93

Table A.5 — First Qualitative Study: Locations and participants of focus group meetings Province District Participants Central Secondary School Boys and Girls (Day School) ORDU Fatsa Mothers Village Primary School Children Bornova Service Providers (Education) ĠZMĠR Bergama Primary School Children Village Fathers Yüreğir Mothers ADANA Kozan Service Providers (Education and Health) Village Primary School Children Central Secondary School Boys and Girls (Boarding School) YOZGAT Sorgun Primary School Children Village Mothers

Table A.6 — Site selection and interviews conducted in the Second Qualitative and Anthropological Study Rural/ Ethnic Region Province District Mahalla/Koy Urban Diversity Blacksea Samsun Merkez Ilyaskoy Mahallesi (C) Urban Medium Blacksea Samsun Tekkekoy Yazilar Koyu (D) Rural Medium Eastern Anatolia Van Merkez Beyüzümü Mahallesi (E) Urban High Eastern Anatolia Van Gurpinar Yolsan Koyu (F) Rural High Southeastern Anatolia Diyarbakir Merkez 5 Nisan Mahallesi (A) Urban High Southeastern Anatolia Diyarbakir Ergani Fatih Mahalessi (B) Urban High

94

Table A.7 — Views of beneficiary households Education Health Description Beneficiaries Beneficiaries (percent) Sources of prior information about the CCT program Television 4.3 2.7 Schoolteacher 47.9 10.5 Muhtar 12.4 23.6 SYDV official 3.2 4.5 Newspaper 0.1 16.8 District administrative office 1.9 1.8 Friends/ neighbor/ relative 28.5 38.7 Others 1.8 1.4

How much do you know about the application and selection criteria of CCT program? Much 1.7 1.0 A little 33.0 29.9 Nothing 65.3 69.0

From where did you get the CCT application form? School 50.9 12.4 Muhtar 16.2 25.1 SYDV 20.7 28.9 Health center 1.0 19.1 District administrative office 5.2 6.5 Other 6.0 7.9

Who completed the CCT application form? Mother 17.3 15.0 Father 13.4 17.4 Family members together 25.3 18.9 With help from SDYV 5.4 7.1 With help from Muhtar 15.3 4.3 With help from others 23.4 21.3 16.0 Where did you submit the application? At district admin office 15.3 12.2 At SYDV center 37.8 46.2 At health center - 6.2 At school 36.9 16.8 To Muhtar 8.8 18.0 Other 1.2 0.6

Who submitted the application? Mother 81.0 74.8 Father, because mother was not present 13.0 21.5 Father, because mother doesn't go out 1.6 1.2 Other household member 3.5 1.6 Other (Not household member) 0.9 0.8 (continued)

95

Table A.7 - Views of beneficiary households (continued) Education Health Description Beneficiaries Beneficiaries (percent) Does the HH get benefit from green card or disability fund? Green card 85.2 77.2 Handicapped fund 0.8 6.7 No 13.9 16.1

How many of your primary education children get benefit from CCT program? 0 8.3 - 1 30.3 - 2 34.3 - 3 19.3 - 4 6.3 - 5 and above 1.5 -

How many of your secondary education children benefit from CCT program? 0 76.0 - 1 17.8 - 2 5.4 - 3 and above 0.8 -

Do you think that the government should provide in-kind support for education instead of cash support? In-kind support 4.8 - Go on with CCT 95.2 -

How do you get CCT money? Ziraat Bank (With bank account) 22.8 22.4 Ziraat Bank (With transfer to name) 41.4 55.8 Ziraat Bank (With ATM card) 1.2 1.9 Post office 17.9 12.4 School 14.7 5.3 Other 1.9 2.2

Under whose name is the Bank account/ATM card/postal account? Mother 92.6 95.7 Father 3.3 3.2 Other 4.2 1.1

How many times did you receive CCT money since your child was enlisted in the CCT program? 0 1.5 2.0 1 10.1 18.9 2 21.3 28.0 3 22.8 23.1 4 17.5 10.9 5 11.7 9.6 6 7.7 3.8 7 times and above 7.3 3.6 (continued)

96

Table A.7 - Views of beneficiary households (continued) Education Health Description Beneficiaries Beneficiaries (percent) Interval of getting CCT money Once in every month 1.0 0.2 Once in every two months 18.8 20.0 Irregular 80.2 79.8

Are you pleased with the payment intervals? Yes 15.3 13.3

How should CCT payments be? Once a year 6.5 6.6 Once in every 6 months 0.1 0.4 Once in every 3 months 2.7 2.6 Once a month 88.5 88.8 Other 2.2 1.5

Who collects the CCT money? Child's mother 96.9 97.4 Child's father 2.1 2.2 Other 1.0 0.4

Has CCT payment been stopped because failed to comply with the conditions? Yes 8.9 11.7

For which condition was CCT payment stopped? Lack of school attendance 53.1 - Lack of child‘s performance in school 9.4 - Not taking children to health center regularly 6.3 63.6 Other 31.3 36.4

Has CCT payment been stopped even though complied with the conditions? Yes 6.5 9.8

Does the respondent know the criteria which his/her children have to fulfill in order to maintain eligibility for CCT? Knows the criteria 8.1 7.4 Does not know the criteria 68.4 72.8 Knows a little 23.4 19.8

Reasons for not knowing the criteria for maintaining eligibility Nobody informed the rules of the program 89.5 86.9 Was not interested in learning 8.2 7.7 Other 2.3 5.4

If CCT program is stopped for boys, what would you do? Won't send children to school or to clinics for regular health checkups 15.0 20.1 Won't send children to school after compulsory education 10.1 - Will do nothing, nothing will change 7.5 14.2 Will send children to school or to clinics for regular health checkups but with difficulties 67.4 59.8 Other - 5.9 (continued)

97

Table A.7 - Views of beneficiary households (continued) Education Health Description Beneficiaries Beneficiaries (percent) If CCT program is stopped for girls, what would you do? Won't send children to school or to clinics for regular health checkups 17.6 19.5 Won't send children to school after compulsory education 10.1 - Will do nothing, nothing will change 6.1 11.4 Will send children to school or to clinics for regular health checkups but with difficulties 66.2 63.0 Other - 6.2

Who spends the CCT money? Mother 93.2 94.8 Mother receives the money but gives to father 5.1 4.8 Others 1.7 0.4

Is CCT money enough to cover all education or health expenses of the child (percent of households agreeing with the statement)? Yes 8.0 5.6

How is the quality of education your children receive at school? Very good 13.5 - Good 63.5 - Not bad 14.7 - Bad 6.0 - Very bad 1.1 - Cannot judge the quality 1.3 -

Why do you think your children receive bad quality of education at school? Too many students in classroom 46.7 - Poor quality of teachers 22.1 - Inadequate/poor school facilities 23.8 - Other 7.4 -

How much CCT money is your primary school boy supposed to receive per month? YTL 25.4 -

How much CCT money is your secondary school boy supposed to receive per month? YTL 37.2 -

How much CCT money is your secondary school girl supposed to receive per month? YTL 37.7 -

How much money in total have you received so far? YTL 365.7 YTL 186.5

Distance have to travel to collect CCT money (kilometers) 4.9 4.8

Commuting time to collect CCT-education money (hours) 2.3 2.5 Source: Based on data from ―Evaluating the Conditional Cash Transfer Program in Turkey: 2005-06 Household Survey.‖

98

Table A.8 — Views of all CCT beneficiaries Description Response (percent) Purpose of spending last CCT money (multiple responses) Made payment to school 13.1 Bought textbooks 16.7 Bought stationary 76.2 Bought clothing for children 70.4 For health care of children 43.0 For going to health center 17.2 For feeding of children 82.3 Kitchen expenses 55.3 Paid a part of house rent 5.6 Paid loan 8.3 Other 11.9

Have you ever heard that CCT program encourages parents to have more children? Yes 3.6

From whom did you hear that? Neighbors/ friends 70.8 Muhtar 20.8 Health center personnel 1.4 Schoolteachers 2.8 Other 4.2

Is any of the female household members pregnant? Yes 5.7

Do mothers think that CCT program encourages having more children? No 1.3

Do fathers think that CCT program encourages having more children? No 2.9

Is mother planning to have more children for getting more money from CCT program? No 0.9 Is father planning to have more children for getting more money from CCT program? No 1.9

Is any of the household members a smoker? Yes 52.4

Has cigarette consumption been increased since started receiving CCT money? Yes 1.6

Does any of the household members use alcoholic beverages? Yes 1.1

Has alcoholic beverage consumption increased since you started receiving CCT money? No 100.0

Does receiving CCT money discourage any of the household members from finding a paid job? No 100.0

Does receiving CCT money discourage any of the employed members from going to work? Yes 0.1 Source: Based on data from ―Evaluating the Conditional Cash Transfer Program in Turkey: 2005-06 Household Survey.‖

99

Table A.9 — Responses of nonbeneficiary-applicants Applicants for: Description Education Health (percent) Source of information about the CCT program Television 3.8 3.4 Radio 0.9 0.5 Schoolteacher 42.9 8.7 Muhtar 7.2 13.4 SYDV official 3.8 6.9 Health center 0.5 16.6 District administrative office 1.6 2.7 Friends/ neighbor/ relative 37.5 45.8 Other 1.8 2.1

From where did you get the CCT application form? School 50.7 10.5 Muhtar 13.8 20 SYDV 23.1 30.4 Health center 0.5 23.3 District administrative office 5.0 7.5 Other 7.0 8.2

Who completed the application form? Mother 23.7 22.2 Father 14.4 20.8 Family members together 24.1 16.6 With help from SDYV 6.1 7.9 With help from health center 0.0 6.3 With help from Muhtar 9.8 10.6 With help from others 21.9 15.5

Where did you submit the application? At district admin office 13.8 16.9 At SYDV center 40.5 48.7 At school 35.8 5.4 At health center 0.6 17.4 To Muhtar 7.6 10.0 Other 1.6 1.6

Who submitted the application? Mother 81.4 74.3 Father, because mother was not present 11.1 19.4 Father, because mother doesn't go out 2.2 2.3 Father, because mother is dead 0.3 0.0 Father, because parents divorced 0.1 0.2 Other HH member 3.8 2.5 Other (Not HH member) 1.2 1.4

Does the household get benefit from green card or disability fund? Green card 74.5 79.2 Handicapped fund 0.6 2.6 None 24.9 18.2 (continued)

100

Table A.9 - Responses of nonbeneficiary-applicants (continued)

Applicants for: Description Education Health (percent) Why do you think your children couldn't get CCT money? Did not comply with selection criteria 16.1 29.1 Selected but not beneficiary yet 72.8 64.6 Other 11.1 6.3

Is CCT selection process fair? Yes 11.6 14.3

Why do you think that the selection process is not fair? Many poor households are out of program whereas many relatively not poor households get selected 87.4 88.7 Selection grading is wrong 3.6 2.4 There are some biases 7.4 7.6 Other 1.6 1.2

Should state give school materials instead of cash? Yes 9.3 -

Response from all nonbeneficiary-applicants

Knows about the application and selection criteria of CCT program Much 1.4 A little 38.6 Nothing 60.1

Does any of the HH members have social security (multiple response)? None 96.0 SSK 8.7 Bað-Kur 1.1 Emekli Sandýðý 0.6 Other 1.3

Ever heard that CCT program encourages parents to have more children? Yes 5.4

Does the household have a pregnant woman? Yes 7.1

Do mothers think CCT program encourages to have more children? Yes 1.3

Do fathers think CCT program encourages to have more children? Yes 1.4

Do interviewed members think CCT program encourages to have more children? Yes 3.3 Source: Based on data from ―Evaluating the Conditional Cash Transfer Program in Turkey: 2005-06 Household Survey.‖

101

102

Figure B.1 — First Qualitative Study: List of key stakeholders

SOCIAL SOLIDARITY FOUNDATIONS * DAY SCHOOLS * BOARDING SCHOOLS - Primary - Secondary EDUCATION SERVICE - High SECTOR * CENTRAL SCHOOLS PROVIDERS

HEALTH SECTOR

LOCAL * GOVERNOR/DEPUTY GOVERNOR AUTHORITIES * DISTRICT GOVERNOR

* EDUCATION NGOs * HEALTH

* HEADMAN * VILLAGE COUNCIL * RELIGIOUS LEADER COMMUNITIES * SCHOOL TEACHER * COFFEE HOUSE ATTENDENTS * PUBLIC TRAINING CENTERS CHILDREN & SATFF

* LOCAL BRANCHES OF MEDIA NEWSPAPERS

BENEFICIARIES APPLICANTS THE POOR NONBENEFICIARIES

103

Figure B.2 — First Qualitative Study: Comprehensive study model

COMMUNITY CHOICES for COMMUNITY BASED KEY STAKEHOLDERS GROUP DISCUSSIONS SOCIAL SOLIDARITY FOUNDATIONS * DAY SCHOOLS PROVINCE (6) * BOARDING SCHOOLS EDUCATIO - Primary SERVICE CENTRE N SECTOR - Secondary PROVIDERS - High * CENTRAL SCHOOLS

HEALTH VICINITY I VICINITY II (high CCT (high poverty) SECTOR concentration) DISTRICT I (poorest) LOCAL * GOVERNOR/DEPUTY GOVERNOR AUTHORITIES * DISTRICT GOVERNOR VICINITY I VICINITY II (high CCT (high powerty) concentration) * EDUCATION NGOs * HEALTH

DISTRICT II (poorest) * HEADMAN * VILLAGE COUNCIL * RELIGIOUS LEADER COMMUNITIES * SCHOOL TEACHER * COFFEE HOUSE VILLAGE (high poverty) ATTENDENTS * PUBLIC TRAINING CENTERS CHILDREN VICINITY * LOCAL BRANCHES OF (high CCT concentration) MEDIA NEWSPAPERS

BENEFICIARIES APPLICANTS THE POOR NONBENEFICIARIES

104

105

Appendix C1: Key Questions for the Second Qualitative and Anthropological Study

RESEARCH QUESTIONS FOR 2006 CCT QUALITATIVE AND ANTHROPOLOGICAL STUDY (IFPRI/Agrin June 2006)

The following questions will be the focus of the anthropological studies. These topics have been developed based on a combination of 1) priorities identified by the Client; 2) issues known to be of interest/concern to CCT programs internationally; 3) issues identified from the 2005 qualitative study as in need of follow-up; 4) issues known to be of importance in Turkey more generally; 5) issues which best lend themselves to study through qualitative research (those better answered through survey methods are excluded). The one source from which questions have not yet been derived is the quantitative survey, as quantitative analysis has not yet been performed at this level. However, as questions arise from the quantitative data related to our key focal areas below, they may be added. Given the time frame of the study, it is not possible to cover every issue of interest; thus the topics and questions below represent a prioritization. Some issues are given stronger emphasis than others, e.g. Women‘s status and gender relations are a primary focus. Note that the questions below are still quite extensive, and represent those that will be pursued with best effort in the anthropological studies; whether and how they will be answered depends on the quality of the data that can be derived in the time period available, and how forthcoming informants are with responses.

The key focal areas of the research are: 1) Household and community description 2) Program communications 3) Beneficiary selection and targeting 4) Attitudes and perceptions with respect to government assistance, social programs and work 5) Education 6) Health and nutrition 7) Pregnancy 8) Use of the grants 9) Other social programs 10) Women‘s status and gender relations

The questions below are research questions for the study, not direct questions to be asked of respondents in this format. Interview instruments in the form of checklists have been developed for this purpose. These issues will be pursued in different forms, as relevant, will be asked of the following informants (household members include those in beneficiary and non-beneficiary households):

o Mothers o Husbands/fathers o Youth o Young children o Others in the household, e.g. grandparents, aunts, uncles o Health staff o School staff o Foundation staff

106

o Local leaders o Shopkeepers o Informal gatherings

1. General information (to be derived from quantitative survey and supplemented with qualitative data)

Livelihoods activities Income, cash and in-kind Assets Shocks Living conditions Household demographics Community-level data 2. Program communications

Do people know when their benefits arrive at the Bank?

o If not, why not? o Who is supposed to inform them? Who has informed them in practice? o Why do they decide to go to the Bank when they go? o What would be an effective system of communication?

Within one family, why are some children enrolled and not others? Why are families enrolled for education benefits and not health, or vice-versa? What are the observed interactions with Foundation staff and how do the nature of these interactions affect the program? 3. Beneficiary Selection and Targeting What do people understand to be the reason people are selected for the program? (e.g. Ties to the government (relatives or friends)? Poverty? Children out of school) What do they think of this basis? (Is it fair?) How are people actually selected? (What do we know about political affiliation, tribal relations, etc.) Do they think there are people who need money and not getting it, or people getting the money who do not need it? How do they think the selection should be done? What would be better methods for targeting? Why is it that some people do not apply for the program?

o Do they not hear about it? o Do they not like the idea of it or aspects of it? (e.g. Do husbands or other relatives object? Stigma of poverty; lost hope) o Do they think they‘re not eligible?

Are people who do not receive benefits unhappy about this?

107

Do non-beneficiaries know who the beneficiaries are? Are there tensions between people in the community because some people are included and others are excluded?

o What forms does this tension take? o What are the reasons for the tension (e.g. beliefs about political influence? Other unfairness; unhappiness over not receiving money)

How does the appeals process work in practice? Who is involved and what are the mechanisms? How responsive is it to appeals?

4. Attitudes and perceptions with respect to government assistance, social programs and work

Work: What are the attitudes of adults and young people about finding a job—their chances, what is needed to do this? Has the program affected their willingness to work? Has it changed their expectations about what type of work they want? Does program have any effect on parents‘ sending their children to work? Increasing it, reducing it or no effect? What do they think of the idea of the government giving people money without a work requirement? o Is there stigma attached to this? o How do they feel about this for themselves?

What do they think about the government giving people money, conditioned on health and education services?

5. Education

What is the level of education of parents and what were the reasons they stopped or continued to this level? How do levels of education and literacy of parents affect their responses to the program? Do their children participate in informal religious studies? Are they paid a fee to participate? Is it replacing their schooling? What are parents and children‘s attitudes toward the importance of education?

o How do they see the purpose of education? o What benefits will it provide the children in the future? o How do parents feel about their own levels of education and how it has affected their lives?

Has the program changed their attitudes toward the importance of education for children? For girls?

108

Are parents sending their children to school because they are getting money or because they believe schooling is important? (Does the first seem to have affected the other?)

o How long do parents intend to keep their children in school? What are their reasons (either for continuing or ending?) o What is the future that parents see for their children (For girls vs. boys) That children see for themselves?

If other programs related to for school rehabilitation or other programs, exists in the locality, how did this affect their attitudes toward school? What are children‘s attitudes toward their families, with respect to their own levels of education vs. other family members? How do teachers treat the children? o Do they treat beneficiary children differently than non-beneficiary children? o Are girls treated differently than boys?

Do children know they are beneficiaries and what do they understand this to mean? How do they feel about it? Does it generate stigma or pride? Have they improved their grades at school? What risks do they face at school? In getting to school? Are teachers regularly marking children as attending even when they do not? What are the reasons?

6. Health and Nutrition:

Health: What are the patterns of health care practices, at home, with traditional healers, or public clinics (expenditures, treatment of illnesses, growth monitoring, vitamins, vaccines, etc.)? What are local beliefs about health care, nutrition, fertility? How are these related to different government services? What do people know/believe about aspects of health care required by CCT? What are their attitudes toward these? Are people participating in the required health services (also for non-beneficiaries)? Why do people not participate in health services? What would be needed to provide better incentives for participation in health services? How do men‘s attitudes toward nutrition affect the behavior of women and children?

o What would help to change men‘s attitudes?

How are people treated with respect by the health staff?

109

How is the quality of the services? Are girls and boys treated differently by the health staff?

Nutrition: What are diets of adults and children? What difference is made by the CCT program? What are their attitudes toward the importance of good nutrition? Has the program changed their attitudes toward its importance for mothers and young children? How do men‘s attitudes toward nutrition affect the behavior of women and children?

o What would help to change men‘s attitudes?

7. Pregnancy

What are attitudes toward family size, toward # of children that people want to have? Has this been affected by the program? How accessible is birth control and are they using it? Does the program seem to be providing an incentive for women to get pregnant?

o Do people have this perception? Do they actually know anyone who has done this? o In order for the family to receive benefits now? o Because they think the government will support the children in the future?

What are the other reasons for families to have additional children?

8. Use of the grant

How is the money spent? Why do they choose these expenditures? Does anyone require them to spend it in a particular way?

9. Other programs

What other similar programs (health, education and nutrition, livelihoods) do people participate in? How do they compare them with the CCT? What other types of programs would they prefer?

110

10. Women’s Status and gender relations:

Status in household: What is the status of women, men, girls and boys in their households? How much power do they have? How to people perceive this status to have changed over time? What is the impact of the program on this status? Who makes decisions in the household about:

o Spending money earned by men? o Spending money earned by women? o Spending money earned by children? o Spending CCT benefits? o Whether children (boys/girls) will attend primary and secondary school? o Whether and when to take children to the clinic? o Other important decisions as observed.

Women‟s status in community and collective activities: What is the status of women in their communities? How often do they go out of the house and for what? What are people‘s attitudes toward the status of women in general? What is the impact of the program on this status?

o Do they go out more now? For what? o What new institutions do women participate in that they did not before the program? (e.g. Registration of marriage and births, bank accounts, credit cards)

How have these changes affected their self-image/confidence? What forms of collective activities exist in the locality? What forms for women in particular?

o How does the program interact with these activities? o When did these activities start/post-CCT? Have there been any changes in these programs? o Are women more or less likely to participate since CCT? o Are there any differences between beneficiaries and non-beneficiaries?

Relation between spouses: How have relationships changed between men and women as a result of the program?

o Are women given more respect? o Are women more assertive o Is there new conflict or domestic violence (to get control of benefits or because of change in women‘s roles)?

Girls education: What are the reasons that children/girls are not sent to school/drop out of school? E.g.

o No use for education

111

o No employment opportunities o To get married o Fear for their physical safety o Concerns about boys o Concerns about rumors/reputation o Too expensive o Influence of external actors (landlords, local leaders) o Children don‘t like school o Other reasons….

What would it take to change each of these conditions? Are their program incentives that would address these issues? Do people understand the reasons for the difference in size of benefit for girls vs. boys, and for primary vs. secondary school?

Status of girls and boys: What is the status of girls vs. boys in the household, especially girls? How are they treated, viewed, by male and female HH members? How do people perceive this to have changed over time? How has the program affected this status?

Visions of the future: For mothers and fathers, what do they want for their daughters‘ futures?

o How do they perceive the role of education in affecting the future? o How do they see their future actually turning out?

When they were younger, what occupations did they want for themselves? For mothers and fathers, at what age did they get married themselves?

o Do they think was a good age? o At what age would they like their daughters to get married? Why?

What are their expectations of qualifications of prospective mates for their children and has the program changed this?

For older girls: Has the program affected how they see their potential, compared to their mothers? What do girls want to do with their futures?

o How much education do they need to fulfill their expectations? o How do they see their futures turning out? o How do they view marriage? At what age do they want to get married? Do they see reasons to get married later or earlier? o Have they changed their expectations for the qualifications of their respective mates?

112

For older boys: What do they think about educated girls? Would they prefer to marry a more educated or less educated girl? For brothers, what do they think about their sister going to school? What do they think about their future daughters going to school?

113

Appendix C2: Interview Guide/Checklist for Households and Key Informants: Second Qualitative and Anthropological Study

KEY

A=all M=mothers F= fathers B=boys (13+) G=girls (13+, unmarried) Y=young children (7-12) O=other HH member ORP=other relevant person H=health staff S=school staff P=foundation staff L=leaders N=non-bens OBS=observation

Issue Informant to ask Check 1. General information Household demographics (HH members, age, M, F, O, N, OBS whether in school) Formal and informal work M, F, O, N, OBS Income: cash and in-kind M, F, O, N, OBS Assets and living conditions M, F, O, N, OBS Recent or relevant shocks (e.g. illnesses, death, fire, M, F, O, N drought, debt problems, job loss, etc.

2. Program communications Reasons why people don‘t know their benefits have M, P arrived at Bank. How they are supposed to be informed and how they M are informed in practice; reasons they go to the Bank. More effective systems of communication. M, O, P

Role of different actors in communications. OBS, M, P

3. Beneficiary Selection and Targeting

People‘s understanding of program selection criteria. M, F, B, G, O, N, P, (e.g. ties to the government through relatives or H, S friends? Poverty? Children out of school) Opinion about this—fair? M, F, B, G, O, N, P, H, S Actual basis for selection (politics, tribe etc.) OBS, P What people think about errors of exclusion. M, F, N, P, H, S, L

114

What people think about errors of inclusion. M, F, N, P, H, S. L Ideas about how to improve selection and targeting. M, F, N, P, L, H, S Reasons why people do not apply (have not heard, N, H, S, L don‘t like idea of program, stigma, discouraged by someone, lost hope, ineligible etc.) Feelings of people who do not receive benefits. N Non-beneficiaries‘ awareness of who beneficiaries N are. Tensions between people included and those M, F, B, G, Y, O, N, excluded. OBS, H, S Form which this tension takes. M, F, B, G, Y, O, N, OBS, H, S Reasons for the tension (e.g. beliefs about political M, F, B, G, Y, O, N, influence? Other unfairness; unhappiness over not OBS, H, S receiving money) Way in which appeals process works in practice M, OBS, P, ORP (persons involved, mechanisms) Responsiveness of appeals process. M, OBS, P, ORP

4. Attitudes and perceptions with respect to government assistance, social programs and work Attitudes of adults and young people about finding a M, F, B, G, N job—their chances, requirements. Effects of program on willingness to seek work. M, F, B, G, OBS Effects of program on expectations about kind of M, F, B, G work they want. Effects of program on parents‘ sending their children M, F, B, G to work (more? less? no effect?) Opinion about the government giving people money M, F, B, G, O, L without a work requirement (stigma? feelings about this for self?) Opinion about the government giving people money M, F, B, G, O, L, H, conditioned on health and education services. S

5. Education Parents‘ level of education and reasons why they M, F, N stopped or continued to this level. Relationship between parents‘ level of M, F education/literacy and their response to the program. Other activities that children participation in to M, F, G, B, Y, N educate them, keep them busy (paid a fee? replacing their schooling?) Attitudes towards importance of education (purpose? M, F, G, B, O, N future benefits?) Parents‘ feelings about their own levels of education M, F, N and how this has affected their lives. Whether program enables school attendance or M, F parents would send their children without program. Effects of program on parents‘ attitudes towards M, F, B, G educating their children, especially girls.

115

Reasons for sending children to school (because M, F, N important in itself? to receive program benefits?) Length of time parents planning to keep kids in M, F, N school and reasons for continuing or ending. Other education or school rehabilitation programs in OBS, M, F, B, G, N, the area (contribute to changing attitude towards S importance of education?) Children‘s attitudes towards the educational level of B, G, OBS, ORP other family members. Teachers‘ treatment of children (different for B, G, N bens/nonbens? girls/boys?) Children‘s understanding of their beneficiary status. B, G, Y Feelings about this (stigma? pride? indifference?) Changes in children‘s grades at school. M, F, B, G, S, O, OBS Risks at school or in getting to school. M, F, B, G, O, S, OBS Regular marking (by teachers) of children as M, B, G, Y, S ‗present‘ even when they are absent. Reasons.

6. Health and Nutrition Health Patterns of health care practices, at home, with M, F, B, G, O, N, H, traditional healers, or public clinics (expenditures, ORP, OBS treatment of illnesses, growth monitoring, vitamins, vaccines, etc.)? Local beliefs about health care, nutrition, fertility. M, F, B, G, O, N, H, Relationship between these beliefs and different ORP, OBS government services. Knowledge/beliefs about aspects of health care M, H required by CCT. Attitudes toward these. Participation of bens and nonbens in required health M, F, N, H services. Reasons for non-participation in health services. M, F, N, H Incentives which would encourage greater M, F, N, H participation in health services. Attitude of health personnel to people attending M, H, OBS health services (respectful? rude? neutral?) Quality of health services obtained. M Differences in treatment of girls and boys. M, B, G, OBS

Nutrition Diets of adults and children. Changes caused by OBS, M, F, B, G, O CCT. Attitudes towards importance of good nutrition. OBS, M, F, N Effects of program on attitudes towards importance OBS, M, F of good nutrition for mothers and young children. Men‘s attitudes towards health care and good OBS, F, M, B, G, O, nutrition. Effects of these attitudes on women and N children.

116

Possible ways of changing men‘s attitudes towards M, N, H health care and good nutrition.

7. Pregnancy Attitudes towards family size, number of children M, F, O, N desired. Effects of program on these attitudes. Accessibility and use of birth control and abortion. M, H Incentive effect of program on women getting M, F, O, H pregnant. Perceptions of people about this. Direct knowledge of someone getting pregnant because of program. Reasons (benefits now? future support of children by government?) Other reasons for having more children. M, F, O, H, OBS

8. Use of grant What money is spent on. M, F, B, G, Y, O, N, ORP, OBS Reasons for these choices. M, F, B, G, Y, O, N Obligation to spend money in a certain way (have to M, F, P show receipts from purchases etc.)

9. Other social programs Other similar programs (health, education and M,F,H,S,N nutrition, livelihoods) that people belong to/benefit from How they compare them to CCT program M,F,H,S Other types of programs they would prefer M,F

10. Women’s Status and gender relations Status in household Status of women, men, girls and boys in their OBS, M, F, O, N households. How much power they have. How people perceive this status to have changed M, F, G, B, N over time. Impact of the program on this status. M, F, G, B. OBS Program-related reasons she goes out of the house M, F Who in HH makes decision about: - spending money earned by men M, F, N, OBS -spending money earned by women M, F, N, OBS -spending money earned by children M, F, N, OBS -spending the money from the CCT benefits M, F, OBS -whether children (girls/boys) attend primary M, F, N school/secondary school -whether and when to take children for medical M, F, N, OBS care -other important decisions as M, F, N, OBS observed/mentioned

Women‟s status in community and collective Activities:

117

What is the status of women in their communities? OBS, M, O, H, N How often they go out of the house and for what OBS, M, N Impact of the program on this status OBS, M, H Whether they go out more now, and for what. OBS, M, F New institutions that women participate in that they M did not before the program(e.g. Registration of marriage and births, bank accounts, credit cards) How these changes have affected their self- M image/confidence Forms of collective activities that exist in the M, F, O, H, S, P, N locality/ forms for women in particular. How the program interacts with these activities M, H, S, P When these activities started/post-CCT? Any M, H, S, P, N changes in these programs related to CCT. Whether women are more or less likely to participate M, H, S, P, N in these activities Differences in participation between beneficiaries M, H, S, P, N and non-beneficiaries?

Relation between spouses Nature of relationship between spouses OBS, M, F, O, H, N How relationships have changed between men and M, F, O, H, G women as a result of the program Whether women are given more respect M, F, O, H, G/B? Whether women are more assertive M, F, O, H, G/B? Whether there is new conflict or domestic violence M, F, O, H, G/B? (to get control of benefits or because of change in women‘s roles) Whether relations have improved (e.g. because of M, F, O new resources) Girls education Reasons that girls/boys are not sent to school/drop M, F, O, G, B, H, S, out of school N -No use for education -No employment opportunities -To get married -Fear for their physical safety -Concerns about boys -About rumors/reputation -Too expensive -Influence of external actors (landlords, local leaders) -Children don‘t like school -Other reasons…. Why the education benefit is/is not enough to compensate for cost of sending child to school What is needed to change these conditions M, F, O, G, B, H, S, N, P Program incentives that could address these issues M, F, O, G, B, H, S, P

118

How people understand the reasons for the M, F, G, B difference in amount of cash transfer for girls vs. boys, and for primary vs. secondary school

Status of girls and boys Status of girls vs. boys in the household How they M, F, O, G, B, H, S, are treated, viewed, by male and female HH N, P members. How do people perceive this to have changed over M, F, O, N time How the program has affected this status M, F, O, G, B, H, S

Visions of the future What mothers and fathers want for the futures of M, F, N their sons vs. daughters How they perceive the role of education in affecting M, F, N the future How do they see their future actually turning out? M, F, N When they were younger, occupations they wanted M, F, O, N for themselves Age at which mothers, fathers, O wanted to get M, F, O, N married themselves, age at which married and what they think of this decision now Age at which they would like their M, F, O, N daughters/granddaughters to get married Current expectations of qualifications of prospective M, F, N mates for their children/For Beneficiary: how the program has changed this

For girls: What girls want to do with their futures G, Y, N Level of education they feel they need to fulfill their G, Y, N expectations How they see their futures actually turning out G, N How they view marriage. Age they want to get G, N married. Reasons for getting married later or earlier Current expectations for the qualifications of their G, N respective mates and how program has changed this. How they see their potential, compared to their G, N mothers/For Beneficiary, how program has affected this

For boys: What boys want to do with their futures B, Y, N Level of education they feel they need to fulfill their B, Y, N expectations How they see their potential, compared to their B, N fathers/For Beneficiary, how program has affected this How they see their futures actually turning out B, N

119

What they think about educated girls. Why they B, N prefer to marry a more educated or less educated girl. For brothers: What they think about their sister B, N going to school through secondary level What they think about their future daughters going B to school through secondary level

120

Appendix C3: Interview Guide/Checklist for Phase 2 Case Study Focus and New SSI Households: Second Qualitative and Anthropological Study

NEW PHASE 2 PRIORITY CHECKLIST HH NAME: ______

NEW=Semi-structured interview Issue Informant Check PRIORITY ISSUE 1: School enrollment 1)-Why are parents not sending their children to NEW –E secondary school hhs (B&NB) secondary school? NEW –E primary school hhs (B&NB) -Why not, even when they receive the benefit (for School staff Bs)? Health staff -Why not, in general (for NBs) Muhtars District Governors Foundation staff 2. Are non-beneficiary households changing their -Your current NBs with enrolled behavior in the hope that this will help them to get the secondary school children benefits/? NEW + E NBs whose children are -Are they sending their children to secondary school? enrolled in high school Primary school? School staff -Are they getting vaccinated or going to health Health staff controls? Muhtars District Governors -What other explanations are there for rise in Foundation staff enrollment of non-beneficiaries? (other programs? T.v., What else?)

3). What better incentives would be needed to All –E HHs: encourage secondary school enrollment among HHs with secondary school dropouts households are resistent? HHs with primary students whose parents say they won‘t continue to secondary What better incentives would be needed to encourage HHs with non-enrolled primary students primary school enrollment among households still -Your NEW HHs with non-enrolled resistent? secondary or primary students B and NB School staff Health staff Muhtars District Governors Foundation staff -How much would the minimum benefit have to be monthly to decided to send your son/daughter to primary/secondary? -If they made schools safer and more proper for girls? What suggestions do they have?

-If they made other improvements at schools? Or with transportation (free, safer route, trustworthy drivers)? Free books, etc.? School closer to home? (Note: This is not about money; it is about changes in schools. -What else would they need? -In response to their specific reasons for not sending children to school, ‗Is there anything at all that would change your mind?‘ -Where children dont want to go to school, B and NB –E HH what might work as incentives to make NEW B and NB –E children more interested in school? Also ask the non-enrolled children who dont want to go to school -Would they send them if they made 12 Only in key informant interviews with

121

years compulsory? Should this be with or school directors and teachers. without CCt?

4) -Where children are not performing well, what are All old and new HHs with current or the reasons? former primary students who have performed poorly Do children understand Turkish very well? Is one reason they have trouble performing well in school because they don‘t understand Turkish well enough to school subjects? -Why else?

PRIORITY ISSUE 2: Pregnancy Incentive -Do they know what the CCT is? All old HHs

-If it is ‗child money‘ what is it for and what are the conditions?

(Find out if they are aware of the health and pregnancy benefit?) How much do they think it pays? All old HHs Health care staff Teachers Muhtars Neighbors/others Would they get pregnant in order to get the CCT? All old HHs Do you know anyone who became pregnant to All old HHs receive this benefit? What is the evidence for this? Health care staff: (midwives, mobile clinics, nurse/doctor, local, district, provincial) Teachers Muhtars Neighbors District governor Others

PRIORITY ISSUE #3: Work Disincentives Does receiving the benefit encourage men to: -Foundation staff -stop working? -Muhtars -not look as hard for work? -Neighbors -Coffee house owners -District governor

PRIORITY ISSUE 4: How could a system of Key informant interviews with: community-level beneficiary liaison work best (a ―promoter‖?) [A promoter is a local woman, a -Foundation staff liaison between the Foundation and the beneficiaries, -Health center staff and mobile clinic selected to follow up with beneficiaries with providers information on: program conditionalities, amount of -Teachers benefits for different age and sex children; notify that -Muhtars/wives money is in the bank, answer other questions] -Others with insight -District governor -Beneficiaries and other community persons How should a promoter be selected? KIs Bens

122

What type of person would be most appropriate? KIs (have the necessary skills, respect, non-controversial, Bens etc.) Should the promoter be paid? KI What method for conveying information? KIs -Could beneficiaries be gathered in a meeting -Or, could the promoter only assist individuals? -Where could they gather? -Would they come? What would be there difficulties?

-Would you go once a month to someone‘s house or Bens community building to get program information? What problems might you face with this? What type of program information should be KIs conveyed at meetings? What issues discussed? Bens Could a nurse or midwife come to meetings with KIs information about health and nutrition (e.g. food for young children, reasons for health controls, other health and hygiene issues; food to buy with program money, etc? What other things would need to be considered in KIs order to have this new system?

PRIORITY ISSUE 5: Targeting Are there very poor people NOT applying for the - New poor NB households who have not benefit? applied -Do they know about the program? -All old HHs -If they know about the program, why don‘t they Interviews with: apply? Teachers Health staff Foundation Muhtars Other [Ask above key informants for help identifying poor non-applicants] How did the targeting, (i.e. beneficiary application Foundation staff and processing) work in practice? Muhtars Teachers Health staff District governor What were the unanticipated problems? KI

Was it done fairly? Is there any manipulation at local KI level of amount of money given to people, or this just a mistaken local perception? How frequently was verification done at household Foundation staff level?

Who is filling out the application forms? Who writes? KI Who participates in determining what to write? Bens What would be a better system for selection? KI

PRIORITY ISSUE 6: Health issues Why are people more aware of Ed benefit than Health Foundation staff Benefit? Health staff School staff

123

What do people know about health controls? All old HHs -Where they know of controls, what is the purpose For households where children are not vaccinated: All old HHs (without vaccines) -Why not? (e.g. Distance, afraid, not seen as important, etc. )

PRIORITY ISSUE 7: What do people think about conditionality?

Is it better to have school and health requirements as All old HHs conditions of the benefit or to give the benefit without Foundation staff conditions? Muhtars Health staff School staff District governor -What other new conditions would be helpful? ― What other behavior change could be encouraged

124