Population and Development Review, Volume 24, Number 4

POPULATION AND DEVELOPMENT REVIEW

John C. Caldwell Malthus and the less

VOLUME 24 NUMBER 4 developed world: The pivotal role of India D E C E M B E R 1 9 9 8 Jere R. Behrman and James C. Knowles Population and reproductive health: An economic framework for policy evaluation Ashley S. Timmer and Jeffrey G. Williamson Immigration policy prior to the 1930s: Labor markets, policy interactions, and globalization backlash Zeba A. Sathar and John B. Casterline The onset of fertility transition in Pakistan Notes and Commentary Population, carbon emissions, and global warming: An exchange Data and Perspectives M.E. Enchautegui on low-skilled immigrants and the changing American labor market Archives Francis Hutcheson on the rights of society Book Reviews Review essays by I. Castles and M. Perlman; reviews by J.R. Wilmoth, T. Dyson, G. Santow, R.D. Retherford, and others Documents Population aging and the US federal budget; the 1998 revision of the UN population projections Population and Development Review seeks to advance knowledge of the interrelationships between population and socioeconomic development and provides a forum for discussion of related issues of public policy.

EDITOR Paul Demeny

MANAGING EDITOR Ethel P. Churchill

EDITORIAL COMMITTEE Paul Demeny, Chair Geoffrey McNicoll Ethel P. Churchill Michael P. Todaro Susan Greenhalgh

EDITORIAL STAFF Robert Heidel, Production Editor Y. Christina Tse, Production/Design Margaret A. Knoll, Circulation Sura Rosenthal / Susan Rowe, Production

ADVISORY BOARD Ester Boserup Akin L. Mabogunje Gustavo Cabrera Milos˘ Macura John C. Caldwell Carmen A. Miró Mercedes B. Concepción Asok Mitra Richard A. Easterlin Samuel H. Preston

Signed articles are the responsibility of the authors. Views expressed in the Review do not necessarily reflect the views of the Population Council. Direct manuscripts, comments on articles, and correspondence to: Population and Development Review Population Council One Dag Hammarskjold Plaza New York, New York 10017 USA Subscription information appears on the inside back cover. Support from the United Nations Population Fund is gratefully acknowledged. Volumes are available on microfilm from University Microfilms, Inc., 300 North Zeeb Road, Ann Arbor, Michigan 48106. The website for Population and Development Review is www. popcouncil.org/pdr The full contents of Volumes 1–19 (1975–93) are available through participating libraries from JSTOR at www.jstor.org/journals/00987921.html Population and Development Review (ISSN 0098-7921) is published quarterly in March, June, September, and December by the Population Council, One Dag Hammarskjold Plaza, New York, NY 10017 USA. Periodicals postage paid at New York, NY, and additional mailing offices. POSTMASTER: Send address changes to Population and Development Review, One Dag Hammarskjold Plaza, New York, NY 10017. © 1998 by The Population Council, Inc. All rights reserved. Printed in the United States of America. Library of Congress Catalog Card Number: 75-27311 ISSN 0098-7921 Information for Subscribers Population and Development Review is available on a paid subscription basis at the following rates: One year (4 issues) US$32.00 Two years (8 issues) US$60.00 To enter a subscription, send payment by check or money order drawn on a US bank, payable to the Population Council, or by Visa or MasterCard (provide card number and expiration date), together with mailing address to: Population and Development Review Population Council One Dag Hammarskjold Plaza New York, New York 10017 USA Credit card orders may also be placed by phone to (212) 339-0514, by fax to (212) 755-6052, or by e-mail to [email protected]

Complimentary subscriptions Complimentary subscriptions are available to qualified applicants residing in developing countries. Individuals and institutions wishing to apply for complimentary subscriptions should send requests on letterhead stationery to the above address stating the nature of professional involvement in development- and population-related issues.

Back issues Casebound (hardcover) volumes are available at $45.00 each. Single back issues are available at $8.00 per issue. Please designate copies desired by volume and issue number. Orders may be placed by mail, phone, fax, or e-mail as indicated above. Complimentary copies are available to qualified individuals residing in developing countries. A twenty-year cumulative index to Volumes 1–20 (1975–94) is available at no charge to subscribers to the Review.

Printed on recycled paper in the USA. POPULATION AND DEVELOPMENT REVIEW VOLUME 24 NUMBER 4 D E C E M B E R 1998

ARTICLES

Malthus and the Less Developed World: The Pivotal Role of India 675

JOHN C. CALDWELL

Population and Reproductive Health: An Economic Framework for Policy Evaluation 697

JERE R. BEHRMAN AND JAMES C. KNOWLES

Immigration Policy Prior to the 1930s: Labor Markets, Policy Interactions, and Globalization Backlash 739

ASHLEY S. TIMMER AND JEFFREY G. WILLIAMSON

The Onset of Fertility Transition in Pakistan 773

ZEBA A. SATHAR AND JOHN B. CASTERLINE

NOTES AND COMMENTARY

Population, Carbon Emissions, and Global Warming: Comment 797

PAUL DE SA

Toward a Per Capita–Based Climate Treaty: Reply 804

FREDERICK A. B. MEYERSON DATA AND PERSPECTIVES

Low-Skilled Immigrants and the Changing American Labor Market 811

MARÍA E. ENCHAUTEGUI

ARCHIVES

Francis Hutcheson on the Rights of Society 825

BOOK REVIEWS

The Mismeasure of Nations: A Review Essay on the Human Development Report 1998

IAN CASTLES 831

Styles of Population Economics: A Review Essay on Mark R. Rosenzweig and Oded Stark (eds.), Handbook of Population and Family Economics

MARK PERLMAN 846

Henri Leridon and Laurent Toulemon, Démographie: Approche statistique et dynamique des populations and Jean Bourgeois-Pichat, La dynamique des populations: Populations stables, semi-stables et quasi-stables

JOHN R. WILMOTH 860

Gavin W. Jones, Robert M. Douglas, John C. Caldwell, and Rennie M. D’Souza (eds.), The Continuing Demographic Transition

TIM DYSON 863

John M. Riddle, Eve’s Herbs: A History of Contraception and Abortion in the West

GIGI SANTOW 869

Ochiai Emiko, The Japanese Family System in Transition: A Sociological Analysis of Family Change in Postwar Japan

ROBERT D. RETHERFORD 875

Short Reviews 878 DOCUMENTS

Population Aging and the US Federal Budget 885

The 1998 Revision of the United Nations Population Projections 891

ABSTRACTS 897

AUTHORS FOR THIS ISSUE 902

CONTENTS TO VOLUME 24 903

ACKNOWLEDGMENTS 909

Malthus and the Less Developed World: The Pivotal Role of India

JOHN C. CALDWELL

MALTHUS’S IDEAS have survived the two centuries since his First Essay was anonymously published, and they remain an intellectual force to be reck- oned with.1 The purpose of this essay is to identify why this has been the case and to explore the unique role of the English-speaking world and of the connection between Britain and India in determining the survival of the principle of population. First it is necessary to examine the kind of society that produced the theory and those of its elements that survived time and transplanting. Thomas Robert Malthus (1766–1834) was a man of his time but also a man who helped shape his time. John Maynard Keynes (1951: 101) placed him with John Locke, David Hume, Adam Smith, and William Paley as his predecessors, Jeremy Bentham as his contemporary, and John Stuart Mill and Charles Darwin as his successors “profoundly in the English tradition of humane science, . . . a tradition marked by a love of truth and a most noble lucidity, by a prosaic sanity free from sentiment or metaphysic, and by an immense disinterestedness and public spirit.” It was essentially a secular tradition, in spite of the fact that Paley and Malthus were Anglican cler- gymen. It was also one in tune with the outlook and interests of the rising commercial middle class. Malthus himself, like most of his contemporaries seeking general social laws, would have added Isaac Newton to this list. Malthus was also in tune with his predecessors in another way that he readily conceded. His argument that population was limited by the availability of food was commonplace in the eighteenth century (Winch 1992: xi), and so he has been attacked as a plagiarist as if this were his sole contribution (e.g. Fryer 1965: 70). In fact, his views on the relationship between food and sustenance were expressed much more strikingly than those of other commentators, as well as in a far more significant context. Malthus lived in England at a time when the older mercantilist economic and demographic policies of protecting domestic production and aug-

POPULATION AND DEVELOPMENT REVIEW 24(4):675–696 (DECEMBER 1998) 675

Click to print article Click to return to Table of Contents 676 MALTHUS AND THE LESS DEVELOPED WORLD menting population growth were being successfully attacked. One reason was the growth of Britain as a trading power. Another was the related growth in power of the middle class, to be given political form by the Great Reform Act of 1832. Mercantilism was essentially the program of auto- cratic, centralized states, and lingered longer on the continent. Adam Smith (1723–90) had enunciated a new economics, suited to emerging capitalist Britain, in 1776 in his Wealth of Nations. But his work lacked the major population component that was subsequently produced by Malthus. The two doctrines complemented each other and provided a “liberal” political agenda. Malthus wrote the First Essay during a decade when population probably grew at twice the rate that had been the case during the decade when the Wealth of Nations was written (12 percent decadal growth compared with 6 percent, according to G. Talbot Griffith’s estimates; see Mitchell and Deane 1962: 5). England had a Poor Law, codified in Elizabethan times, whereby magistrates spent parish rates meeting the needs of destitutes. Uniquely, this allowed a monetary estimate to be put on the cost of providing for the very poor and the additional cost imposed by their children. But it also became a greater political issue as the eighteenth century drew to a close because of inflation associated with the French Wars. Malthus also wrote at a time when parents in at least some Western countries were probably beginning to feel that large families represented a greater burden than before, and when some deliberately used contraception to reduce ultimate family size. Abbé J. A. Dubois, who left France in 1792 to spend over 30 years in South India, described, in contrast to India, pre-Revolutionary France: “Other nations [those in Western Europe] . . . regard the fruitfulness of their women with aversion. They are, moreover, not ashamed to resort to wicked and disgusting means of reducing or de- stroying it altogether . . . that they might not deprive themselves of the means of satisfying their ambition or of procuring the luxuries of life . . . ” (Dubois 1906: 592–593). It is unlikely that the British of the 1790s were any less apprehensive than the French of fruitfulness, but they, like Malthus, were ashamed of resorting to birth control and, in contrast to the French, it was to be another three-quarters of a century before their fertility fell. The importance of Malthus was that he perceived a self-regulating system explaining population fluctuations. This was a system that Charles Darwin was to generalize through all species. Human fecundity was so great that fertility could easily outstrip the growth in food production. This would lead to starvation and a rise in mortality, bringing population growth back to the level of the food supply. Such crises were usually averted because, in “civilized” regions like Europe, a preventive check came into operation: marriage was either postponed or averted, thus reducing women’s level of childbearing. There were other positive checks, especially outside Chris- tian Europe: warfare, epidemic disease, the frequenting of prostitutes lead-

Click to return to Table of Contents J OHN C. CALDWELL 677 ing to sterilizing disease, polygamy, and such vices as infanticide, abortion, and contraception. If one check declined, others would inevitably become more forceful. The model was more sophisticated than this simple description suggests, and it was the additional complexities that carried a political message. Although Malthus believed that the total societal level of mortality had not changed for centuries, he held that, in a class society, the lowest classes, nearer the breadline, were characterized by the highest mortality and were likely to die first in a subsistence crisis. Furthermore, the most ignorant were the most likely to spurn the preventive check and were likely to marry as soon as they had the possibility of providing the ensuing children with the barest sustenance. This meant that financial transfers under the Poor Laws allowed them access to more food and encouraged them to marry earlier and breed more rapidly. The additional children put greater strains on the food supply and inflated its price, thus restoring population equilibrium by making it harder for other sections of society, especially the social class next above them, to buy food, and so raising their mortality. The intended benevolent functioning of the Poor Laws thus did nothing to reduce mortality or misery but rather spread these conditions to worthier parts of society. To force such profligate couples to take responsibility, their abandoned illegitimate children should be allowed to die (Malthus 1992: 263). This interpretation ultimately led to the Poor Law Amendment Act of 1834 with its establishment of urban workhouses to replace rural out- door relief. Curiously, it was to be the first of Malthus’s theses to have an impact on the administration of India. Malthus considered the only satisfactory escape from the population– subsistence trap to be the restraint of marriage accompanied by sexual con- tinence before and outside marriage. He began by regarding such restraint as typifying the behavior of civilized societies, primarily Christian Europe, especially northern Europe. Increasingly he regarded it as a criterion of civilization and even as a major element in the civilizing process. He did, however, believe that passion between the sexes, marriage, and childbearing were of great importance, and that it was unfortunate that natural law meant they had to be constrained. Nevertheless, such conflicts were part of a greater equilibrium, or, better, near-equilibrium model, that had both beauty and a deeper meaning. Adam Smith had promoted “the thesis that man, in following the prompt- ing of his nature, unconsciously gives substantial expression to some parts of the . . . Plan” (Campbell and Skinner 1976: 4). This was certainly natural law, but Smith never makes it clear whether it was also Divine Law. There is no such ambiguity in the case of Malthus. He, like many of his Cambridge generation, was greatly influenced by William Paley, the theo- logian and philosopher (1743–1805, author of Principles of Moral and Politi- cal Philosophy, 1785), and both believed in God’s design of a world tending

Click to return to Table of Contents 678 MALTHUS AND THE LESS DEVELOPED WORLD toward equilibrium (see Winch 1992: xix–xx). The historian Henry Buckle (author of History of Civilization in England, 1857) said that Adam Smith’s model would still remain valid without the statistics and illustrations (Campbell and Skinner 1976: 56), and Malthus clearly regarded the principle of population in the same light. Malthus’s belief that the principle of population was part of the divine plan is the key to our understanding of the aspect of his work that later generations found so incomprehensible: his opposition to contraception. He certainly regarded contraception as vile, insulting to womanhood, and un-Christian—the consensus view in his day, shared by all but a few peripheral dissidents. The preventive check needed and taught self-control. It was a civilizing instrument. It was necessary to overcome the indo- lence of the poor and the uncivilized. God’s plans to elevate his people could be thwarted by devices that kept starvation at bay while leaving in- dolence to reign. The long road to civilization might not be begun (see Breton 1983: 249–250). There was a strong case against “immediate grati- fication,” marriage could be looked upon as a prize for industry and virtue, chastity had a solid moral foundation, and Christians should avoid temp- tation (Malthus 1992: 218–221). Malthus’s refusal to link his identifica- tion of the threat of population growth with an encouragement of contraception greatly increased the possibility that his views would win widespread public acceptance.

The acceptance of Malthus’s doctrines

The impact of Malthus on the world depended at first on how he was accepted by his British contemporaries. A generation after the publication of the First Essay, the principle of population was little debated. The Irish famine of 1846–47 was rarely treated as a test of Malthus’s ideas. His first biog- rapher, Bonar (1924: 363), looking back from 1885, believed that this was because most influential people, and indeed a broad swathe of society, were almost immediately converted to the central Malthusian thesis—a view subsequently taken also by Keynes (1951: 100–101), Glass (1953: 30), and Hollingsworth (1983: 213–215). England was a cost-accounting “nation of shopkeepers,” predominantly Protestant and increasingly scientific and secular in much of its thought, growing crowded, with most families fearing too many births. Bonar reported that Malthus soon won over leading Whig politicians (William Pitt, the younger, Henry Brougham, James Mackin- tosh, and Samuel Whitbread), theologians (William Paley and Edward Copleston), economists (David Ricardo and Nassau Senior), and historians (Henry Hallam). In opposition were the poets, writers, and romantics (Rob- ert Southey, William Hazlitt, and William Cobbett). Many supporters of the working class, however, remained hostile specifically because of the

Click to return to Table of Contents J OHN C. CALDWELL 679 argument that acceptance of the principle of population led to hostility toward poor relief, and generally because of the judgmental attitude toward the value of the various social classes. A survey of nineteenth-century En- glish diaries and novels showed that, throughout the century, “the British never saw very large families as particularly desirable. . . . [A]t least from the beginning of the nineteenth century, those who had substantially more children than was typical were seen as comic, inept, or sad victims of fate.... [T]he theme is one of reproduction outstripping resources” (Kane 1995: 152–153). Hollingsworth (1983: 217–218) believes that Malthus’s thesis that disaster occurred when population growth outstripped food supply was so taken for granted by the 1840s that few thought of the Irish famine as testing Malthus. But it is likely that the acceptance of Malthusian concepts explained the delayed reaction of the British government to the famine. Even after passage of the 1834 Poor Law Amendment Act, Malthus’s influential successors spelled out his message about the dangers of helping the unfortunate in the clearest terms. In 1848 John Stuart Mill concluded in his Principles of Political Economy: “Everyone has a right to live. We will suppose this granted. But no one has a right to bring creatures into life, to be supported by other people” (Mill 1848: 252). Two decades later Walter Bagehot, the editor of The Economist, was applying “the principles of natural selection to political society”:

The most melancholy of human reflections, perhaps, is that, on the whole, it is a question whether the benevolence of mankind does most good or harm. Great good, no doubt, philanthropy does, but then it also does great evil. It augments so much vice, it multiplies so much suffering, it brings to life such great populations to suffer and to be vicious. (Bagehot 1869: 188–189)

John Carey (1992) has shown that by the last decades of the nineteenth century and the first decades of the present century the mainstream view of the English literary intelligentsia was strongly Malthusian. Popula- tion growth, especially of the poor, was dangerous. Perhaps Malthus’s most effective support came from where he least sought it, those who advocated artificial birth control. By 1822, with Malthus still alive and producing new editions, Francis Place in England had drawn the conclusion in his Illustrations and Proofs of the Principle of Popu- lation (1994) that the solution to the Malthusian difficulty was contraception, and he was probably the author of the “Diabolical Handbills,” which circulated the following year with clear instructions to working-class women on how to practice contraception. By 1832 Charles Knowlton in the United States had published a book on the subject. The birth control movement saw where Malthus’s views led, and it was not inappropriate that Charles Bradlaugh should found the Malthusian League in the early 1860s or that

Click to return to Table of Contents 680 MALTHUS AND THE LESS DEVELOPED WORLD first the Dutch and then the English should describe contraception as neo- Malthusianism from the late 1870s on.

Malthus on the “less civilized” parts of the world

In the First Essay Malthus’s treatment of the “less civilized” world, variously and unsatisfactorily now known as the third world or the less developed countries, was limited. Indeed, like most educated Englishmen of his time, he knew more about the Classical World. He drew on literature on both to draw contrasts with contemporary, Christian Europe. The latter was “civilized,” not least, as he noted in the fourth edition, because “in modern Europe the positive checks to population growth prevail less, and the preventive checks more than in past times, and in the more uncivilized parts of the world” (quoted in Wrigley 1983: 115). In the uncivilized lands there was more warfare, epidemic disease, and vice, as well as famine. In- deed the first three more often than not kept starvation at bay. He identified un-Christian practices like infanticide, and even polygyny, as taking the place in Asia of Europe’s restraint from marriage. Malthus was well aware that in sparsely settled lands with few resources population was just as likely to press on resources as in the most densely settled countries. Indeed, he devoted more attention to North American Indians, Australian Aborigines, and Polynesians than to the heart- land of agrarian Asia—partly, admittedly, because explorers and voyagers with a scientific bent were more frequently reporting on the former in the second half of the eighteenth century. But it was inevitable that later generations would most readily identify Malthusian crises as existing in densely populated agrarian societies, such as those of India and China, where famines and epidemics could kill millions. Malthus wrote little about India, in spite of the fact that—or perhaps because—he was Professor of Political Economy, teaching India’s future administrators, at the East India Company’s College at Hertford and then Haileybury from 1805 until his death in 1834. He referred more frequently to East Asia. In the First Essay he pointed out that China was entirely occu- pied by tillage so that even the marginal lands, producing less, were being employed. He noted: “It is said, that early marriages very generally prevail through all the ranks of the Chinese. Yet Dr Adam Smith supposes that population in China is stationary” (Malthus 1986a: 25). If this were the case, then the explanation was famine and the “exposure” of children, for there did not appear to be a high level of disease. With regard to infanticide, he noted: “it is difficult to avoid remarking, that there cannot be a stronger proof of the distresses that have been felt by mankind for want of food, than the existence of a custom that thus violates the most natural principle of the human heart” (Malthus 1986a: 25). By 1826, in the sixth edition, he identified with alarm “the extraordinary encouragements that

Click to return to Table of Contents J OHN C. CALDWELL 681 have been given [in China] to marriage” (Malthus 1986b: 128). He gath- ered evidence also of early and nearly universal marriage in India, as a result of which: “The population would thus be pressed hard against the limits of the means of subsistence, and the food of the country would be meted out to the major part of the people in the smallest shares that could support life. In such a state of things every failure in the crops from unfavourable seasons would be felt most severely; and India, as might be expected, has in all ages been subject to the most dreadful famines” (Malthus 1986b: 119–120). From the second edition onward, Malthus became increasingly interested in the lack of prudence about marriage in the uncivilized nations, and in the conditions that nevertheless stopped them from being in a per- petual state of famine. Apart from disease, he found vice: fighting, unchas- tity and sterilizing disease, polygyny, infibulation, infanticide (see Godelier 1983: 134), as well as prolonged lactation and sexual abstinence within marriage. He saw the possibility that one day “the diffusion of education and knowledge” might bring into force the preventive check, the delay of marriage. But those who administered the British colonies, especially India, drew three major lessons from Malthus. The first was that colonial administration reduced the level of the positive checks, certainly warfare and infanticide, and probably disease, with the result that populations would tend to grow faster, calling more frequently on famine as the ultimate check. The second was that interventions in cases of famines might, like poor relief in England, do more harm than good. The third was that the check of delayed marriage might not come into being for a very long time. In time, some were to argue that the Malthusian theses led directly to the need for birth control.

The Anglo–Indian link

The link between Britain and India, unique in colonial history, is of great importance for understanding the repercussions of Malthus and his doctrines (see Caldwell and Caldwell 1986: 37–43). The East India Company was chartered at the beginning of the seventeenth century and had established a presence in India within a decade. Two-and-a-half centuries later, in 1858, the British Government assumed direct control of the country, to relinquish it in 1947 to independent India and the newly created state of Pakistan. Even before 1858 the British Government had ultimate responsibility for India, and the affairs of the East India Company and the Gov- ernment were intermeshed. India was not an ordinary colony; its population, standing at the middle of the nineteenth century at seven times greater than Britain’s and constituting over 90 percent of the population of the colonial empire, ensured that. It had an ancient history recorded in litera-

Click to return to Table of Contents 682 MALTHUS AND THE LESS DEVELOPED WORLD ture and fascinated most Englishmen who had contact with it. And most of those who had been influential in the acceptance of the principle of population had such contact. Malthus worked nearly his whole adult life for the East India Company, as did John Stuart Mill for the India Office (where Keynes later worked briefly); Walter Bagehot’s father-in-law was the Secretary for the Treasury in India. The administration of India, partly based on Moghul procedures, influenced the organization of Britain’s own civil service. Indian society was hierarchical, and upper-class Englishmen understood India’s royal families and upper castes. They did not have to hurdle cultural and linguistic barriers in doing so, especially after the 1835 resolution within the framework of the Government of India Act with its educational provisions molded by Lord William Bentinck, the Governor-Gen- eral, and Lord Thomas Macauley, which resolved to spend the Education Fund solely on English-medium education to promote European literature and science. Within a century 220,000 Indian schools and colleges were teaching 13 million students (Cunningham 1941: 149–150). Seeley (1885: 253) described the decision as “the great landmark in the history of the British Empire considered as an institution of civilization.” English had been a medium of communication between the rulers and the ruled long before, but the 1835 resolution ensured that a huge English-speaking segment of the population would be conversant with British writings and thought, and would impinge on the evolution of Britain much as the Brit- ish would leave their mark on the Indian intelligentsia (see O’Malley 1941; Rawlinson 1941). No other densely settled underdeveloped country had such a relationship with a European nation: the Dutch did not develop it with Java and neither China nor Japan was colonized. The material circumstances of India had begun to make the British see it in a Malthusian way even before the publication of the First Essay: Sir Thomas Munro, in a 1795 letter to his sister, talked of famine and the check imposed by food production on population growth (Ambirajan 1976: 5). The British and Indians had common ground in Malthusianism. Both were densely settled. Once, densely settled lands had been regarded as evidenc- ing prosperity, as was the case when the fourteenth-century Arab traveler, Ibn-Batuta, had reacted to the then most closely settled part of India by calling it “Golden Bengal.” Malthus ensured that educated Englishmen and Indians would never again see the world this way. Generations of British administrators and generations of reports ensured that India was known to Western researchers in a way that no other agrarian civilization was. There was no equivalent elsewhere in the underdeveloped world to the demographic statistical information that flowed from the Indian census beginning in the late 1860s. In India the Malthusian model could be tested at work in a non-European country as nowhere else.

Click to return to Table of Contents J OHN C. CALDWELL 683

India before the censuses

The first impact of Malthus’s ideas on India’s administrators came from his theses about the Poor Law, as has been well demonstrated by the historian of this phenomenon, S. Ambirajan (1976). A starkly clear parallel existed between the payment of Poor Law relief from Britain’s upper classes to the destitute poor, and the payment of famine relief in India from revenues ultimately deriving from taxes contributed disproportionately by the better-off and upper castes to those at the bottom of the social ladder. Malthus’s views and language were well known from early in the century, and in 1829 the Madras Board of Revenues of the East India Company commented on regional data appearing to show food production increasing faster than population: This is “at variance with one of the best established truths of political economy” (quoted in Ambirajan 1976: 6). In the first two decades after Britain assumed direct responsibility for the administration of India, parts of the country were subject to devastating famine, particularly in Orissa in 1866 and Bengal in 1874. Ambirajan found that during these critical years many in the administration were converted to a Malthusian view of the situation. Lord Edward Lytton, Gover- nor-General during 1876–80, was convinced that famine should be solved by the market, telling the Legislative Council in 1877 that the Indian population “has a tendency to increase more rapidly than the food it raises from the soil” (quoted in Ambirajan 1976: 6). Many reports during the 1870s were written in Malthusian terms, arguing that famine was becoming commoner because the Pax Britannica had diminished many of the other direct checks, certainly warfare, infanticide, and suttee, and possibly epidemic disease. This view was expressed by the Report Consequent to the Partial Census Conducted in India after the 1877 Famine and by the 1880 Famine Commission. An 1881 report concluded that 80 percent of famine victims were drawn from the poorest 20 percent of the population, and if such deaths were prevented this stratum of the population would still be unable to adopt prudential restraint. Thus, if the government spent more of its revenue on famine relief, an even larger proportion of the population would become penurious. Ambirajan (1976: 7– 11) concluded that, although the administration was divided about the value of famine relief, these Malthusian views undoubtedly affected its volume and timing. Although the language employed originated in Malthus’s works, the authorities cited were more often his midcentury interpreters, especially Mill (1848) and Bagehot (1869). Sir Robert Giffen, statistician, Walter Bagehot’s assistant editor of The Economist, and later head of the British Board of Trade, said he was expressing the view of most leading public men and economists when he wrote in the late nineteenth century that “‘from the point of civilization and progress’ millions of money spent in

Click to return to Table of Contents 684 MALTHUS AND THE LESS DEVELOPED WORLD pauperizing Indians could not ‘be contemplated with any satisfaction’” (quoted in Ambirajan 1976: 9–10). British Empire censuses, like those of Britain itself, were planned to take place in the first year of each decade. In fact, India’s first census, that of the 1870s, spanned the years from 1868 in the Punjab to 1876 in Ben- gal. But from 1881 on, censuses were held simultaneously in all parts of the country. The series of censuses, now spanning 120 years to 1991, are a demographic record for the less developed world without parallel. There was nothing similar for China until 1984 when the 1953, 1964, and 1982 censuses were released simultaneously. The Indian census reinforced interest in Malthusian equilibria, and later played a central role in creating the world’s first national family planning program. The reports of the censuses of the 1870s and 1881 were published separately by state or territory, usually under the auspices of a British census-taker in directly ruled British India and an Indian census-taker in the Indian States. Their reports were almost identical, those of persons well educated in the British tradition and understanding the population question as it was debated by British nineteenth-century political economists. From 1891 there was also an overall Indian report written by a British official, often recruited from Britain to take the census. The census reports took as a major theme the race between population and India’s ability to produce food. The 1868 Punjab report (Indian Census 1870: 17) warned that “as the limit of cultivation is approached, the rate of increase of population will halt.” The report of Bengal (Indian Census 1878: 84) explored the central Malthusian issues: “British rule has established peace and security throughout the country, and so far has re- moved some of the causes which were at work to check the natural increase of the people. But Bengal is not more fertile than it used to be, and, if there has been an extension of cultivation, there are also large exports of agricultural products which were unknown at the beginning of the century.” In less densely settled British Burma, it was largely the relaxation of the checks that was of interest: “These tracts of country came into our pos- session in the last war [the Third Burmese War], and in 1872 they had for 15 years been enjoying the benefits of a secure government and abundant material prosperity. One inevitable result of such a change is a rapid natural increase among the indigenous population” (Indian Census 1875: 19). In progressive Cochin, A. Sankariah noted, “The subjects of the Rajah of Cochin have therefore no need to fear of their numbers outstripping the capacity of the earth to sustain them if they will only exhibit ordinary prudence and enterprise” (Indian Census 1877: 56). The outlook of the 1881 census reports was inevitably affected by the famines of the 1870s. The report for the Northwest Province and Oudh commented that the Benares Division “as a whole suffered less than the

Click to return to Table of Contents J OHN C. CALDWELL 685 rest of the province in 1878–79 and consequently the positive checks to population have been less active” (Indian Census 1882: 26). The Baroda Territories report discussed “the laws of increase” and “the law of growth” and commented: “As a rule, in crowded cities and towns the preventive and also the positive checks operate in a greater degree than they do in the rural districts. Diseases and epidemics make their appearance more frequently in crowded cities and towns than they do in the rural districts. The preventive causes, namely moral restraint and vice, or rather compulsory restraint and vice, have a greater operation in the cities and towns than they have in the districts” (Indian Census 1883a: 73). The 1881 Madras report went beyond marriage to comment on famine and fertility: “It may be accepted that when food is scarce there are fewer births; whether this is exclusively the result of prudence, and whether that prudence is deliberate or instinctive, it is not here necessary to enquire” (Indian Census 1883b: 25–26). The Mysore report noted that famine mortality was particularly high among the Indian poor because even less help is transferred between castes than between classes (Indian Census 1884: 40). In the all-India or General Report of the 1891 Census, J. A. Baines took a particular interest in “The Movement of Population” in Chapter 3. He believed that “population tends to equilibrium,” and noted: “The conclusions arrived at by Malthus . . . have been severely criticised, and often misquoted or misapprehended, but, in the main, they have not been dis- proved“ (Indian Census 1893: 58). This came from a man who had in his keeping the records of a census of over 200 million persons and two previous decennial censuses for comparison. When Baines wrote the report, Brit- ain had been experiencing a contraception-induced fertility decline for twenty years. He knew of the decline but in common with most educated Englishmen of his time not its causes, and followed Herbert Spencer in believing that fecundity was being reduced by “the increased demand upon the nervous and intellectual qualities of life” (Indian Census 1893: 58). However, he followed Malthus in believing that a long period of enlightenment is needed “before the preventive checks can come into being.” He noted that in India at that time, female marriage was early, but the cause of “the enormous number of births” was its universality (Indian Census 1893: 59–60). The 1911 Census Report tried to relate the cultivable area in different parts of India to the rate of population growth in order to explore Malthusian pressures (Caldwell and Caldwell 1986: 37). The 1921 Superintendent of the Census, in an era when the impact of deliberate birth control in the West had become recognized, wrote in his report that India “has abandoned—or more or less abandoned—the old- fashioned methods of limiting population to an optimum, viz. periodic ab- stention from intercourse, abortion and infanticide and she has not yet adopted the methods of advanced countries, viz. postponement of mar-

Click to return to Table of Contents 686 MALTHUS AND THE LESS DEVELOPED WORLD riage and voluntary birth control. She is at a point where the population is controlled by disease and disease only” (Indian Census 1924: 48–49). By the 1930s, J. H. Hutton, the Superintendent of the 1931 census and the author of a major study of the Indian caste system, wrote in his Report:

It appears to be the general opinion of Indian economists who discuss the population problem of this country that the only practical method of limiting the population is by the introduction of artificial methods of birth control, though it is not easy to exaggerate the difficulties of introducing such methods in a country where the vast majority of the population regard the propagation of male offspring as a religious duty and the reproach of barren- ness as a terrible punishment for crimes committed in a former incarnation.... Nevertheless, a definite movement towards artificial birth control appears to be taking place and is perhaps less hampered by misplaced prudery than in some countries which claim to be more civilized. (Indian Census 1933: 31–32)

He also echoed the later editions of Malthus in noting that improvements in child survival and education would assist fertility restraint. The next full census was in 1951, in an independent but partitioned India, with an Indian head of the census, R. A. Gopalaswami. His attitudes were similar to those of his English predecessors, but sharpened by the fact that he was a citizen of this newly independent country, by his experience as Secretary of the Bengal Famine Inquiry, and by “the incredible population increase of the past 30 years.” He spoke of “improvident maternity” and quantified the race between population and food (Huxley 1960: 72). He devoted his Chapter 4 to “Growth of Population: Checked and Un- checked” (Indian Census 1953). He drew attention to the inadequacy of “the natural checks on population growth, famine and disease” and to the fact that the principal and more satisfactory check in Europe was now contraception. It had been on the receipt of the preliminary results of the 1951 census together with Gopalaswami’s interpretation of them that Prime Min- ister Nehru made his final decision to announce the world’s first national family planning program. Others also drew on the censuses to examine the Indian population situation. P. K. Wattal, of the Indian Audit and Account Service, published two books almost half a century apart, both concentrating on census data and reports, and both starting with Malthus and the Law of Population (Wattal 1916, 1958).

The impact on India of Malthusian and neo-Malthusian thought

The British and Indian elites were in intimate contact: both spoke English, they were educated from the same syllabuses, and their newspapers drew

Click to return to Table of Contents J OHN C. CALDWELL 687 on the same news services. University education in the Presidency colleges, molded on English institutions, went back to the 1850s. The proportion of Indians in the higher levels of government, the Indian Civil Service (In- dian Administrative Service, after Independence), increased steadily, but particularly rapidly after the 1935 Government of India Act. Interviews with retired members of the service suggest that by the 1930s few of either the English or Indian members doubted that the huge and growing Indian population threatened India’s food supplies and progress. The emerging political elites shared this view: Jawaharlal Nehru had often been heard to say that “India would have been a much more advanced nation if its population were about half its actual size” (quoted in Mamoria 1961: 388). There was little in the strongly secular thought of either the British or Indian ruling classes or administrative elites that opposed Malthusian or neo- Malthusian solutions. Even Gandhi’s reservations about the methods to be used echoed Malthus’s feelings. Documents that the Civil Service regarded as their own began to favor government intervention to slow population growth to an extent that was unparalleled anywhere else in the world. Hutton, writing the report of the 1931 Indian Census (1933), had raised the issue of government involvement in family planning. Shortly afterwards, Sir John Megaw, the former Director of Medical Services in India, wrote a report saying that Britain should introduce family planning into India (Sanger 1938: 464). Such views were supported in 1945 by the Final Report of the inquiry into the Bengal Famine (India Famine Inquiry Commission 1945: 385–386), and in 1946 by the prestigious Bhore Committee’s investigation of India’s health needs (India Health Survey and Development Committee 1946: vol. 2, pp. 483–487). These documents were all drawn upon for a 1947 report by the population subcommittee of the Congress Party’s National Planning Commit- tee, which was published as a book, and which advocated government intervention in favor of family planning in the medical and informational sectors (Shah 1947: 14, 113). All these documents drew on the Malthusian discussions in the censuses and other official documents, but, as in Britain in the mid-nineteenth century, the Malthusian conclusions had become so much a part of mainstream thought that there was often neither mention of, nor debate with, their progenitor. An American, Kingsley Davis (1946), wrote a paper that was to be influential in Indian planning circles and that posed the Malthusian dilemma of either “a calamitous rise in the death rate or a fall in fertility” (p. 243); he expected the former (p. 253) although he did identify some restriction of fertility arising from the Hindu taboo on widow remarriage (p. 254). These writings set the stage for Prime Minister Nehru’s announcement in 1951 that India would establish a national family planning program from the beginning of the 1952 five-year planning period— an announcement met without objection except by the Gandhian Minis- ter for Health, who objected to artificial means of fertility control.

Click to return to Table of Contents 688 MALTHUS AND THE LESS DEVELOPED WORLD

Parallel developments occurred within influential nongovernmental sectors of the population. In 1916 a students’ Eugenics Association was formed at Madras Presidency College. Two members of the association, the Maha- rajah of Mysore and Lady Rama Rao, thereby received their first exposure to neo-Malthusian ideas, which were to lead to the former establishing the world’s first government birth control clinics in Mysore from 1930 and the latter becoming the founding president of the International Planned Par- enthood Federation in 1953 (Caldwell and Caldwell 1986: 39). D. Ghosh (1946) published Pressure of Population and Economic Efficiency in India, re- plete with Malthusian concepts and references to “checks” and “moral re- straints” (Ghosh 1946: 102 ff). Baljit Singh (1947) expressed the Malthu- sian and Gandhian view that the program for population growth constraint should center on “social attitudes towards marriage, sex relationship and family based on an elevation of social values “ (Singh 1947: 128). The 1935 annual meeting of the All-India Women’s Conference called for a birth control program and invited Margaret Sanger to address their 1936 meeting “to assist them to put teeth into” their proposals (Sanger 1938: 461). India was the key to developing-world movements toward fertility control. Thus, although an unsuccessful meeting was held in 1948 in Cheltenham to attempt to set up an international family planning movement, a very successful meeting was held in Bombay (Mumbai) four years later. It gained great strength from its location in India, the strong delegation that Lady Rama Rao was able to bring from the All-India Women’s Conference, the organizing involvement of Margaret Sanger, and above all, the fact that Nehru had already announced that India would have a national population program. He sent a sympathetic, but carefully worded message to the meeting. As a result of the 1952 conference the Interna- tional Planned Parenthood Federation was established the following year; it had three developing-world members, India and two Asian British colonies, Singapore and Hong Kong.

Discussion

I want to elaborate upon two points addressed above: (1) Malthus and the Anglo–Indian connection and (2) the significance of Malthus in the move to curb developing-world population growth over the last half-century. It is no accident that Malthus wrote in English. The way had been prepared by Adam Smith and others. But the way had been prepared for Smith by the growth of trade in the English-speaking world: in North America, throughout the British Empire, and in the North Atlantic Trian- gular Trade system. It had also been prepared by the success of the trading middle classes in England in the 1642–45 civil war, the 1688 Glorious Revolution, and

Click to return to Table of Contents J OHN C. CALDWELL 689 the 1832 Great Reform Act, and in the 1776–83 American Revolution and the 1789 Constitution. All these changes and Malthus’s own writings occurred in a Protestant religious context, the latter specifically in an Angli- can one. Without Thomas Robert Malthus, there would soon have been another, but he might not have written with such clarity and almost certainly without such a conjunction of ideas. The English-speaking world became the nineteenth-century cradle for antinatalist thought. A partial exception to that generalization is the neo- Malthusian activities in the Netherlands, but they were backed by little indigenous political economic philosophy and the Dutch had no close in- terrelations with Indonesia of the type that developed between Britain and India. The main theater for playing out these antinatalist interests was Britain’s huge colony, India. The establishment of the Indian national family planning program in 1952 was, in turn, of prime importance in the development of the world antinatal movement; its pattern of moral leadership by political and social elites and its top-down organization became a model for Asian programs (see Caldwell 1993). By this time, links in the fertility control field were being developed with India by experts from America as well as from Britain. Margaret Sanger had chosen India and Japan in the 1930s as the non-Western countries most likely to heed her message. When Nehru set up a committee in 1951 to advise on the need for, and design of, the family planning content in the 1952–56 Five Year Plan, they sought advice and reports from three Americans, Kingsley Davis, Pascal Whelpton, and William Ogburn (Caldwell and Caldwell 1986: 40). Davis had just published his Population of India and Pakistan (Davis 1951). The impulse begun by Malthus continued and partly explains the central importance of English-language writers in stirring interest in world population growth after World War II and the early involvement of Ameri- can and British technical aid in family planning. By 1965 the citizens of former British colonies in sub-Saharan Africa knew far more about family planning than did those in other countries in the region (Caldwell 1966); and indeed four of these countries, South Africa, Zimbabwe, Botswana, and Kenya, led the regional fertility decline, South Africa from 1960, the other three from the 1980s. In spite of rapidly declining death rates worldwide after World War II and evidence that food production was increasing as fast as population, the most widely read literature with probably the greatest policy impact focused on the race between population and food supplies. Malthus’s name was frequently employed. William Vogt wrote in his Road to Survival (1949: 72): “And hand in hand with famine will stalk the shade of that clear-sighted English clergyman, Thomas Robert Malthus.” Indeed, the clergyman stalks much of the book. Sir John Russell in World Population and Food Supplies (1954: 5) named,

Click to return to Table of Contents 690 MALTHUS AND THE LESS DEVELOPED WORLD as thinkers in the field, Malthus and Sir William Crookes, who in an 1898 address pointed out that wheat eaters were increasing faster than wheat production. Russell’s analysis did, however, present a fairly optimistic picture, at least in the short term. Karl Sax began Standing Room Only (1955) with the “Malthusian Laws.” By 1968 Paul Ehrlich opened The Population Bomb with the sentence, “The battle to feed all humanity is over.” Interestingly, although he pursued Malthusian themes, even that of the class differential in starvation (p. 17), he, unlike most other popular authors on this theme, made no reference to the inventor of the principle of population. In contrast, Garrett Hardin, in “The tragedy of the commons,” attributes much to Malthus’s leadership. He exceeds the master in declaring that “Freedom to breed is intolerable” (Hardin 1968: 1246) and that the solution is “Mutual coercion mutually agreed upon” (p. 1247), but comes closer to Malthus’s belief in self-restraint in the section of his essay on “How to legislate temperance?” (p. 1245). More-recent writings are distinguished by an approach that is understandably more global and less national than Malthus’s, but that adds to the problem of food those of water, erosion, air pollution, absorption of waste, and increasing scarcities in the forest and energy sectors. This was the approach of the Club of Rome (Meadows et al. 1972). It is also that of Lester Brown, as, for instance, in “The new world order” (Brown 1991), which does, however, give food supplies centrality with ample reference to Malthus (e.g., Brown 1991: 15 ff). The divergence from Malthus has been greater in the academic literature, where most competing theory dates only from the 1940s. Most demographers accept some form of Frank Notestein’s demographic transition theory, which placed a greater emphasis on the role of fertility in deciding whether societies survived or not. That theory was summarized by Ansley Coale and Edgar Hoover in the introduction to their book Population Growth and Economic Development in Low-Income Countries (1958: 9–10):

[In an agrarian peasant economy] death rates usually fluctuate in consequence of variations in crops, the varying incidence of epidemics, etc. In such an economy birth rates are nearly stable at a high level. Death rates are high as a consequence of poor diets, primitive sanitation, and the absence of effective preventive and curative medical practices. High birth rates result from social beliefs and customs that necessarily grow up if a high death-rate community is to continue in existence. These beliefs and customs are reinforced by the economic advantages to a peasant family of large numbers of births.

They went on to comment that an agrarian low-income society “has a mortality and fertility pattern that fits pretty closely the conditions which Malthus thought . . . to be a universal tendency: high birth and high death

Click to return to Table of Contents J OHN C. CALDWELL 691 rates. Growth of population is usually slow” (p. 10). This emphasis on the primary role of fertility decisionmaking was congenial in circumstances of twentieth-century interventions in such matters, but it does give rise to some anomalies. It implies that in traditional societies the only determinant of equilibrium in mortality–fertility levels was mortality, with fertility being raised to meet its levels. It ignores the more hopeful formulation of Malthus that marriage restraint by reducing fertility could in turn reduce the force of the positive checks and hence the mortality levels, resulting in a more tranquil society with moderate fertility and mortality levels. Coale and Hoover’s is not strictly an equilibrium model because it allows for societies that fail to keep fertility sufficiently high; but these disappear, leaving only societies where mortality alone determines both mortality and fertility levels. This is the Malthusian model without the advantage of the preventive check. In the same book, Coale and Hoover offer a theory for contemporary developing-world societies, one where a reduction of the birth rate, achieved at least partly by government intervention, slows the rate of population growth. The gains from this are not confined to raising nutritional levels but extend to allowing investment in broad economic development. Ester Boserup published, in 1965, The Conditions of Agricultural Growth, often taken to be a theory that dispenses with the need for the principle of population. Indeed, the author herself took this view:

The neo-Malthusian school has resuscitated the old idea that population growth must be regarded as a variable dependent mainly on agricultural output. I have reached the conclusion . . . that in many cases the output from a given area of land responds far more generously to an additional input of labour than assumed by neo-Malthusian authors. If this is true, the low rates of population growth found (until recently) in pre-industrial communities cannot be explained as the result of insufficient food supplies due to overpopulation, and we must leave more room for other factors in the explanation of demographic trends . . . —medical, biological, political etc.—which may help to explain why the rate of growth of population in primitive communities was what it was. (Boserup 1965: 14)

In fact, although Malthus regarded the potential for food production as setting the bounds for population growth, he held that in many societies famines are rare because the positive checks hold population numbers below the ceiling imposed by potential food production. If the checks are reduced, the extra labor will indeed result in greater food production until the agricultural limit is reached, famine results, and, because of starvation and heightened levels of epidemic disease, population numbers decline. Boserup argues that there is an alternative path, namely that the subsistence crises force human beings to innovate by turning to discoveries al-

Click to return to Table of Contents 692 MALTHUS AND THE LESS DEVELOPED WORLD ready known but underutilized, and so change the agricultural mode of production to allow greater production from a given unit of land provided that there are greater labor inputs. The argument that population growth has not been steady over the millennia but that there have been periods of relatively fast growth is undoubtedly true. But whether this observation does more than add to Malthusian theory is far less certain. The fact that population rises rapidly only when new modes of cultivation allow food supplies to increase unusually fast seems to support the concept of a Malthu- sian food-based population ceiling. Boserup cannot mean that every famine produces a change in the modes of production, or else all of human history, or at least its changes in agricultural systems and vast increases in population numbers, would have taken centuries rather than millennia. Nor can she mean that population growth has been determined solely by free-floating mortality checks, determined by nothing beyond themselves, for, in this case, the human race might again have compressed its whole history or have become extinct. Many anthropologists have been much taken with the view that primitive food-collecting and hunting populations were not in a high-mortality state of misery as demanded by Malthusian theory, but had access to such ample food that they spent relatively little time meeting their nutritional needs. This derived partly from Boserup’s (1965) thesis that the simpler modes of cultivation needed smaller labor inputs, partly from Richard Lee and Irven DeVore’s (1968) Man the Hunter, and partly from Marshall Sahlins’s (1974) Stone Age Economics, although much of the argument can be found in Alexander Carr-Saunders’s 1922 work, The Population Problem. Sahlins went so far as to describe this situation as primitive affluence. It is unlikely that these descriptions of the mortality and state of welfare among premodern hunting and gathering cultures are the equal of Malthus’s. We have criticized the anthropological data sources and conclusions elsewhere (Caldwell, Caldwell, and Caldwell 1987: 31–32) and have identified a failure to fully comprehend the unrepresentative situation of niche hunting and gathering groups in the modern world, and a selective use of the older reports, as helping to present an unsound, and far too optimistic picture. Primitive affluence theorists are hard put to explain long-term near-stationary population among hunting and gathering populations, a group to which Malthus devoted considerable attention. Malthus was actually closer to Notestein with regard to fertility and to Boserup with regard to mortality than either seemed to realize, as is shown by this passage from the 1830 modification of the 1824 supplement to the Encyclopaedia Britannica:

The frequency of wars and the dreadful devastations of mankind occasioned by them, united with the plagues, famines, and mortal epidemics . . . must

Click to return to Table of Contents J OHN C. CALDWELL 693

have caused such a consumption of the human species that the exertion of the utmost power of increase must, in many cases, have been insufficient to supply it; and we see at once the source of those encouragements to marriage and efforts to increase population which . . . distinguished the legislation and general policy of ancient times. (Malthus 1960: 41)

Finally, it is pertinent to ask which political theories induced developing-world leaders to try to control population growth. I have been associated over the last 30 years with interrelated projects seeking information from those involved at the political or governmental levels with these decisions.2 There is little doubt that in the densely settled agrarian lands of South Asia the dominant factor was the concept of a race between population and food, and of the need for higher nutritional levels. Among the survivors of the pre-1947 Indian Civil Service, which subsequently provided the upper echelons of the administrative services of both India and Pakistan, and after 1970 of Bangladesh, attitudes, shared with the now expatriated Brit- ish members of the service, were usually quite consciously Malthusian. The genesis of this essay springs from that discovery. President Ayub Khan, when interviewed in early 1969, explained the emphasis his administration in Pakistan placed on curbing fertility in terms of feeding the people. Subsequent research on the origins and success of the Bangladesh family planning program showed that officials first in East Pakistan, and then in Bangladesh, were convinced that the family planning program had to be pursued vigorously because of the visible density of the population. The 1974–75 Bangladesh famine strengthened these feelings, just as memo- ries of the 1943–44 Bengal famine had been a factor in the 1951 Indian decision to opt for a family planning program. Chinese officials explain their present program more often than not in terms of needing the capacity to feed themselves. Elsewhere, in Southeast Asia, North Africa, and sub-Sa- haran Africa and in other less densely settled countries, the justifications for programs are undoubtedly more complex with a greater emphasis on broader economic development. But in the successor states of ancient agrarian civilizations, Malthusian beliefs and fears, while diminished, remain of significance and ready to reappear when crisis threatens.

Notes

Most of the work underlying this essay was National Academies Forum and the National undertaken with Pat Caldwell, and some of Library of Australia to mark the 200th An- it with Bruce Caldwell, Barkat-e-Khuda, and niversary of the publication of Malthus’s First Indrani Pieris. Assistance has also been pro- Essay. vided by Wendy Cosford, Jeff Marck, and 1 Where possible the quotations from Rita Tanusoski. This essay was prepared for Malthus have been taken from The Works of the conference organized by the Australian Thomas Malthus, ed. E. A. Wrigley and David

Click to return to Table of Contents 694 MALTHUS AND THE LESS DEVELOPED WORLD

Souden, Pickering, London, 1986, although 2 This research began as a 1968–69 the series edited by Patricia James, Cam- Population Council project and later was a bridge University Press, 1989, and Donald Ford Foundation–supported project in 1981– Winch’s version of Malthus’s second edition 83. It is reported in Caldwell and Caldwell (1803), Cambridge University Press, 1992, (1986). The East Pakistan/Bangladesh inter- have also been consulted. Malthus 1960 views carried out from 1969 and again in- (Mentor Books, Three Essays on Population, tensively in 1995–96 are reported in Khuda New American Library, New York) has been et al. (in press). used for the 1830 version of the 1824 supplement to the Encyclopaedia Britannica.

References

Ambirajan, S. 1976. “Malthusian population theory and Indian famine policy in the nineteenth century,” Population Studies 30, no. 1: 5–14. Bagehot, Walter. 1869. Physics and Politics, or The Thoughts on the Application of the Principles of “Natural Selection” and “Inheritance” to Political Society. London: Kegan Paul, Trench, Trübner. Bonar, James. 1924 [1885]. Malthus and His Work. London: Allen and Unwin. Boserup, Ester. 1965. The Conditions of Agricultural Growth: The Economics of Agrarian Change under Population Pressure. Chicago: Aldine. Breton, Yves. 1983. “Malthus and underdevelopment,” in Dupâquier, Fauve-Chamoux, and Grebenik 1983, pp. 247–253. Brown, Lester R. 1991. “The new world order,” in Lester R. Brown et al., The State of the World 1991: A Worldwatch Institute Report on Progress Toward a Sustainable Society. New York: Norton, pp. 3–20. Caldwell, J. C. 1966. “Africa,” in Family Planning and Population Programs: A Review of World Developments, ed. B. Berelson et al. Chicago: University of Chicago Press, pp. 163–181. ———. 1993. “The Asian fertility revolution: Its implications for transition theories,” in The Revolution in Asian Fertility: Dimensions, Causes and Implications, ed. R. Leete and I. Alam. Oxford: Clarendon Press, pp. 299–316. Caldwell, John and Pat Caldwell. 1986. Limiting Population Growth and the Ford Foundation Contribution. London: Pinter. Caldwell, John, Pat Caldwell, and Bruce Caldwell. 1987. “Anthropology and demography: The mutual reinforcement of speculation and research,” Current Anthropology 28, no. 1: 25–43. Campbell, R. H. and A. S. Skinner. 1976. “General introduction,” in Smith 1976, pp. 1–60. Carey, John. 1992. The Intellectuals and the Masses: Pride and Prejudice among the Literary Intel- ligentsia, 1880–1939. London: Faber and Faber. Carr-Saunders, Alexander M. 1922. The Population Problem: A Study in Human Evolution. Ox- ford: Clarendon. Coale, Ansley J. and Edgar M. Hoover. 1958. Population Growth and Economic Development in Low-Income Countries: A Case Study of India’s Prospects. Princeton: Princeton University Press. Cunningham, J. R. 1941. “Education,” in O’Malley 1941, pp. 138–187. Davis, Kingsley. 1946. “Human fertility in India,” American Journal of Sociology 52: 243–254. ———. 1951. The Population of India and Pakistan. Princeton: Princeton University Press. Dubois, J. A. 1906 [1816]. Hindu Manners, Customs and Ceremonies, trans. H. K. Beauchamp. Oxford: Clarendon. Dupâquier, J., A. Fauve-Chamoux, and E. Grebenik (eds.). 1983. Malthus Past and Present. London: Academic Press.

Click to return to Table of Contents J OHN C. CALDWELL 695

Ehrlich, Paul R. 1968. The Population Bomb. New York: Ballantine. Fryer, Peter. 1965. The Birth Controllers. London: Secker and Warburg. Ghosh, D. 1946. Pressure of Population and Economic Efficiency in India. Bombay: Oxford Uni- versity Press for Indian Council of World Affairs. Glass, D. V. 1953. “Malthus and the limitation of population growth,” in Introduction to Malthus, ed. D. V. Glass. London: Watts, pp. 25–54. Godelier, Maurice. 1983. “Malthus and ethnography,” in Dupâquier, Fauve-Chamoux, and Grebenik 1983, pp. 125–150. Hardin, Garrett. 1968. “The tragedy of the commons,” Science 162: 1243–1248. Hollingsworth, T. H. 1983. “The influence of Malthus on British thought,” in Dupâquier, Fauve-Chamoux, and Grebenik 1983, pp. 213–221. Huxley, Julian. 1960. “World population,” in Malthus 1960, pp. 63–81. New York: Mentor (first published in Scientific American, March 1956). India Famine Inquiry Commission. 1945. Final Report. Madras: Government Press. India Health Survey and Development Committee [Bhore Committee]. 1946. Report of the Health Survey and Development Committee. Delhi: Government of India Press. Indian Census. 1870. Report of the Census of the Punjab 1868. Lahore. ———. 1875. Report on the Census of British Burma 1872. Rangoon. ———. 1877. Report on the Census of Native Cochin 1875. Madras. ———. 1878. Report on the Census of Bengal 1876. Calcutta. ———. 1882. Report on the Census of the Northwest Province and Oudh 1881. Allahabad. ———. 1883a. Report on the Census of the Baroda Territories 1881. Byculla. ———. 1883b. Report on the Census of the Presidency of Madras 1881. Madras. ———. 1884. Report on the Mysore Census 1881. Bangalore. ———. 1893. General Report on the Census of India 1891, by J. A. Baines. London: Eyre and Spottiswoode. ———. 1924. Report on the Census of India, 1921, by J. T. Marten. Calcutta. ———. 1933. Report, Census of India 1931, by J. H. Hutton. Delhi: Government of India. ———. 1953. Report on the Census of India 1951, by R. A. Gopalaswami. New Delhi. Kane, Penny. 1995. Victorian Families in Fact and Fiction. London: Macmillan. Keynes, John Maynard. 1951. Essays in Biography. London: Rupert Hart-Davis. Khuda, Barkat-e-, John C. Caldwell, Bruce K. Caldwell, Indrani Pieris, Pat Caldwell, and Shameem Ahmed. In press. “The global fertility transition: New light from the Bang- ladesh experience,” in Comparative Perspectives on Fertility Transition in South Asia, ed. IUSSP Committee on Fertility and Family Planning. Oxford: Clarendon Press. Lee, Richard B. and Irven DeVore (eds.). 1968. Man the Hunter. Chicago: Aldine. Malthus, Thomas R. 1960 [1830]. “A summary view of the principle of population,” in Three Essays on Population. New York: Mentor, pp. 13–59 (first published 1830 as a slightly abridged version of the 1824 supplement to the Encyclopaedia Britannica). ———. 1986a. [1798]. The Works of Thomas Robert Malthus, vol. 1, An Essay on the Principle of Population, ed. E. A. Wrigley and David Souden. London: William Pickering. ———. 1986b [1826]. The Works of Thomas Robert Malthus, vol. 2, An Essay on the Principle of Population, 6th edition, ed. E. A. Wrigley and David Souden. London: William Pickering. ———. 1989. An Essay on the Principle of Population, 2 vols., ed. Patricia James. Cambridge: Cambridge University Press. ———. 1992. [1803]. An Essay on the Principle of Population, 2nd edition, selected and introduced by Donald Winch. Cambridge: Cambridge University Press. Mamoria, C. B. 1961. India’s Population Problem. Allahabad: Kitab Mahal. Meadows, Donella H., Dennis L. Meadows, Jørgen Randers, and William W. Behrens. 1972. The Limits to Growth: A Report for the Club of Rome’s Project on the Predicament of Mankind. London: Earth Island. Mill, John Stuart. 1848. Principles of Political Economy with Some of Their Applications to Social Philosophy. London: Standard Library.

Click to return to Table of Contents

696 MALTHUS AND THE LESS DEVELOPED WORLD

Mitchell, B. R. and Phyllis Deane. 1962. Abstract of British Historical Statistics. Cambridge: Cambridge University Press. O’Malley, L. S. S. (ed.). 1941. Modern India and the West: A Study of the Interactions of Their Civilizations. London: Oxford University Press. Ortega y Gasset, José. 1932. The Revolt of the Masses. New York: Norton. Place, Francis. 1994 [1822]. Illustrations and Proofs of the Principle of Population: Being the First Work on Population in the English Language Recommending Birth Control. London: Rout- ledge/Thoemmes. Rawlinson, H. G. 1941. “Indian influence on the West,” in O’Malley 1941, pp. 535–566. Russell, E. John. 1954. World Population and Food Supplies. London: Allen and Unwin. Sahlins, Marshall. 1974. Stone Age Economics. London: Tavistock. Sanger, Margaret. 1938. An Autobiography. New York: Norton. Sax, Karl. 1955. Standing Room Only: The Challenge of Overpopulation. Boston: Beacon. Seeley, John. 1885. The Expansion of England. London. Shah, K. T. (ed.). 1947. Population: Report of the Population Sub-committee of the National Plan- ning Committee. Bombay: Vora. Singh, Baljit. 1947. Population and Food Planning in India. Bombay: Hind Kitabs. Smith, Adam. 1976 [1776]. An Inquiry into the Nature and Causes of the Wealth of Nations. Oxford: Clarendon. Vogt, William. 1949. Road to Survival. London: Gollancz. Wattal, P. K. 1916. The Population Problem in India: A Census Study. Bombay: Bennett, Coleman and Co. ———. 1958. Population Problem in India: A Census Study. New Delhi: Minerva. Winch, Donald. 1992. “Introduction,” in Malthus 1992, pp. vii–xxiii. Wrigley, E. A. 1983. “Malthus’s model of a pre-industrial economy,” in Dupâquier, Fauve- Chamoux, and Grebenik 1983, pp. 111–124.

Click to print article Click to return to Table of Contents

Population and Reproductive Health: An Economic Framework for Policy Evaluation

JERE R. BEHRMAN JAMES C. KNOWLES

THIS ARTICLE PRESENTS and illustrates the use of a standard simple economic framework for the evaluation of policies on population and reproductive health services and contrasts this framework with current discussions of several issues in these fields. There is a long tradition of economic analysis in the population and reproductive health fields. Nevertheless, a number of current policy discussions could benefit from some realignment to become more consistent with the broader field of human resource policy analysis. We think that there are two gains from such a realignment. First, the use of the standard economic framework for policy analysis would facilitate communication between analysts and policymakers working in the population and reproductive health community and economists and policymakers not working in these fields who are unfamiliar with the way policy issues are conceptualized by population and reproductive health sector specialists. Improving this communication is important because of ongoing policy discussions related to the extent to which scarce public resources should be directed toward population and reproductive health versus a large number of competing claims. Such discussions are likely to take place increasingly with ministries of finance and planning in developing countries, with key staff of development banks, and with policy and program specialists in bilateral donor agencies. Absent a realignment of the kind advocated in this article (or at least the ability to adapt analyses to these audiences), such interactions run the risk of becoming increasingly frustrating. Researchers working in population and reproductive health will be more effective policy analysts, and policymakers using their work will be more effective advocates for policies promoting family planning and reproductive health services, if the standard economic framework is used more

POPULATION AND DEVELOPMENT REVIEW 24(4):697–737 (DECEMBER 1998)697

Click to print article Click toClick return to returnto Table to of Table Contents of Contents 698 POPULATION AND REPRODUCTIVE HEALTH in the future. Second, even discussions among policymakers and analysts in the population and reproductive health community generate some confusion because similar terms are used in different ways and not always within a unified analytical framework. In our judgment the standard economic policy framework is a useful one with which to contemplate policy issues in population and reproductive health, and its use will increase clarity in some areas in which there now seems to be confusion. The intended audience of this article is economists engaged in preparing policy analyses for the benefit of the population and reproductive health community. In addition, the article may be of interest to a broader audience of policy analysts and policymakers in population and reproductive health who are interested in how economists think about policy issues and the implications of such thinking for the programs that concern them. The article is not intended to convince noneconomists that a standard economic framework is superior to that of any other discipline in conceptual- izing problems and formulating effective policies in the population and reproductive health field. We make no effort to compare alternative frameworks for analyzing policies. Instead, our objective is to clarify where some of the differences arise in the hope that this will lead to more effective communication across disciplines. Of course, with improved communication comes a high probability that policies will be more carefully conceived, implemented, and evaluated. The first section of the article presents a simple standard economic framework for thinking about policy in the social sectors generally and in population and reproductive health specifically. The material in this section is not new, and many readers with prior training in economics will be familiar with its contents. The section pulls together material from many sources, however, and provides the conceptual framework for the rest of the article. The second section illustrates how the framework can be applied in practice, using case studies from Indonesia and Vietnam. We then apply the policy framework to several key population and reproductive health policy issues (viz., pricing of services and cost recovery; access and quality; the role of the private sector; targets, incentives, and unmet need; decentralization; sustainability; and evaluation) and identify where current analysis in the field appears to be inconsistent with the policy framework. Finally, we provide some tentative conclusions about the direction of economic policy in population and reproductive health and identify potentially fruitful directions for future analyses.

Economic framework for evaluating policies in population and reproductive health

Often analyses of social-sector policies are undertaken without consideration of the general rationale for policies. It is presumed that policies that,

Click to return to Table of Contents J ERE R. BEHRMAN / JAMES C. KNOWLES 699 say, increase access to family planning or education or improve health and nutrition must be good. But such analyses are of little help in convincing skeptics that scarce resources should be allocated for these purposes, given the many competing uses to which they can be put. Moreover, they may not provide much in the way of guidelines for choosing among policy alternatives in the population and reproductive health fields. It is useful to begin by asking why policy interventions in these areas (and in other areas of human resource development) might be desirable. For most economists the two possible justifications for governmental policy interventions in these and other areas are: 1) to increase efficiency and productivity; and 2) to redistribute resources.

Efficiency and productivity

Resources are used efficiently in the economic sense of the term if they are used to obtain the most product possible given the quantities of the resources and the available production technologies and if the composition of that product increases the welfare of members of society as much as possible given the resource and technological constraints and the distribution of resource ownership. An investment (or expenditure) in population and reproductive health services is efficient if the additional (marginal) social benefits of the last unit of that investment just equal its additional (marginal) social cost.1 If the social marginal benefit of a particular investment is greater (less) than the social marginal cost, society is not investing enough (is investing too much) and would benefit from increasing (decreasing) the level of investment until the social benefits and costs become equal on the margin. Although applying the above rule maximizes social gains, private maxi- mizing behavior leads to population and reproductive health investments at the level at which the private marginal benefit of the investment equals its private marginal cost under the assumption that, given the information available to them and the constraints that they face, individuals act in what they perceive to be their best interests. Figure 1 provides an illustration for one individual. The private marginal benefit curve depends on the expected private gains as perceived by consumers from investing in one type of reproductive health service. These may reflect a desire for improved health, but they are also likely to reflect such other factors as the time required to consume the service, convenience, social customs, and imperfect information. The private marginal benefit curve is downward-sloping because of diminishing returns to such investments (i.e., benefits do not increase proportionately with additional expenditure as a consequence, for example, of fixed genetic health endowments). The private marginal cost curve may be increasing if there are increasing opportunity costs to devoting additional time to such investments, but the analysis applies equally to cases in

Click to return to Table of Contents 700 POPULATION AND REPRODUCTIVE HEALTH

FIGURE 1 Private marginal benefits and private marginal costs of human resource investments

Marginal benefits Marginal costs

Private marginal benefits

Private marginal costs

H* Human resource investments which they are constant. Private returns net of costs are maximized at level H*, where private marginal benefits are equal to private marginal costs. Now consider what happens if the private incentives for population and reproductive health investments differ from the social incentives for such investments, first with respect to the marginal benefits and then with respect to the marginal costs. Let the dashed line in Figure 2 represent the social marginal benefits for population and reproductive health investments that are drawn to be greater than the private marginal benefits.2 In this case the private incentives are to invest at level H*, which is less than the socially optimal (efficient) level of investment at level H**. Therefore there is an efficiency argument for policies to induce or to require private investments at level H** instead of level H*. Why might social marginal benefits exceed private marginal benefits for population and reproductive health investments? Among the most frequent answers are: 1. Some health and nutrition interventions, in addition to their direct impact on the individual’s productivity and welfare, may reduce suscepti-

Click to return to Table of Contents J ERE R. BEHRMAN / JAMES C. KNOWLES 701

FIGURE 2 Private marginal benefits and private marginal costs of human resource investments, with social marginal benefits

Marginal benefits Marginal costs

Private marginal benefits

R** R*

Private marginal costs Social marginal benefits

H* H** Human resource investments bility to contagious diseases that, if contracted, may spread to others.3 This is particularly relevant in the case of investments in the prevention and treatment of infectious diseases, such as HIV and other sexually transmitted diseases. 2. Information on which private investment decisions are made may misrepresent the private rates of return to these investments because it is incomplete or incorrect.4 This is an important consideration in the case of population and reproductive health investments because actual and potential consumers are often not well informed about the benefits and costs.5 3. There may be social gains beyond the private gains to population and reproductive health investments if such investments are viewed by society as inherently valuable (e.g., to facilitate catching up with other rapidly growing economies) in addition to their effects on individual economic productivity and consumption. 4. The combination of uncertainty, risk aversion, and imperfect insurance markets may result in private incentives to underinvest in health from a social point of view because from a social point of view the risks are pooled.

Click to return to Table of Contents 702 POPULATION AND REPRODUCTIVE HEALTH

FIGURE 3 Private marginal benefits and private marginal costs of human resource investments, with social marginal costs

Marginal benefits Marginal costs

Private marginal Private benefits marginal costs

Social R** marginal costs

H* H*** Human resource investments

5. In the case of fertility, governmental subsidies to health care, schooling, and other social services may reduce the private costs of having additional children, thereby reducing the private marginal benefits to parents of expenditure on family planning services. 6. Household decisionmakers may not duly consider the preferences of all household members in allocating household resources. For example, the preferences of women for fewer or more widely spaced children or for safer childbirth may not be taken into consideration by husbands.6 Now let the dashed line in Figure 3 represent the social marginal costs for population and reproductive health investments that are drawn to be less than the private marginal costs.7 In this case the private incentives are to invest at level H*, which is less than the socially optimal (efficient) level of investment at level H***. Therefore there is an efficiency argument to consider the possibility for policies to induce or to require private investments at level H*** instead of at level H*. Why might social marginal costs be less than private marginal costs for population and reproductive health investments? Among the most frequent answers are:

Click to return to Table of Contents J ERE R. BEHRMAN / JAMES C. KNOWLES 703

1. There may be capital market imperfections for investments in health (in part because human capital is not recognized as collateral) such that the private marginal costs for such investments exceed their true social marginal costs, probably more so for individuals from poorer families who cannot easily self-finance such investments. This may pose a problem, for example, for risky investments in such areas as safe motherhood (e.g., emergency obstetric services) that involve substantial outlays with some of the benefits reaped only after considerable delays under conditions in which capital and insurance markets do not function well or are nonexistent. 2. The sectors that provide population and reproductive health services may produce them inefficiently because institutional arrangements do not induce efficient production of an efficient basket of commodities. This is mainly a problem where services are provided directly (and often inefficiently) by governments, or are heavily regulated; but it can also be an outcome of monopoly pricing.

Distribution

Distribution is a major policy motive distinct from efficiency. Society well might want to ensure that everyone has basic social services even at some cost in reduced efficiency and productivity. Sometimes family planning, education, health, and nutrition services are viewed as “merit goods” that are socially desirable in themselves (e.g., the UNDP Human Development Re- ports take this approach).

Policy choices to increase efficiency and to improve distribution

Consider first efficiency. If all other markets in the economy are operating efficiently and there are differences between marginal private incentives and marginal social incentives in population and reproductive health service markets so that private incentives are to invest at level H* instead of at level H** in Figure 2 or level H*** in Figure 3, policies that increase the human capital investment to the socially efficient levels increase efficiency and productivity. If all other markets in the economy are not operating efficiently, then policies that narrow the differences between private and social incentives in population and reproductive health service markets do not necessarily increase efficiency and productivity. It is conceivable that the distortion between private and social incentives in this market is simply offsetting some distortion elsewhere, so that the economy will become less efficient and less productive if this distortion by itself is lessened.8 Because there always are distortions between private and social incentives elsewhere in the economy, this may seem to be bad news for establishing an efficiency/productivity basis for policy. But, in the absence of specific infor-

Click to return to Table of Contents 704 POPULATION AND REPRODUCTIVE HEALTH mation to the contrary, such as the existence of two counterbalancing distortions, a reasonable presumption is that lessening any one distortion between social and private incentives is likely to increase efficiency and productivity. The foregoing still does not indicate what policies would be best to induce population and reproductive health investments at level H** in Fig- ure 2 or level H*** in Figure 3. There is a large set of possibilities, including governmental fiats, governmental provision of social services such as schools and health clinics at heavily subsidized prices, price incentives in the markets for population and reproductive health services, price incentives in other markets, and changing institutional arrangements in various markets. To choose among alternatives, there are two important considerations. First, it is necessary to realize that policies have costs. These costs are not only the direct costs of implementing and monitoring the policy, but also the distortionary costs introduced by policies that may encourage socially inefficient behavior (including rent-seeking9 by both public and private entities).10 In fact the costs may be sufficiently high that it is not desirable to try to offset some market failures by introducing policies.11 But, if it is desirable to do so, there is a case for making policy changes that are directed as specifically as possible to the distortion of concern because that tends to lessen the distortion costs. An efficiency policy hierarchy can be defined in which alternative policies to attain the same improvement in efficiency in a particular market are ranked according to their social marginal costs, including direct and distortion costs. A policy is higher in the policy hierarchy if it has lower social marginal costs. This hierarchy indicates the preferential ordering of policies to deal with particular divergences between private and social incentives. Second, there are far-reaching information problems regarding what effects policies have, particularly in a rapidly changing world. This is an argument in favor of policies that are as transparent as possible, which generally means higher in the policy hierarchy with regard at least to distortion costs because more direct policies are likely to be more transparent. This also is an argument for considering an experimental approach to evaluating policy alternatives—that is, rather than introducing a reform country-wide, for introducing variants of reforms for health clinics, schools, and other social services in randomly selected sites with careful monitoring of the results for both the experimental groups and the control groups. Infor- mation problems also provide an argument for price policies (taxes or subsidies) because shifts in the underlying relations are likely to be more visible to policymakers if they have an impact on the governmental budget than if they only change the distortions faced by private entities, as tends to happen with quantitative regulatory policies such as quotas and restrictions on production or use.12 Finally information problems in the presence of heterogeneities across communities suggest the desirability of decentralization and empowerment of users of social services in order to increase

Click to return to Table of Contents J ERE R. BEHRMAN / JAMES C. KNOWLES 705 the efficacy of the provision of those services, although such considerations must be balanced against possible economies of scale, higher quality of staff, and possibly lower levels of corruption at more centralized levels, as well as intercommunity distributional concerns. Thus, for efficiency/productivity reasons, particularly given that information is imperfect and changes are frequent, there is an argument for choosing policies as high as possible in the policy hierarchy defined by the extent of marginal direct and distortionary costs—and thereby using interventions that are focused as directly on the problem as possible.13 In particular, this means that if there is a good efficiency/productivity reason for public support for a population or reproductive health investment, the best way to provide such support is not necessarily through governmental provision of the services. Higher in the policy hierarchy than direct governmental provision of such services, for example, may be subsidies or taxes that create incentives for the efficient provision of these services whether the providers are public, private, or some mixture. On the other hand, policies that discriminate against one type of provider—for example, by making the availability of such subsidies dependent on the provider being public—are generally likely to be lower in the policy hierarchy than policies that do not have such conditions. Now consider distribution. Generally speaking the subsidization of specific goods and services (and even less, the direct provision by government of goods and services at subsidized prices) is not an efficient way of lessening distributional problems. Because subsidies are designed to lower prices to consumers, they induce inefficient consumption behavior (i.e., consumption at a point to the right of H* in Figure 1). Instead, it generally is more efficient (and thus has lower costs in terms of alternative resource uses) to redistribute income to consumers, allowing them to allocate the income in ways that lead to efficient patterns of consumption.14 Of course, there are some cases in which subsidization of selected goods and services may be defensible to attain distributional objectives. For example, in cases where it is difficult (and therefore costly) to target the poor, subsidizing certain goods and services that are mainly consumed by the poor may be the most efficient policy alternative. A second example occurs when policymakers believe there is a serious problem of intrahousehold distribution (e.g., women and children are disadvantaged relative to adult males). In such circumstances, subsidies directed to goods and services consumed mainly by those who are viewed as disadvantaged within households may serve to improve their welfare more than cash payments directed to the household.15 Rather than being concerned with the command over resources of its poorer members, society, as noted above, may deem it desirable that everyone enjoy basic population, reproductive health, and other services.16 Such an objective might be obtained through many means. But presumably it is desirable to ensure that everyone has basic options for population

Click to return to Table of Contents 706 POPULATION AND REPRODUCTIVE HEALTH and reproductive health care at as little cost as possible in terms of productivity. Therefore, rather than ignoring efficiency considerations, it is desirable to choose policies as high as possible in the efficiency policy hierarchy and still ensure that the basic objectives are met. Efficiency considerations thus play an important role in interaction with the pursuit of distributional goals.

Project evaluation

Cost-benefit analysis and cost-effectiveness analysis have been widely used to evaluate governmental investments, including investments related to population and reproductive health.17 For example, cost-benefit analysis has been used extensively to evaluate family planning investments in developing countries since the 1960s, and variants of it are still being used.18 In the context of a market economy most economists presume that governmental intervention should be considered only if there are efficiency or distributional reasons for such policies and that particular policies should be adopted only if they are high in the policy hierarchy and have expected social marginal gains that exceed the social marginal costs (Folland, Goodman, and Stano 1993; Devarajan, Squire, and Suthiwart-Narueput 1997; Hammer 1997). This paradigm invokes principles of welfare economics to evaluate the need for governmental intervention in markets, including those delivering population and reproductive health services (Preston 1986; Behrman 1993). The application of cost-benefit analysis within this paradigm involves a redefinition of benefits to become the net social welfare gains (if any) resulting from governmental intervention in the form of either a project or a policy reform.19 Figure 4, which depicts the same situation as Figure 2, illustrates this idea. If a subsidy is used to lower the private cost to consumers sufficiently that they increase their investment (e.g., use of contraceptives) from H* to H**, the net welfare gain from such a policy would be represented by the area formed by the triangle ABC. In a cost-benefit analysis, this net welfare gain would constitute the benefits, while the cost of administering the policy change (i.e., the cost of administering the subsidy as well as any distortionary costs) would constitute the policy’s costs.20 Cost-effectiveness analysis has also been used widely to evaluate international population programs since at least the 1960s.21 Cost-effectiveness analysis was originally developed as an alternative to cost-benefit analysis for evaluating competing investments in sectors such as national defense in which the output is not exchanged in markets and hence cannot be easily “valued.” Cost-effectiveness analysis has also been used effectively as a managerial tool to evaluate alternative investments designed to achieve the same purpose.22 However, it has also been used recently to evaluate dissimilar competing investments;23 in the context of the present policy framework cost-effectiveness analysis used in this way has several limita-

Click to return to Table of Contents J ERE R. BEHRMAN / JAMES C. KNOWLES 707

FIGURE 4 Private marginal benefits and private marginal costs of human resource investments, with social marginal benefits and welfare gain (ABC) of moving from H* to H**

Marginal benefits Marginal costs

Private C marginal benefits

R** B R* A

Private marginal costs Social marginal benefits

H* H** Human resource investments tions (Hammer and Berman 1995). First, and perhaps most importantly, it fails to consider which services merit governmental interventions in the first instance.24 Second, the benefits are too narrowly defined (e.g., improved health) as compared to the much broader concept of “welfare” that underlies the framework in this section.25 Third, cost-effectiveness analysis completely neglects the demand side. If the reason for preparing cost-effectiveness analysis estimates is to determine which health services are wor- thy of public subsidization, price elasticities of demand effectively constrain the extent to which subsidies can increase levels of use (and therefore their ultimate potential to produce positive health effects).

Illustrations of the policy framework: Examples from Indonesia and Vietnam

In this section we illustrate the application of the policy framework presented above, using examples from two Asian countries. The first example, focusing on efficiency, is from the cost-benefit analysis of a recently devel-

Click to return to Table of Contents 708 POPULATION AND REPRODUCTIVE HEALTH oped safe motherhood project in Indonesia (World Bank 1997; Knowles 1996, 1997a). Benefits are estimated solely on the basis of the changes in economic welfare that result from the project’s interventions (i.e., changes in consumer surplus, risk reduction). Although the example refers to safe motherhood, similar analysis can be done for other reproductive health or family planning investments. The second example, focusing on distribution, is drawn from a recent “benefit incidence” analysis of Vietnam’s family planning program that estimates the share of governmental subsidies received by different income groups (Behrman and Knowles 1998a).

Cost-benefit analysis of an Indonesian safe motherhood project

Although Indonesia has for many years registered impressive gains in family planning use, it lags behind other countries in the ASEAN region (as well as other middle-income developing countries) in maternal health. Indonesia’s maternal mortality rate is estimated to be about 390 maternal deaths per 100,000 live births, and there is no evidence that it has declined much in recent years. The government of Indonesia committed itself as early as 1988 to reducing the maternal mortality rate by one-half by the year 2000. An accelerated midwife training program was initiated in 1993 to place a professionally trained midwife in every village of the country (World Bank 1997). By 1997, 55,000 village midwives had been trained and placed in villages (out of an estimated total of 60,000 villages nationwide). Despite improved access, many Indonesian women fail to use the services of a trained midwife.26 One explanation may be inadequate information within the population about the risks of their current consumption patterns (i.e., use of unqualified traditional birth attendants). Maternal mortality is a relatively rare event. At Indonesia’s current maternal mortality and fertility rates, a village of average size (3,000 inhabitants) experiences a maternal death only once on average every five years. Even if villagers are careful observers, and include maternal morbidity and perinatal mortality outcomes as well as maternal mortality in evaluating their choices, they are hindered by the rarity of observations in making choices other than the traditional ones (i.e., births at home attended by traditional birth attendants).27 The project addresses this information gap through a strong IEC (information, education, communication) component. However, even well-trained village midwives are not able to care for all obstetric deliveries. Approximately 10 percent of women need to deliver in a hospital, either as a precautionary measure or in response to problems encountered during the delivery itself (emergency cases). Most rural Indonesian women do not have health insurance, so the high cost of hospital deliveries (particularly those requiring surgical intervention) is a barrier to many seeking the care they need. The project addresses this prob-

Click to return to Table of Contents J ERE R. BEHRMAN / JAMES C. KNOWLES 709 lem by providing social insurance to pregnant women to cover the cost of emergency obstetric care. The cost-benefit analysis of these two components of the project pro- ceeds as follows: The IEC component. The potential net social benefit resulting from the IEC component consists of the additional consumer surplus that results from shifting the current demand curve outward so that it is equal to the social marginal benefits curve corresponding to the demand of fully informed consumers (as in Figure 5). It is possible to obtain at least order-of-magnitude estimates of these benefits (triangle ABC in Figure 5) by making the following assumptions: —The marginal cost of midwife obstetric delivery services is assumed to be constant. The main costs of professionally attended deliveries, most of which occur in the woman’s home, are the midwife’s time and a limited amount of consumable supplies. In the absence of this assumption, an estimate of the elasticity of supply would be required.

FIGURE 5 Welfare gains from the IEC component of the Indonesian safe motherhood project

Marginal benefits Marginal costs

Social marginal benefits (demand P1 curve of fully informed consumers) B Current demand curve (poorly informed consumers) Marginal costs P 2 AC

Human resource M1 M2 investments

Click to return to Table of Contents 710 POPULATION AND REPRODUCTIVE HEALTH

—The full-information demand for professionally attended births can be approximated by current patterns of use by women with at least some secondary education. —The price elasticity of demand for attended births is –0.3. Most estimates of the price elasticity of demand for ambulatory medical care, both internationally (Folland, Goodman, and Stano 1993; Phelps 1992) and among rural Indonesian adults (Gertler and Molyneaux 1997), are in the range of 0.0 to –0.4 (use of an elasticity closer to zero increases the magnitude of the estimated benefits). Under these assumptions, annual benefits at the end of the project are estimated to be approximately 889 rupiah (US$0.38) per capita (see Appendix for a detailed description of the estimation process). These estimated benefits should be compared to the cost of the project’s IEC component, but not to the cost of the additional health services consumed (since these are matched by an equivalent welfare value, i.e., rectangle M1ACM2 in Figure 5), to obtain a benefit-cost ratio. Emergency obstetric care insurance. The social financing of emergency obstetric care improves consumer welfare by allowing consumers to pool the risk of a major health expenditure. The welfare gains from risk pooling (i.e., the benefits) are greatest for risks such as the cost of emergency obstetric services that are relatively rare but involve large costs. Offsetting these benefits are costs in the form of: 1) welfare losses through increased consumption of services resulting from the price changes to an insured individual (in the health economics literature these losses are referred to, somewhat misleadingly, as “moral hazard”28 and are depicted in Figure 6 as the triangle CDE, i.e., the difference between the cost of the extra services consumed and the valuation consumers place on them); and 2) the administrative (or “loading”) costs of the insurance (Phelps 1992: 289–291). The following assumptions are used to facilitate the estimation of these benefits and costs: —Ten percent of women are assumed to be at risk of needing hospitalization for complications related to pregnancy or delivery, 70 percent of whom require nonsurgical interventions (at an average cost of 345,000 rupiah per admission) with the remaining 30 percent requiring surgery (at an average cost of 805,000 rupiah per admission).29 —The insurance provided involves a 20 percent coinsurance rate (i.e., the insurer reimburses the hospital for only 80 percent of costs, with the insured paying the remaining 20 percent). —The price elasticity of demand for emergency obstetric services is 0.1. Most international studies indicate that the demand for hospital services, especially emergency services, is less elastic than that for outpatient services and in the range of 0.0 to –0.2 (Phelps 1992; Folland, Goodman, and Stano 1993).

Click to return to Table of Contents J ERE R. BEHRMAN / JAMES C. KNOWLES 711

FIGURE 6 Welfare losses attributable to moral hazard

Marginal benefits Marginal costs

Social marginal benefits

CDMarginal costs P1

Copayment P 2 E

M M Human resource 1 2 investments —The income elasticity of the marginal utility of income is –0.4 (i.e., a one percent increase in consumer income is associated with a 0.4 percent decline in the marginal utility of income).30 Under these assumptions the benefits of emergency obstetric care insurance are estimated to be $4.58 per insured pregnancy, while the costs are estimated to be $2.48 per insured pregnancy, i.e., the sum of moral hazard costs of $0.67 and loading costs of $1.81 (see Appendix). The estimated annual benefit-cost ratio is therefore ($4.58/$2.48) = 1.85.

Distributional analysis of Vietnam’s family planning program

Although Vietnam has had a strong antinatalist population policy since 1963, only since 1993 has the government made substantial investments in family planning. The program recently has made available more contraceptive methods, has implemented a large IEC component,31 and has developed an administrative structure with paid staff down to the commune level and with thousands of volunteers at the village level (Mai Ky 1994). The total fertility rate has fallen from about 6.4 births per woman in 1960–

Click to return to Table of Contents 712 POPULATION AND REPRODUCTIVE HEALTH

64 to about 3.3 in 1989–93 (General Statistical Office 1991, 1995). Cur- rent demographic objectives include limiting couples to one or two children, achieving replacement-level fertility by the year 2015, and achieving a stable population by the year 2050. The government regards the family planning program as one of its most important development programs. Vietnam has an extensive, but poorly funded governmental health system. At the top of the system are a limited number of relatively well- funded central governmental hospitals operated by the Ministry of Health, almost all of which are located in urban areas. These facilities are generally acknowledged to provide the best quality health care. Although their in- patient services are in principle available to the entire population, their extensive outpatient services are practically accessible only to urban and suburban populations, and, because of the relatively high fees they charge, are mainly used by upper-income groups and senior governmental officials. Provinces operate a system of provincial and district hospitals, with the former being considerably larger and providing a broader range of higher quality services than the latter. Provinces also operate polyclinics (health centers staffed by several doctors) in a few areas having poor access to district hospitals. The lowest level of the governmental health system (and in fact not formally a part of it) consists of commune health centers operated and financed almost entirely by communes. They are typically staffed by one doctor or assistant doctor and one or two nurses and/or midwives. The government’s family planning program provides services through the existing health infrastructure, and in contrast to most other health services (including prenatal care, obstetric delivery care, induced abortion, and menstrual regulation) there is usually no charge for either family planning services or contraceptives. Table 1 provides estimates of the use of family planning services by per capita consumption quintile.32 These estimates indicate that current contraceptive use is 66 percent among the poorest women, 75–78 percent among women in the next three quintiles, and 70 percent among women in the richest quintile. Surprisingly, no systematic variation is observed between quintiles in rates of modern contraceptive use, which is 57 percent among the poorest women and 55 percent among the richest. Little variation is found as well in the method mix between women in different per capita consumption quintiles, although poorer women tend to rely more on the IUD and less on traditional methods. There is, however, substantial variation in the source of contraceptives. More than half of the poorest women (54 percent) obtained their contraceptives from commune health centers, compared to only 21 percent of women in the richest quintile. In contrast, about one-quarter (26 percent) of the rich obtained contraceptives from hospitals, compared to only 11 percent of poor women. Since the unit costs of family planning services (which are equal to unit subsi-

Click to return to Table of Contents J ERE R. BEHRMAN / JAMES C. KNOWLES 713

TABLE 1 Use of family planning and reproductive health services by consumption expenditure quintile, Vietnam 1996 Per capita consumption quintile All Poorest 2nd 3rd 4th Richest Vietnam

Percent of married women aged 15–49 currently using a contraceptive 65.9 78.0 74.6 76.0 69.5 72.7 Percent of married women of reproductive age using a modern contraceptive 57.3 63.8 59.1 57.2 55.0 58.0 Method currently used (percent) Pill 6.1 5.0 5.6 8.9 6.7 6.5 IUD 64.7 63.6 59.7 54.4 55.1 59.2 Condom 6.2 5.2 7.6 4.6 7.4 6.2 Male sterilization 1.4 0.0 0.4 0.0 0.3 0.4 Female sterilization 8.0 8.0 6.0 6.7 8.9 7.5 Other modern methods 0.5 0.0 0.0 0.7 0.7 0.4 Traditional methods 13.0 18.2 20.8 24.8 20.9 19.8 Total 100.0 100.0 100.0 100.0 100.0 100.0 Contraceptive source (percent) Central hospital 0.8 0.0 0.0 0.0 0.3 0.2 Provincial hospital 1.6 4.4 5.8 5.5 11.8 6.1 District hospital 8.9 11.0 10.9 5.9 14.0 10.2 Polyclinic family planning center 14.9 20.7 19.0 17.8 23.3 19.4 Commune health center 54.2 41.6 37.2 35.6 21.4 37.1 Privatea 19.6 22.3 27.1 35.1 29.3 27.0 Total 100.0 100.0 100.0 100.0 100.0 100.0 aIncludes traditional methods. SOURCE: Behrman and Knowles (1998a).

dies in this case) are much higher in hospitals, these data suggest that the rich may be capturing proportionately more family planning program subsidies than the poor. Since one of the reasons most often given for subsidizing these services is to help the poor, an important question is what share of the total subsidy is reaching its target. In the public economics literature, the share of governmental subsidies received by different income groups is referred to as “benefit incidence” (Jimenez 1995; van de Walle and Nead 1995; Pradhan 1996). Estimates of the benefit incidence of family planning services are presented in Table 2.33 In cases where the percentage of subsidies received by the poorest quintile exceeds their share of the population (20 percent)—as is not the case in Table 2—the subsidies are considered to be strongly pro-poor (or relatively efficient in targeting the poor). The Lorenz

Click to return to Table of Contents 714 POPULATION AND REPRODUCTIVE HEALTH

TABLE 2 Benefit incidence of Vietnam’s family planning program, by consumption expenditure quintile, 1996 Per capita consumption quintile All Poorest 2nd 3rd 4th Richest Vietnam

Married women of reproductive age as percent of total population 14.4 15.1 14.9 15.3 17.8 15.5 Per capita monthly consumption (000 VND, adjusted for regional price differences) 60.03 89.09 111.48 143.32 265.84 134.02 Annual service utilization rates per 1,000 population Central hospital 0.759 0.000 0.000 0.000 0.374 0.225 Provincial hospital 1.518 5.182 6.447 6.395 14.703 6.874 District hospital 8.446 12.956 12.116 6.861 17.444 11.494 Polyclinic 14.140 24.380 21.119 20.698 29.032 21.861 Commune health center 51.434 48.996 41.349 41.396 26.666 41.806 Private 18.600 26.265 30.123 40.814 36.508 30.425 Benefit incidence (annual family planning subsidies received per 1,000 population, VND) Central hospital 995 0 0 0 490 295 Provincial hospital 1,207 4,118 5,123 5,082 11,684 5,462 District hospital 4,544 6,971 6,519 3,691 9,385 6,184 Polyclinic 1,828 3,152 2,730 2,676 3,753 2,826 Commune health center 6,948 6,619 5,586 5,592 3,602 5,648 Private 82 116 133 180 161 134 Total benefit incidence 15,604 20,976 20,091 17,222 29,075 20,550 Targeting efficiency (percent of total family planning benefits received) 15.2 20.4 19.5 16.8 28.4 100.0 Percent of total population 20.0 20.0 20.0 20.0 20.0 100.0 Lorenz (cumulative) distribution of family planning benefits 15.2 35.6 55.1 71.9 100.0 — Lorenz distribution of per capita consumption 8.9 22.3 38.9 60.2 100.0 — Lorenz distribution of population 20.0 40.0 60.0 80.0 100.0 —

NOTE: VND = Vietnamese dong. SOURCE: Behrman and Knowles (1998a).

distribution of subsidies is the cumulative percentage of subsidies received by a given quintile as well as by all poorer quintiles. A completely equal Lorenz distribution would have the same cumulative distribution as the quintile populations (i.e., 20 percent, 40 percent, 60 percent, 80 percent, 100 percent).34 If the Lorenz distribution of subsidies is more equal than the corresponding distribution of total consumption per capita—as is the case in Table 2—the subsidies are considered to be weakly pro-poor (i.e., the subsidies in this case improve the overall distribution of consumption).

Click to return to Table of Contents J ERE R. BEHRMAN / JAMES C. KNOWLES 715

In Vietnam’s weakly pro-poor family planning program, the poorest per capita consumption quintile receives 15 percent of the total governmental family planning benefits (i.e., public subsidies). Although this is less than their share of the population (20 percent, by definition of a quintile), it is substantially more than their share of total private consumption (9 percent).35 In Table 2 the poorest two quintiles receive 36 percent of the total benefits, almost as much as their share of the population (40 percent) and substantially more than their share of total private consumption (22 percent). These results for the poorest 40 percent of the population are directly comparable to similar benefit incidence estimates for public health programs in other Asian countries (Pradhan 1996), for example, Malaysia in 1974 (47 percent), Philippines in 1975 (27 percent), Sri Lanka in 1978 (46 percent), and Indonesia in 1978 (19 percent) and to an estimate of the benefit incidence of curative care subsidies received by the poorest 40 percent of the population (28 percent) in Vietnam (World Bank 1995). In Vietnam, the main source of inequality in the distribution of family planning benefits received stems from the greater use of more heavily subsidized governmental providers (i.e., hospitals) by the higher income quintiles. As noted, for example, 26 percent of contraceptive users in the richest quintile obtain their contraceptives from governmental hospitals, compared to only 11 percent of those in the poorest quintile. In contrast only 21 percent of contraceptive users in the richest quintile obtain their services at a commune health center (the least expensive source), compared to 54 percent of those in the poorest quintile. In addition to differences in the source mix, differences in contraceptive prevalence (relatively small in Vietnam) and in the proportion of married women of reproductive age in the population (18 percent of the richest quintile, compared to 14 percent of the poorest quintile) also contribute to inequality in the distribution of benefits. Partly offsetting these factors is the observation that 29 percent of those in the richest quintile obtain their services from the lightly subsidized private sector (i.e., private-sector subsidies are limited to the government’s social marketing program). A policy of charging fees for the relatively high-cost family planning services provided by hospitals would be likely to improve the distribution of family planning benefits in Vietnam.36 At present, the government provides much higher subsidies (in absolute terms) to family planning services obtained from hospitals than it provides for services obtained from commune health centers, polyclinics, and family planning centers. Since the relatively well-off urban population lives close to hospitals, they are in a better position to (and do in fact) capture the bulk of these generous subsidies. If instead the subsidy per acceptor at hospitals were reduced to the level received by acceptors at commune health centers and family planning centers (i.e., about 130,000 VND per acceptor), the share of subsidies

Click to return to Table of Contents 716 POPULATION AND REPRODUCTIVE HEALTH received by the poorest two quintiles would increase from 36 to 41 percent (i.e., to more than their share of the population), while the share received by the richest quintile would decrease from 29 to 21 percent (i.e., program benefits would be distributed almost proportionately across quintiles). In addition, such a policy change would be likely to shift some rich acceptors away from more expensive sources of family planning services (since they would be receiving more appropriate price signals), simultaneously improving overall program efficiency.

Key policy issues in population and reproductive health

In this section we apply the policy framework developed earlier to a set of policy issues in the areas of population and reproductive health that have important economic dimensions. These include: pricing and cost recovery; access and quality; the role of the private sector; targets, incentives, and unmet need; decentralization; sustainability; and evaluation.

Pricing and cost recovery

Many family planning and reproductive health programs provide services free of charge to the entire population, and financing these programs has become burdensome in many developing countries. One response has been to consider charging fees for these services. Although charging fees has the potential to recover a significant share of costs, concern is often expressed that it will reduce prevalence levels and impose a burden on the poor. Other issues relate to how high fees should be (i.e., what percentage of costs should be recovered) and how fees should be set for different contraceptive methods. The issues raised with respect to family planning financing apply equally to reproductive health services. According to the policy framework, all health services (including family planning services), whether produced by the government or by the private sector, should be priced so that the social marginal benefit of the last unit consumed is equal to its social marginal cost. Unless a situation such as that depicted in Figures 2 or 3 exists, the normal operation of competitive markets should ensure that this occurs. If not, methods such as those illustrated in the Indonesian example should be used to find an appropriate level of subsidy (or tax). If a case can be made for subsidizing governmental services on efficiency grounds, the same subsidy should be extended to private providers. Failure to extend the same subsidies to the private sector as the public sector receives puts the private sector at a competitive disadvantage and risks losing the potential gains in efficiency that may be obtained by strengthen- ing competition between public- and private-sector producers.

Click to return to Table of Contents J ERE R. BEHRMAN / JAMES C. KNOWLES 717

The same principles should be used to determine subsidies, if needed, for individual contraceptive methods. In general, they imply that prices (and subsidies) for different methods should be roughly proportional to their marginal cost, so that the fee charged for a sterilization or implant should in most cases be significantly higher than that for an IUD insertion. For a given method, the same absolute subsidy should be given to each provider (including private providers). Actual pricing policy often departs from these principles, however, as the Vietnam example above illustrates; more expensive methods and services typically receive higher subsidies to make them affordable. In many programs, all services regardless of cost are provided free. Varying the subsidies from one method or source to another, in ways other than those dictated by applying the principles presented in the policy framework, creates the wrong incentives for consumers and encourages inefficient contraceptive choices.37 It can also exacerbate distributional inequities, as the Vietnam example illustrates. The level of subsidy should not depend on the perceived “cost-effectiveness” of the method unless society is sufficiently committed to the goal of reducing fertility so that it is prepared to define social marginal benefits very narrowly as a method’s effectiveness in averting births.38 Certainly, pricing population and reproductive health services according to efficiency principles may hurt the poor in comparison with free provision (just as competitive pricing of food, housing, and clothing may hurt the poor in comparison with free provision). They will be hurt whether or not they reduce their consumption of the service. At times it is suggested that they will be hurt more if they reduce their use of such services substantially because of price increases (e.g., Gertler, Locay, and Sanderson 1987). But in fact they will probably be hurt more by increased prices because of “cost recovery” applied to services that induce little or no reduction in use (i.e., demand is inelastic) than they are by price increases that result in large reductions of quantities demanded (i.e., for which there are close substitutes so demand is elastic). If the poor can readily substitute other goods and services (e.g., additional food purchases) for family planning services if the price of the latter increases, this may be of concern to the public health specialist and the family planning worker. But it may be of less concern to the poor themselves because they are able to substitute consumption of other goods and services with little loss of welfare (see Deaton 1997: 206–210 for a discussion of related issues). That the poor may reduce their use of family planning and reproductive health services because of price increases as part of cost recovery efforts, however, does not imply that continuing to subsidize all such services necessarily helps the poor. If certain services have a high income elasticity of demand, for example, most of the subsidy will be captured by upper-income groups, as in the case of hospital-based subsidies noted earlier.

Click to return to Table of Contents 718 POPULATION AND REPRODUCTIVE HEALTH

In such cases, the poor would be better off if the funds were used to subsidize the goods and services of which they consume proportionately more.

Access and quality

The family planning and reproductive health community frequently identifies “access” and “quality” as program objectives in addition to efficiency, distribution (or equity), and “sustainability” (discussed below). “Access” is usually defined as the absence of economic and physical barriers that keep a prospective client away from a service delivery point; “quality” is usually defined as the factors that determine whether the care is effective and acceptable to the client once she or he walks through the door. Particularly in the family planning literature, optimal levels of access and quality are defined as those that maximize use and impact on fertility. Thus, for example, contraceptives are delivered free of charge to a client’s doorstep (taking access to the limits), and governmental family planning programs are criticized for their poor quality if they do not provide a full range of contraceptive methods (the “cafeteria approach”), citing evidence that continuing use increases with the number of methods available. Although these concepts are undoubtedly useful to policymakers within the population and reproductive health communities, they are confusing from the perspective of our economic framework because “access” and “quality” considerations are included in the concept of efficiency. Stan- dard economic principles support improved access (e.g., bringing services closer to clients) up to the point where the social marginal benefits of doing so are equal to the social marginal cost. Goods and services of varying qualities are effectively distinct goods and services, and the normal functioning of markets (assuming consumers are well informed) provides an optimal mix of goods and services of different qualities (i.e., the mix that responds to consumer demand). Unless a situation such as that depicted in Figure 2 or 3 is present, the normal operation of competitive markets should result in optimal “access” and “quality.” If this is not the case, the methods illustrated in the Indonesian example can be used to determine optimal levels of subsidies (or taxes) that should then be applied to all producers, public and private. If consumers cannot be relied upon to make “good” quality choices in the market for health services, this is a problem of lack of information and should be addressed as such. If governments have an effective monopoly in the provision of certain services, which can easily occur if they are financed mainly out of governmental budgets, problems of poor access and quality often arise. De- cisions about the location of governmental health facilities are often made on the basis of political considerations, rather than efficiency and distribution. Governmental providers have little incentive to offer good quality services; and they tend to provide a more limited range of service quality than

Click to return to Table of Contents J ERE R. BEHRMAN / JAMES C. KNOWLES 719 do private providers. The poor quality of governmental services often is a consequence of a lack of competition from private providers. Policies that subsidize governmental providers, but not private providers, are the source of the problem. The solution is to eliminate subsidies to governmental providers (i.e., charging fees that reflect full marginal costs) or, if justified on either efficiency or distributional grounds, provide subsidies equally to all providers. Special quality improvement programs directed to governmental providers are not nearly as high in the policy hierarchy as solutions to problems of poor-quality governmental services.

The role of the private sector

The policy framework presented in this article suggests that most goods and services should normally be provided by an approximately competitive private (or mixed public and private) sector with numerous suppliers. Because of the difficulties in assuring the right incentives for governmental providers, direct governmental provision of services is unlikely to be high in the policy hierarchy except under unusual circumstances. In markets where social marginal benefits exceed private marginal benefits, the preferred solution is to offer subsidies to private providers (or to consumers) rather than to have governments provide services directly. The main exception to this rule occurs when the relevant market is too small to support more than one seller of a good or service that is not easily imported from other regions or from abroad, so that a monopoly is likely to result. Even in this case, however, regulated private monopolies are likely to be preferred to a governmental producer because of the problems in creating incentives for efficient behavior for governmental producers, problems that are exacerbated by limited information. If governments are to provide services efficiently, in general they must be encouraged to compete on a level playing field with the private sector. There is a common perception that the private sector is more efficient than the public sector. The profit motive is credited with providing private managers with the incentive to be more efficient. Governmental providers may be efficient (and there are numerous examples of efficient governmental providers of goods and services), but when they are it is usually because they function like private providers in a competitive market environment. Monopolistic governmental agencies financed out of a central budget are rarely efficient and are often highly inefficient. Their payrolls are usually swollen with redundant, poorly trained and equipped personnel, often with ties to the political leadership. As noted above, they are often incapable of providing the kinds and variety of services that consumers demand (i.e., services of the “right” quality). These are some of the reasons that governments around the world are rapidly “privatizing” often enormously inefficient government-owned and -operated enterprises in a

Click to return to Table of Contents 720 POPULATION AND REPRODUCTIVE HEALTH wide range of industries. It is because of the typically poor performance of many governmental providers of social services that many economists favor the use of demand-side subsidies (e.g., voucher schemes) to consumers that “empower” them to choose freely between public and private providers. In contrast, the widespread view within many family planning and reproductive health programs is that the government’s mission is to serve the poor while those able to pay for services should be served by the private sector. This view mistakenly confuses a perceived need to finance services for the poor, which may be an appropriate mission of governments on distributional grounds, with the need to produce the services themselves, which should not normally be the mission of governments. It is sometimes argued that governmental providers alone are located in the remote rural areas where most of the poor live. One reason for this, however, is that governmental subsidies (i.e., selective supply-side subsidies) are only provided to governmental providers. If subsidies were instead provided directly to consumers (i.e., demand-side subsidies in the form of vouchers), so that private providers would be able to compete with governmental providers, a more balanced mix of public and private providers might be found even in relatively remote locations. The resulting competition, by promoting greater efficiency, would benefit the predominantly poor populations of such communities. In summary, the implication of the economic framework is for policy to be neutral regardless of ownership of producers of goods and services. If the private sector produces the same (or better) services at lower cost (i.e., through its greater efficiency) than governmental producers, then private suppliers should dominate for all income groups, including the poor. The role of the private sector should not be limited to saving the government’s money by attracting rich and middle-income clients away from highly subsidized government-provided services. On the contrary, if subsidies are provided to consumers to improve efficiency (e.g., because social marginal benefits exceed private marginal benefits) the private sector should receive the same subsidies.39 Higher up in the policy hierarchy, the government can reduce the use of scarce resources by ceasing to provide subsidized services to groups capable of paying for them through improved targeting.40

Targets, incentives, and unmet need

Many population programs in Asia have aggressively pursued demographic targets in an effort to reduce their fertility rates as quickly as possible. Such target-setting often reflects a strong belief on the part of policymakers that reducing population growth would significantly increase the pace of the country’s development. In several countries, women and men have been offered incentive payments to be sterilized, and women in some cases have

Click to return to Table of Contents J ERE R. BEHRMAN / JAMES C. KNOWLES 721 been offered payments to have an IUD inserted (Ross and Isaacs 1988).41 In terms of the policy framework, these governments apparently perceive that the social marginal benefit curve lies so far to the right of the private marginal benefit curve (compare to Figure 2) that using subsidies to reduce the price to zero would not be sufficient to achieve the level of contraception at which the social marginal benefit is equal to the social marginal cost (i.e., to induce use at H** implies a negative price). Although some countries still pay incentives to contraceptive acceptors (Vietnam is an example), the practice has not spread much in recent years. Family planning supporters generally frown on the use of incentives for several reasons. First, they are difficult to administer (because they involve cash payments), and the task of financing such payments increases the burden on limited governmental budgets. Second, incentives are often viewed as coercive because the poor find it difficult to refuse such cash payments, so that the payment of incentives effectively infringes on their reproductive rights. Since most of those expressing such concerns favor subsidizing family planning and reproductive health services, the position seems to be that subsidies are acceptable as long as the price of services does not drop below zero, at which point they become coercive.42 Although it is granted that both subsidies and incentives may affect the behavior of the poor more than the rich, most subsidies would be coercive according to such a definition (Mason 1994). For example, even providing modest subsidies to family planning and reproductive health services instead of using the same funds to subsidize food or other types of health services would be coercive because it is more likely to affect the choices made by the poor than by the rich. Taking a strong position against the payment of incentives prevents their possible use in ways that might promote reproductive rights, such as using cash payments to effectively “buy off” men’s opposition to women’s expressed desires to use contraceptives in settings where other measures (e.g., improved information to men on the potential economic and health benefits of child spacing) may prove to be ineffective. At the 1994 Cairo International Conference on Population and De- velopment the international population community rejected the use of demographic targets in population programs. Instead, programs were encouraged to reorient themselves to the goal of satisfying “unmet need” (Cohen and Richards 1994). Although conceptually this reorientation moves population policy away from its more narrow focus on demographic targets to one closer in spirit to “welfare” or “utility” (i.e., programs are designed to meet the “needs” of their clients), the concept of unmet need is sufficiently different from the concept of demand that it cannot be integrated easily into the policy framework presented above (and is accordingly difficult to use as a basis for project evaluation of the type illustrated in our Indone- sian example).43 Although the concept of unmet need is relevant in establishing that a need exists (at least technically, if not in the mind or behav-

Click to return to Table of Contents 722 POPULATION AND REPRODUCTIVE HEALTH ior of the survey respondent) and may be useful in mobilizing political and financial support for family planning programs from some constituencies, it is unlikely to be convincing to economists in ministries of finance or in international agencies, who are likely to question why this particular unmet need should be funded by the government (as compared to unmet needs for items such as food, clothing, fuel, or housing). The policy framework is potentially far more useful as a basis for convincing such economists to support funding for population and reproductive health programs.

Decentralization

Health and population programs are being decentralized in many countries. At one extreme, such decentralization may be limited to the administration of otherwise centrally directed governmental programs. For example, a provincial department may be authorized to incur expenses for some items against a centrally determined budget. More substantial, but still limited, forms of decentralization may permit local authorities to develop their own budgets (usually subject to central approval). In some de- centralizations, local administrations (and even individual health facilities) may also be permitted to establish and collect fees for health services and be permitted to retain most of the revenue for uses they determine. Typi- cally one of the last powers to be devolved from central to local authorities is the power to hire, fire, and determine the compensation levels of personnel; this function is usually retained by central or regional governments. Full decentralization would also include the power to supplement central revenues with locally levied and collected taxes, so that localities can effectively decide how much and on what to spend. Viewed from an economic perspective (as distinct from a political or managerial perspective), decentralization has potentially important implications for efficiency and distribution. In terms of efficiency, arguments have been advanced to support the idea that decentralization leads to improved efficiency and effectiveness in the provision of governmental services (Tanzi 1995). First, different geographical areas may have distinct preferences for governmental services. If the level and composition of services are determined locally, consumers may be able to attain higher levels of welfare using fewer resources. Second, with decentralization, one locality may compete with other localities to provide the best public services for given taxes. Third, decentralization may encourage experimentation with alternative approaches to providing public services. Fourth, decentralization can provide better accountability if local officials are directly responsible for the provision of locally consumed services.44 Against these potential efficiency advantages of decentralization, a number of counter-arguments advanced in the economic literature may

Click to return to Table of Contents J ERE R. BEHRMAN / JAMES C. KNOWLES 723 be relevant to the decentralized provision of population and reproductive health services. First, central governments should be able to obtain sufficient information about local preferences to tailor services appropriately (although the costs in doing so may be considerable). Second, the argument that decentralization provides a better means of registering local preferences assumes that local governments are responsive to preferences of constituents—that sometimes is translated to mean that they are demo- cratically elected.45 Third, corruption may be greater with decentralization if, as has been suggested, local bureaucrats tend to be less scrupulous than central bureaucrats (Prud’homme 1995; Tanzi 1995). Fourth, the quality of local bureaucrats, and consequently their effectiveness as managers, may be lower than that of central bureaucrats. Fifth, on the revenue side, some types of local taxes may be relatively expensive to collect and may distort economic activity more than taxes collected at the central level. In terms of distribution, it is frequently argued that decentralization is likely to worsen regional disparities (Prud’homme 1995). On the expenditure side, under conditions in which there are wide differences in average income levels among localities, decentralization can reduce the capacity of the central government to redistribute resources from rich localities to poor ones. On the revenue side, some types of local taxes may be inequitable in their incidence. At a minimum, decentralization may pose major challenges for population and reproductive health programs. Local governments may not view social benefits and costs in the same way as central governments (and donors), and their views will eventually be expressed in the allocation of resources at the local level. Central governments may need to resist such changes by developing matching grant programs that attempt to influence local resource allocations; but it is often difficult to monitor compliance with the conditions of such grants. A more common solution is to provide some types of services (e.g., those where social benefits and costs differ significantly from private benefits and costs) through centrally funded and directed vertical programs (e.g., Vietnam’s family planning program).46 De- centralization also requires an effective formula for revenue sharing between the central government and local governments to preserve distributional objectives.

Sustainability

The concept of sustainability concerns the capacity of a donor-funded project to continue producing the benefits it is expected to provide (e.g., those that were projected in its cost-benefit analysis) beyond the period for which financing and other inputs are provided by a donor. For example, if the project provides funds for a capital investment, will the beneficiary agency

Click to return to Table of Contents 724 POPULATION AND REPRODUCTIVE HEALTH

(usually a government or nongovernmental organization) be able to mo- bilize the recurrent expenditure to operate the capital investment when project support terminates? Although the primary focus of sustainability is financial, the concept has been broadened to consider the beneficiary’s institutional and managerial capacity to continue generating projected benefits of an investment beyond the life of an externally funded project. It is widely accepted that projects that in their initial stage provide support also for recurrent expenditures are more difficult to sustain than those that provide support only for capital investments. Sustainability is a major problem for many population projects, and concern is often expressed also for the sustainability of family planning and reproductive health programs. One reason for such concerns is that donor projects in these areas often provide support for recurrent expenditure in the form of donated contraceptives and other supplies. Some projects also supplement staff salaries and administrative costs, particularly in the case of projects involving NGOs. A desire to provide services to clients as quickly and reliably as possible is often the reason for funding a broader range of inputs. It is also true that in the early stages of governmental programs, domestic support is often weak and there is no presumption that the necessary recurrent expenditure would be forthcoming in the absence of donor funding. Given concerns for the sustainability of family planning and reproductive health programs, some donors have funded entire projects designed to promote sustainability. The problem is perceived to be particularly acute in the case of NGOs, which tend to receive a very high share of their recurrent budgets from donor funding and have only limited opportunities for obtaining alternative sources of funding. Cost recovery is often advocated or strengthened as one way to replace donor funding. The problem is that as fees rise, the characteristics of the clients served tend to change, with the result that a smaller share of the remaining subsidies reaches the poor (as can be examined through benefit incidence studies, such as that presented above on Vietnam). “Cross-subsidizing” is another common strategy frequently supported by sustainability projects. As an example, an NGO may establish a clinic in a higher income location and charge fees that more than cover its costs, using the surplus to support its operations in low-income areas. Cross-subsidizing activities, although they are sometimes generously funded by donors, rarely succeed (Janowitz and Gould 1993).47 In the context of the policy framework, however, sustainability may be a less serious problem. To begin with, governments should not in most cases provide family planning and reproductive health services directly.48 As argued above, the ultimate objective should not be to have heavily subsidized governmental providers serving the poor while a self-sufficient private sector serves the rich and middle-income groups. If governmental financing is justified on either efficiency or distribution grounds, the preference should be for demand-side subsidies (e.g., vouchers) provided

Click to return to Table of Contents J ERE R. BEHRMAN / JAMES C. KNOWLES 725 directly to target groups of consumers (e.g., the poor). In addition to fos- tering competition and therefore efficiency, the use of targeted demand- side subsidies also helps to resolve the problem of sustainability. In the case of NGOs, for example, those that are able to control their costs and compete effectively for consumers’ business (including the business of those consumers who are equipped with vouchers) will survive; the rest will fail. Governmental programs probably will (and probably should) disappear if they show themselves less able to compete with commercial and NGO providers. The issue of sustainability also raises questions about the long-run role, if any, for donors in the area of population and reproductive health. It is easy to justify a role for donors when situations such as those depicted in Figures 2 and 3 are present but where governments either refuse to intervene because of perceived political, religious, or cultural constraints or are unable to intervene because of financial constraints. There is a danger in such cases, however, that governments will become dependent on continuing donor funding, with the necessary governmental resources never becoming available. In such cases it is important to establish realistic targets for governmental funding, with the stipulation that funds be made equally available to private providers (so that one group of providers does not become dependent on donor resources), ideally through the use of demand-side subsidies. It is also important for donors to continually reassess the need for subsidies in the context of the policy framework. If, for example, the original justification for subsidies was to improve the flow of information to consumers, subsidies can be reduced or eliminated when consumers demonstrate an acceptable knowledge of the services.49

Evaluation

The careful evaluation of governmental policies in population and reproductive health and other human resources can be one of the most cost- effective uses of scarce resources. Nevertheless, a major problem from the perspective of our framework is that much of the existing evaluation of family planning and reproductive health service impact focuses exclusively on changes in fertility and health status. This is consistent with the traditional emphasis placed by programs on narrowly defined goals, as opposed to improving individual and social welfare more broadly conceived. Fu- ture evaluations of population and reproductive health programs should focus not only on fertility and health impacts but also on whether resources are allocated in such a way as to improve efficiency and equity. For example, if the rationale for governmental funding of population programs is that private markets do not produce enough information about family planning options, evaluations should focus on how much program funding is allocated to effective information-increasing activities and whether improvements in consumer knowledge are registered. Whether this leads to reduc-

Click to return to Table of Contents 726 POPULATION AND REPRODUCTIVE HEALTH tions in fertility may not be important from a broader welfare perspective, as long as consumers are making informed decisions. If individuals act in their own best interests, what is important is that they face appropriate private incentives to make fertility and other decisions that reflect true marginal social costs and benefits. If, as another example, the primary rationale for subsidizing family planning and reproductive health services is to ensure that the poor have access to them, evaluations should focus on the share of program subsidies actually received by the poor (as illustrated in the Vietnam example). In the past, such evaluations have been conducted only in rare instances. However, they are likely to be the kinds of evaluations that are persuasive to economists in ministries of finance and in international organizations with whom advocates of family planning and reproductive health are increasingly likely to have to negotiate over resource availability.

Summary and conclusions

Our main objective in this article has been to provide a standard economic framework to evaluate policies in the population and reproductive health field and to illustrate its use. We hope that this will facilitate cross-disciplinary exchanges between economists and others working in the population and reproductive health fields and will enable those working in population and reproductive health to be more effective in communicating their analytical insights and concerns to those outside the field with a primary background in economics. The standard economic framework presented justifies governmental policy interventions in population and reproductive health (and in other areas) in terms of two distinct objectives: 1) to increase efficiency and productivity; and 2) to redistribute resources. We have suggested that it is both practical and informative to use this framework to evaluate projects and policies. The article leads to a number of policy-relevant conclusions, which are summarized below. First, it is desirable to analyze the efficiency gains of projects and policy changes in the population and reproductive health field using a standard welfare economics framework in the context of cost-benefit analysis. Use of such a framework is consistent with current practice in other areas of human resource economics and is preferable to the use of traditional cost- benefit and cost-effectiveness methodologies that emphasize either narrower objectives (e.g., per capita income growth, governmental financial savings) or more diffuse ones (e.g., poverty reduction, environmental improvement, human capital formation, health, empowerment of women). The article has illustrated how this can be done using as an example the cost-benefit analysis of a safe motherhood project in Indonesia. Second, more attention should be given to the evaluation of distributional objectives in family planning and reproductive health programs. Im-

Click to return to Table of Contents J ERE R. BEHRMAN / JAMES C. KNOWLES 727 proving access of the poor to such services is often a stated justification for these programs, yet little attention has been given to evaluating the degree to which the poor benefit from these programs. The article has illustrated how such distributional assessments can be done, using a “benefit incidence” study of Vietnam’s family planning program. Third, application of the policy framework to a number of important resource- and finance-related issues in the population and reproductive health field (pricing and cost recovery, access and quality, role of the private sector, use of targets and incentives, decentralization, sustainability, and evaluation) suggests that a significant program bias favors publicly provided services and hinders the emergence of a more efficient mix of private and public providers competing on an equal basis. The article argues that many of the problems that economists and others working in these fields confront could be lessened or resolved by shifting from policies that emphasize continuing supply-side subsidies to public providers through governmental budgets to policies that make greater use of targeted demand- side subsidies to consumers (e.g., vouchers) that would empower them to choose between competing governmental, NGO, and commercial providers.

Appendix: Cost-benefit analysis of an Indonesian safe motherhood project

1. IEC Component. If the marginal cost of professionally assisted obstetric deliveries is constant (a reasonable assumption, since the main variable costs are the time of the village midwife and consumable supplies), welfare gains (triangle ABC in Figure 5) can be approximated by the following formula (Phelps 1992: 291):

1 Welfare gains = ( /2) (P1 – P2) (M2 – M1), (1) where M1 in Figure 5 refers to the equilibrium level of use under conditions in which consumers are not fully informed, and M2 refers to the equilibrium level of demand under conditions of full information. P2 refers to the equilibrium price under conditions of full information (equal to marginal cost in this case), and P1 refers to the price at which fully informed consumers would demand M1 services. For small changes in price (dP) and service use (dM), and using an estimate of the price elasticity of demand

(e), the term (P1 – P2) in equation (1) can be approximated by: dP = (dM/M) P (1/e). (2)

In a district of one million people in which the crude birth rate is 20 per 1,000 population, there would be approximately 20,000 births annu-

Click to return to Table of Contents 728 POPULATION AND REPRODUCTIVE HEALTH ally. According to data from the 1994 Demographic and Health Survey, the proportion of births in East and Central Java (the project site) that are attended by a trained provider (a doctor or nurse/midwife) are 35 and 27 percent respectively. However, the proportion of births attended by a trained provider among those with some secondary (or higher) education is 72 percent (also reported to be 71 percent in Bali and 70 percent in West Sumatra). If the proportion of births attended by a trained provider increases from 30 to 70 percent as the result of information provided by the project, the number of professionally attended deliveries would increase from about 6,000 annually to about 14,000 annually (i.e., dM = 8,000). Assuming that the price elasticity of demand for such services is –0.3 and that village midwives charge 50,000 rupiah to attend a delivery,50 the annual welfare gains for a single district of one million people can be estimated approximately as:

Welfare gains = 0.5 [ (8,000/6,000) 50,000 (1/0.3) ] 8,000 = 889 million rupiah = $378,573 per million population ($0.38 per capita)51

2. Emergency obstetric care insurance. Assuming 10 percent of pregnant women have a risk of being hospitalized for complications related to pregnancy or delivery, that 7 percent of these will require nonsurgical interventions during three days of hospitalization at a cost to the woman of $50 per day, or a total of $150 (345,000 rupiah), and that the remaining 3 percent will require surgery (e.g., cesarian section) and will be charged $350 (805,000 rupiah) for seven days of hospitalization at $50 per day, a pregnant woman has an expected uninsured loss (risk) of $21 with a variance of $4,809.52 The project is assumed to provide insurance to pregnant women to cover these risks with a 20 percent copayment (i.e., the insurer reimburses only 80 percent of the referral hospital costs, or $16.80, with the woman or her family being responsible for paying the remaining 20 percent, or $4.10). However, because such an insurance policy would change the effective price of hospital care to the woman (i.e., from $50 per day to $10 per day), there is likely to be some increase in the use of hospital services (e.g., in the average length of stay). If the price elasticity of demand for hospital services is –0.1 (fairly inelastic), the average length of stay would increase by 8 percent (e.g., from 3 to 3.24 days, or from 7 to 7.56 days, on average). The expected loss (risk) to the insured person in this example becomes $4.54 (i.e., increases by 8 percent, from $4.20 to $4.54) and the variance of the risk becomes $224 (i.e., insurance reduces the expected risk by $16.46 and its variance by $4,585). The welfare gains from insurance (i.e., the risk premium) can be estimated using the following formula (Phelps 1992: 286):

Click to return to Table of Contents J ERE R. BEHRMAN / JAMES C. KNOWLES 729

Welfare gains from insurance = 0.5*r(I)*(reduction in the variance of the risk), where r(I) is a measure of the degree of the average consumer’s risk aversion.53 Assuming that the poor rural consumers who would benefit from this insurance program have an average annual income of $200, a value of r(I) = 0.002 would correspond to an income elasticity of the marginal utility of income of 0.4. Under these assumptions, the welfare gains from insurance are estimated to be $4.58 per insured person. There are two components to the cost of providing such insurance. The first is the welfare loss from the 8 percent increase in service use that is attributed to moral hazard (i.e., triangle CDE in Figure 6). This can be estimated using the formula in equations (1) and (2) above. In this case, the consumer’s expected loss attributed to moral hazard is approximately $0.67. The second component of cost is the administrative, or “loading,” cost of providing the insurance. If loading costs are assumed to be 10 percent of the insurer’s paid benefit ($18.14 in the present case), they would be $1.81 per insured person. Total cost, therefore, is estimated to be $2.48; and the benefit-cost ratio is ($4.58/$2.48) = 1.85.

Notes

The original version of this article was prepared other individuals in ways other than those for the Program in Population Sciences of the that are reflected in market prices, economists Rockefeller Foundation. Behrman and say that the investment generates “externali- Knowles alone, and not the Rockefeller Foun- ties”—i.e., effects that are external to what is dation, have full responsibility for the contents transmitted through markets. If at least some of the article. Correspondence should be sent individuals do not fully consider the exter- to Jere R. Behrman, Economics, McNeil 160, nal effects of their own decisions on the wel- 3718 Locust Walk, University of Pennsylva- fare of others, a purely private market solu- nia, Philadelphia, PA 19104-6297; telephone tion leads to underinvestment from society’s 215 898 7704, fax 215 898 2124, e-mail standpoint in activities that generate positive [email protected]. externalities (e.g., immunizations) and an 1 Note that economic efficiency is not overinvestment in the case of negative exter- the same as engineering efficiency because nalities (e.g., use of antibiotics). Such “exter- of the incorporation of marginal benefits and nalities” are one case of a broader category of marginal costs rather than an exclusive fo- conditions that economists call “market fail- cus on technological efficiency. ures” (other examples are discussed below). 2 The social marginal benefits also could 4 A necessary condition for efficient out- be lower than the private marginal benefits comes with purely private markets is that so that the social marginal benefits curve is consumers have complete information, the below the private marginal benefits curve, absence of which is considered to be another and policies to attain efficiency would have example of “market failure.” to reduce the private incentives to the so- 5 In purely competitive markets (i.e., cial levels. many producers of a uniform product), in- 3 When investments in one individual’s dividual producers have no incentive to pro- health affect the productivity and welfare of vide information to consumers because the

Click to return to Table of Contents 730 POPULATION AND REPRODUCTIVE HEALTH benefits of such information provision accrue tities to try to obtain these rents (“rent-seek- to all producers equally. In noncompetitive ing”). Rent-seeking uses resources and trans- markets, there is some incentive for indi- fers income among various groups, but does vidual producers to provide information to not add to production. consumers, but (except in the case of pure 10 Below, we present an example of monopoly) the incentive exists mainly to at- distortionary costs arising in connection with tract consumers to a particular producer (i.e., the provision of emergency obstetric care in- to promote brand loyalty), rather than to surance in Indonesia. provide generic information bearing on the 11 If the policies involve public expen- issue of whether and how much of the good ditures, as most do, it is important to con- or service a consumer should purchase from sider the cost of raising the necessary tax rev- a social perspective. Under these circum- enue to finance the policy. In the United stances, one might expect that a private mar- States, for example, it has been estimated ket providing information directly to con- that the distortionary cost (often called the sumers would fill the gap. However, the “deadweight loss”) of raising a dollar of tax “public good” nature of information (i.e., revenue ranges from $0.17 to $1.00, depend- that the marginal cost of providing informa- ing on the type of tax involved (e.g., Feld- tion to another consumer is virtually zero) stein 1995). leads to underproduction of information from a social point of view by private mar- 12 Nevertheless there are likely to be kets because private providers cannot cover some cases, such as providing information their costs if they price information at the on the quality of goods and services related social marginal cost as required for efficiency. to population and reproductive health investments, for which quantitative regula- 6 These are major reasons why a diver- tions may be higher in the policy hierarchy gence may arise between private and social than price policies because of the nature of marginal benefits for investments in popu- the information requirements. lation and reproductive health, although there may be other factors as well (e.g., the 13 Below, we provide an example of a social discount rate may be lower than the fertility control policy that is not well focused private discount rate, wage and price rigidi- (viz., charging fees for obstetric delivery care ties may preclude wages and prices from re- only for third and subsequent births) and flecting social marginal benefits and costs, that has negative distributional and health income taxes may cause private marginal re- effects. turns to human capital investments to be 14 Even redistributing income may lead lower than social marginal returns). to inefficiency because it can affect the work 7 The social marginal costs also could be effort of those on both the tax-paying and higher than the private marginal costs, in tax-receiving sides. which case policies to attain efficiency would 15 However, household decisionmakers have to reduce the private incentives to the may reallocate resources, so that the in- social levels. tended recipients of such policies receive 8 These considerations are part of what much smaller benefits than intended (e.g., economists refer to as the “theory of the sec- children receiving subsidized food at school ond best.” For example, eliminating mo- may be fed less at home). nopoly pricing that restricts output in an in- 16 In the context of such an objective it is dustry that is a heavy polluter would be important for governments to decide whether unlikely to improve efficiency. success is to be measured in outcomes (health 9 Economic “rents” are returns to fixed status), utilization (health clinic use), or access supplies. Because the supplies are fixed, the (ability to consume a package of basic services rents gained by their owners do not affect regardless of income level). the quantities supplied. If supplies are fixed, 17 For an introduction to and an exten- for example, by regulations, rents are cre- sive overview of both methods, see Warner ated by the supply restrictions, so there are and Luce (1982), Sirageldin, Salkever, and Os- incentives for various private and public en- born (1983), and Evans and Hurley (1995).

Click to return to Table of Contents J ERE R. BEHRMAN / JAMES C. KNOWLES 731

For a recent review of the use of both meth- voted to population and reproductive health odologies in family planning, see Knowles interventions, but that once this share has (1997b). been determined (or if it is politically moot), 18 See for example Coale and Hoover cost-effectiveness analysis becomes the ap- (1958), Enke (1966, 1971), Robinson and propriate technique for choosing among in- Horlacher (1969), Simon (1969, 1977), dividual projects and program activities. A Zaidan (1969, 1971), Sommers and Suits problem with such a policy strategy is that (1971), Suits et al. (1975), Sommers (1980), it fails to consider on a case-by-case basis the and Chao and Allen (1984). In contrast, rationale for intervening in private markets. there have been few applications of cost-ben- 25 For example, in the case of family efit analysis to reproductive health, with the planning services, if cost-effectiveness ratios exception of analyses focused on AIDS (e.g., use a health measure in the denominator (e.g., Bloom and Lyons 1993, Bloom and Mahal disability-adjusted life years, or DALYs), they 1997, Cuddington 1993, Cuddington and completely neglect the non-health-related Hancock 1994, Kambou, Devarajan, and benefits to couples from preventing or de- Over 1992). laying births. If, instead, a measure of births 19 Such a welfare economics approach averted is used in the denominator, health to defining the benefits of family planning effects are ignored, such as when (as often projects was proposed over two decades ago has occurred in such analyses) condoms are by Haveman (1976), but it has not been found not to be cost-effective because their widely practiced in health and population role in reducing sexually transmitted diseases project appraisal and evaluation until re- is ignored or when the sterilization of older cently. (relatively infecund) women is found not to be cost-effective because its role in prevent- 20 The costs should include both the di- ing maternal and neonatal mortality and rect costs of the subsidy program (e.g., the morbidity is ignored. administrative costs of a voucher scheme) and the distortionary costs involved in fi- 26 Data from the 1993 Indonesia Fam- nancing the subsidy by raising additional tax ily Life Survey (IFLS) show that midwives revenue (as noted above, these costs are fre- working in both the public and private sec- quently estimated to be large relative to the tors are the modern maternal health care pro- actual tax revenue raised). The costs should viders of choice for most Indonesian women not include the subsidy itself (which is a (Serrato and Melnick 1995). However, the transfer payment) or the cost of the family IFLS shows that Indonesian women also planning services themselves (which are al- have a strong preference for giving birth at ready reflected in the social marginal cost home (84 percent of rural births and 44 per- curve—i.e., equal to the pre-subsidy private cent of urban births); and traditional birth marginal cost curve). attendants are the most common providers attending home births (74 percent of home 21 Even Enke’s pathbreaking cost-ben- births, compared to 19 percent of home efit analysis (1966) devoted more space to births attended by midwives). an extensive cost-effectiveness analysis of alternative contraceptive methods. 27 Data on patterns of use of safe motherhood services from the 1993 IFLS and 1994 22 Recent examples include Simmons, Indonesia Demographic and Health Survey Balk, and Faiz (1991), Vernon et al. (1994), show that educated women exhibit very dif- Foreit, de Castro, and Duarte Franco (1989), ferent patterns of use from those of unedu- Coeytaux et al. (1989), Foreit et al. (1993). cated women. For example, data from the 23 As in applications of the Global Bur- 1993 IFLS indicate that 85 percent of women den of Disease methodology (World Bank without any education had their births at- 1993a; Cochrane and Sai 1993; Walsh et al. tended by a traditional birth attendant, com- 1993; Cochrane, Guilkey, and Akin 1994). pared to only 18 percent of women who had 24 It is sometimes suggested that cost- completed middle school (Serrato and benefit analysis is appropriate to use in con- Melnick 1995). These marked differentials by sidering the overall share of resources de- education are in contrast to the situation in

Click to return to Table of Contents 732 POPULATION AND REPRODUCTIVE HEALTH family planning where, for example, data of the country. It included a survey of 1,890 from the 1994 DHS indicate that 83 percent households and a series of commune, school, of women without any schooling know at and health facility surveys. The VNSSFS was least one modern contraceptive method and one module within a much larger multi- where to obtain it (Central Bureau of Sta- round survey being conducted at the time tistics 1995). by the GSO, i.e., the Multi-Objective House- 28 Strictly speaking, “moral hazard” re- hold Survey (MOHS), that covers 45,000 fers to an insured person engaging in more households in 1,500 communes located in risky behavior as a result of being insured all 53 provinces of Vietnam. For more infor- (e.g., an insured bank makes riskier loans); mation, see Behrman and Knowles (1998a). in the health economics literature, the term 33 The estimates were obtained as fol- refers to a simple movement along the de- lows. The annual service utilization rates by mand curve in response to a price change. service delivery source were calculated as the 29 This is an example of a distortionary product of the contraceptive prevalence rate cost attendant on a governmental policy in- (Table 1), the proportion of married women tervention, as discussed above. of reproductive age in the total population 30 This parameter affects the relative val- (Table 2), and the proportion of the target ues a consumer places on large losses (e.g., the population using a given source (Table 1). cost of an expensive hospitalization) as com- Benefit incidence was calculated as the prod- pared to relatively small losses (e.g., payment uct of annual service utilization rates and the of an insurance premium) and is at the heart unit (per acceptor) public subsidy (i.e., = unit of the valuation of welfare gains associated public cost + unit incentive paid – fee rev- with risk pooling. Unfortunately, there is no enue) by source (i.e., Central hospital VND empirical basis for this assumption (because 1,310,441, Provincial hospital VND 794,647, utility is not measurable). However, the esti- District hospital VND 538,029, Polyclinic mates can be calculated for a range of plau- VND 129,281, Commune health center VND sible values of this parameter, and they can be 135,092, and Private VND 4,418). Targeting checked against responses in surveys to ques- efficiency was calculated by computing each tions measuring demand using the technique quintile’s percentage share of total benefits. of contingent valuation. See Behrman and 34 As with income, the Gini coefficient Knowles (1998b) for an application of contin- (i.e., the ratio of the area between the plot- gent valuation to estimate the demand for ted Lorenz distribution of income and a 45- health insurance. degree line from the origin to the area of the 31 Although knowledge of some contra- triangle bordered by the 45-degree line) is a ceptive methods is widespread (IUDs, for ex- useful summary measure of the degree of ample), knowledge of other methods is lim- inequality in benefit incidence. However, ited (Knodel et al. 1995). whereas the Gini coefficient for income varies between 0 (complete equality, i.e., every- 32 Consumption expenditures are used one receives the same income) and 1 (com- to represent income in much of the litera- plete inequality, i.e., all income is received ture because they better represent longer- by one person), the Gini coefficient for ben- run income constraints than does measured efit incidence ranges between –1 (all subsi- annual income if there are large transitory dies received by the poorest person) and 1 income fluctuations and if households can (all subsidies received by the richest person). smooth consumption over time. The principal data source is the Vietnam Social Sector 35 It is also about the same share as the Financing Survey (VNSSFS), which was con- poorest quartiles of the population were es- ducted by the General Statistical Office timated to receive in two other studies of (GSO) of the Government of Vietnam in family planning program benefit incidence January 1996 in collaboration with us, with in Cebu, Philippines and in Indonesia (Com- funding from the Asian Development Bank. mittee on Population 1995). The VNSSFS was conducted in seven prov- 36 In making such a policy change, how- inces, one from each administrative region ever, it is important to introduce fees that re-

Click to return to Table of Contents J ERE R. BEHRMAN / JAMES C. KNOWLES 733 cover most, if not all, program costs. Other- two lowest groups are eligible to receive fam- wise, the effect of introducing fees may fur- ily planning services free (Knowles 1996). ther limit access of the poor to the most heavily Indonesia also has developed a sophisticated subsidized services (in absolute terms), so that system of geographical targeting of rural dean even higher share of program subsidies ac- velopment loans and other interventions at crues to the rich. Such an adverse consequence the village level (World Bank 1993b). of user fees actually occurs in the case of safe 41 Providers and community motivators motherhood services in Vietnam (Behrman also have received incentive payments in and Knowles 1998a). some countries, usually in connection with 37 In some cases there may be a strong clients’ acceptance of long-term methods rationale for subsidizing certain contracep- (i.e., sterilization, IUD). This practice is tive methods more heavily. The condom, for viewed as a way to compensate for low sala- example, provides protection against sexu- ries by motivating public-sector staff to pro- ally transmitted diseases and AIDS. Its use vide additional services. It is open to abuse, generates significant externalities to other however, if providers try to steer clients to- consumers in the form of reduced risk of ward those contraceptives for which incen- contracting these illnesses. tive payments are received. A policy higher 38 Even if, for example, long-term in the policy hierarchy for addressing the methods are more heavily subsidized be- problem of low salaries in the public sector cause they are perceived to be more cost-ef- is to pay higher salaries (reducing the num- fective in reducing fertility, the subsidy will ber of employees, if necessary). be ineffective if demand for the method is 42 USAID policy, for example, has long highly inelastic (i.e., a price reduction does been that “No AID funds can be used to pay not encourage much additional demand). potential acceptors of sterilization to induce Even in a program with a narrowly defined their acceptance of voluntary sterilization” goal of reducing fertility, optimal levels of (quoted in Ross and Isaacs 1988). subsidies can be determined only by consid- 43 Unmet need is a concept developed ering both supply and demand factors (Ham- by demographers (and still being refined by mer and Berman 1995). them) that is broadly defined as the number 39 Indonesia’s family planning program of women who want to limit or space births has for many years offered subsidies to both but who are not currently using contraceptives governmental and private providers, but it (Westoff 1997). Although demographers has often been criticized for offering subsi- sometimes refer to it as “unmet demand” or dies to the latter because this is perceived as “potential demand” for contraceptives, it en- threatening the program’s “sustainability.” tails a definition of “demand” different from However, there is strong justification in our that used in standard economics and policy framework for offering subsidies equally to analysis (i.e., willingness to purchase a given the public and private sectors, as long as sub- quantity and quality of a good or service at sidies can be justified, and for terminating sub- a given price). sidies to both sectors if they are not justifiable. 44 All of these efficiency advantages of 40 If means testing is impractical, the decentralization, however, are also generally government may be able to target the poor assured if services are provided by a com- through subsidizing goods and services that petitive private sector. are mainly consumed by the poor or by fo- 45 It is not clear, however, that democrati- cusing subsidies on geographical areas in cally elected officials necessarily are appropri- which the poor are concentrated. In Indo- ately responsive to local preferences. Median nesia’s family planning program, for ex- voter models, for example, imply that elected ample, every household in the country is officials are responsive to the preferences of classified in January of each year into one median voters, but such responsiveness may of several “family welfare” groups on the ba- ignore the preferences of minorities—and ig- sis of characteristics of their housing and noring the preferences of minorities may be lifestyles. Persons from households in the viewed as socially inappropriate.

Click to return to Table of Contents 734 POPULATION AND REPRODUCTIVE HEALTH

46 For example, where services are of a feel have even higher social marginal ben- public good nature, the same arguments can efits than are perceived by recipient govern- be advanced against purely local governmen- ments. For example, donors may choose to tal finance and administration of these ac- support family planning out of a desire to tivities as opposed to individual financing reduce global rates of population growth, be- through user fees. Similarly, where exter- cause of a perception that this will improve nalities are present, they may not be limited the environment. When this is the case, to the population of localities. sustainability depends basically on the main- 47 There is little reason to expect they tenance of donor will. would succeed, because in competitive (and 50 Most price elasticity estimates for adult even monopolistically competitive) markets ambulatory care in rural Indonesia are in the there is no reason to expect producers to en- range 0.0 to –0.4 (Gertler and Molyneaux joy long-run profits. If activities in high-in- 1997). Most village midwives charge 40,000– come markets are profitable (i.e., generate 70,000 rupiah (Satoto et al. 1995). revenue above actual costs), other investors 51 US$1=2,348 rupiah at the time the are attracted into the market and the in- project was designed. creased supply drives prices down to the 52 This is a conservative estimate of the point where price is equal to average cost variance because it does not incorporate (i.e., zero profit). This has in fact been the variation in risk within each group. experience of many donor-funded cross-subsidizing ventures. 53 The expression I*r(I) is called the consumer’s income elasticity of the marginal 48 If governments do provide these ser- utility of income, i.e., the percentage change vices directly (and most do), they should be in the consumer’s marginal utility of income made to compete equally with the private for a one percent change in income. For ex- sector: that is, the private sector should re- ample, if I r(I) = –0.4, a one percent increase ceive the same level of subsidies received by * in the consumer’s income is associated with governmental providers. a 0.4 percent decline in the marginal utility 49 Of course, some donors may wish to of income. provide continuing subsidies to services they

References

Behrman, Jere. 1993. “Past and current development paradigms and future directions in development policy and planning,” UNFPA Round Table on Population and Develop- ment Strategies. Bangkok. Behrman, Jere and James C. Knowles. 1998a. “The distributional implications of government family planning and reproductive health services in Vietnam,” mimeo., Univer- sity of Pennsylvania. ———. 1998b. “The demand for health insurance in Vietnam,” mimeo., University of Penn- sylvania. Bloom, David E. and J. Lyons (eds.). 1993. Economic Implications of AIDS in Asia. New York: United Nations Development Programme. Bloom, David E. and Ajay S. Mahal. 1997. “Does the AIDS epidemic threaten economic growth?” Journal of Econometrics 77, no. 1: 105–123. Central Bureau of Statistics, National Family Planning Coordinating Board, Ministry of Health, and Macro International. 1995. Indonesia Demographic and Health Survey 1994. Calverton, MD: CBS and MI. Chao, Dennis N. and Karen B. Allen. 1984. “A cost-benefit analysis of Thailand’s family planning program,” International Family Planning Perspectives 10, no. 3: 75–81. Coale, Ansley J. and Edgar M. Hoover. 1958. Population Growth and Economic Development in Low-Income Countries. Princeton: Princeton University Press.

Click to return to Table of Contents J ERE R. BEHRMAN / JAMES C. KNOWLES 735

Cochrane, Susan and Frederick Sai. 1993. “Excess fertility,” in Dean T. Jamison et al. (eds.), Disease Control Priorities in Developing Countries. New York: Oxford University Press. Cochrane, Susan H., David H. Guilkey, and John S. Akin. 1994. The Cost-Effectiveness of Fam- ily Planning in Reducing the Mortality of Women and Their Offspring. Washington, DC: The World Bank. Coeytaux, Francine, Dayl Donaldson, Touhami Aloui, Taoufik Kilani, and Habib Fourati. 1989. “An evaluation of the cost-effectiveness of mobile family planning services in Tunisia,” Studies in Family Planning 20, no. 3: 158–169. Cohen, Susan A. and Cory L. Richards. 1994. “The Cairo consensus: Population, development and women,” International Family Planning Perspectives 20, no. 4: 150–155. Committee on Population. 1995. “Resource allocation for family planning in developing countries.” Report of a Meeting. Washington, DC: National Research Council. Cuddington, John T. 1993. “Modeling the macroeconomic effects of AIDS, with an application to Tanzania,” World Bank Economic Review 7: 173–189. Cuddington, John T. and John D. Hancock. 1994. “Assessing the impact of AIDS on the growth path of the Malawian economy,” Journal of Development Economics 43, no. 2: 363–368. Deaton, Angus. 1997. The Analysis of Household Surveys: A Microeconometric Approach to Devel- opment Policy. Baltimore: The Johns Hopkins University Press for the World Bank. Devarajan, Shanatayanan, Lyn Squire, and Sethaput Suthiwart-Narueput. 1997. “Beyond rate of return: Reorienting project appraisal,” World Bank Research Observer 12, no. 1: 35–46. Enke, Stephen. 1966. “The economic aspects of slowing population growth,” The Economic Journal 76 (March): 44–56. ———. 1971. “The economic consequences of rapid population growth,” Economic Journal 81: 800–811. Evans, D. and S. Hurley. 1995. “The application of economic evaluation techniques in the health sector: The state of the art,” Journal of International Development 7, no. 3: 503– 524. Feldstein, Martin. 1995. “Tax avoidance and the deadweight loss of the income tax,” Cam- bridge, MA: National Bureau of Economic Research Working Paper No. 5055. Folland, Sherman, Allen C. Goodman, and Miron Stano. 1993. The Economics of Health and Health Care. Englewood Cliffs, NJ: Prentice-Hall. Foreit, Karen G., Marcos Paulo P. de Castro, and Elaine F. Duarte Franco. 1989. “The impact of mass media advertising on a voluntary sterilization program in Brazil,” Studies in Family Planning 20, no. 2: 107–116. Foreit, Karen G., James R. Foreit, Gloria Lagos, and Alfredo Guzman. 1993. “Effectiveness and cost-effectiveness of postpartum IUD insertion in Lima, Peru,” International Fam- ily Planning Perspectives 19, no. 1: 19–24. General Statistical Office. 1991. Viet Nam Population Census—1989: Detailed Analysis of Sample Results. Hanoi: Statistical Publishing House. ———. 1995. Viet Nam Intercensal Demographic Survey 1994: Major Findings. Hanoi: Statistical Publishing House. Gertler, Paul, Luis Locay, and Warren Sanderson. 1987. “Are user fees regressive? The welfare implications of health care financing proposals in Peru,” Journal of Econometrics 36: 67–88. Gertler, Paul and John W. Molyneaux. 1997. “Experimental evidence on the effect of raising user fees for publicly delivered health care services: Utilization, health outcomes, and private provider response.” RAND, Santa Monica, CA. Hammer, Jeffrey S. 1997. “On economic analysis for health projects,” World Bank Research Observer 12, no. 1: 47–171. Hammer, Jeffrey S. and Peter Berman. 1995. “Ends and means in public health policy in developing countries,” in Peter Berman (ed.), Health Sector Reform in Developing Coun- tries. Cambridge, MA: Harvard University Press.

Click to return to Table of Contents 736 POPULATION AND REPRODUCTIVE HEALTH

Haveman, Robert H. 1976. “Benefit-cost analysis and family planning programs,” Popula- tion and Development Review 2, no. 1: 37–64. Janowitz, Barbara and Brian J. Gould. 1993. “Options for financing family planning.” Draft report for United Nations Population Fund. Family Health International, Research Tri- angle Park, NC. Jimenez, Emmanuel. 1995. “Human and physical infrastructure: Public investment and pricing policies in developing countries,” in Handbook of Development Economics, ed. Jere R. Behrman and T. N. Srinivasan, Vol. III, Chapter 43. Amsterdam: Elsevier. Kambou, G., Shanta Devarajan, and Mead Over. 1992. “The economic impact of AIDS in an African country: Simulations with a computable general equilibrium model of Cameroon,” Journal of African Economics 1: 109–130. Knodel, John, Phan Thuc Anh, Truong Viet Dung, and Dao Xuan Vinh. 1995. “Why is oral contraceptive use in Vietnam so low?” International Family Planning Perspectives 21, no. 1: 11–18. Knowles, James C. 1996. “Options for expanding private sector reproductive health services in Indonesia.” Consultant report to the World Bank, Abt Associates Inc. ———. 1997a. “Economic analysis of the safe motherhood project.” Draft consultant report to the World Bank, Abt Associates Inc. ———. 1997b. “Cost-benefit and cost-effectiveness analysis in family planning.” Unpub- lished mimeo. Paper prepared for Joint IUSSP/EVALUATION Project Meeting on Meth- ods for Evaluating Family Planning Programme Impact, Costa Rica, 14–16 May. Mai Ky. 1994. “Statement as Head of the Vietnamese Delegation at the International Con- ference on Population and Development,” Cairo, Egypt, 5–13 September. Mason, Karen Oppenheim. 1994. “Do population programs violate women’s human rights?” Asian Pacific Issues. No. 15. East-West Center, Honolulu, HI. Phelps, Charles E. 1992. Health Economics. New York: Harper-Collins. Pradhan, Sanjay. 1996. “Evaluating public spending: A framework for public expenditure reviews.” World Bank Discussion Paper No. 323. Washington, DC: The World Bank. Preston, Samuel H. 1986. “Are the economic consequences of population growth a sound basis for population policy?” in Jane Menken (ed.), World Population and U.S. Policy: The Choices Ahead, New York and London: W. W. Norton. Prud’homme, Rémy. 1995. “The dangers of decentralization,” World Bank Research Observer 10, no. 2: 201–220. Robinson, Warren C. and David E. Horlacher. 1969. “Evaluating the economic benefits of fertility reduction,” Studies in Family Planning 39: 4–8. Rochjati, Poedji, Agus Abadi, Budi Rahayu, Benny Soegianto, and Wasis Budarto. 1996. “A report on simulation model: A model of service package and parturition cost package in safe motherhood program at regency level,” Reproductive Health Research/Study Center, University of Airlangga, Surabaya, East Java. Ross, John A. and Stephen Isaacs. 1988. “Costs, payments, and incentives in family planning programs.” PHN Working Paper Series No. 88. Washington, DC: The World Bank. Satoto, Dr. et al. 1995. “Final report: Research on the implementation of the role of village midwives in the mother and child health (MCH) services in Central Java,” Diponegoro University, Semarang, Central Java. Serrato, Carl A. and Glenn Melnick. 1995. “The Indonesian Family Life Survey: Overview and descriptive analysis,” Report submitted to USAID/Jakarta, RAND Corporation, Santa Monica, CA. Simon, Julian. 1969. “The value of avoided births to underdeveloped countries,” Population Studies 23, no. 1: 61–68. ———. 1977. The Economics of Population Growth. Princeton: Princeton University Press. Simmons, George B., Deborah Balk, and Khodezatul K. Faiz. 1991. “Cost-effectiveness analysis of family planning programs in rural Bangladesh: Evidence from Matlab,” Studies in Family Planning 22, no. 2: 83–101.

Click to return to Table of Contents

JERE R. BEHRMAN / JAMES C. KNOWLES737

Sirageldin, Ismail, David Salkever, and Richard W. Osborn. 1983. Evaluating Population Pro- grams: International Experience with Cost-Effectiveness Analysis and Cost-Benefit Analysis. Lon- don: Croom Helm. Sommers, Paul M. 1980. “The value of birth control,” Applied Economics 12, no. 2: 195–207. Sommers, Paul M. and Daniel B. Suits. 1971. “A cross-section model of economic growth,” Review of Economics and Statistics 53: 121–128. Suits, Daniel B., W. Mardfin, S. Paitoonpong, and T. Yu. 1975. “Birth control in an econometric simulation,” International Economic Review 16: 92–111. Tanzi, Vito. 1995. “Fiscal federalism and decentralization: A review of some efficiency and macroeconomic aspects,” Washington, DC: International Monetary Fund, mimeo. (Pa- per presented at World Bank Annual Economic Development Conference, 1995). van de Walle, Dominique and Kimberly Nead (eds.). 1995. Public Spending and the Poor: Theory and Evidence. Baltimore and London: Johns Hopkins University Press. Vernon, Ricardo, Anne Staunton, Mario García, Raúl Rosenberg, and Juan José Arroyo. 1994. “A test of alternative supervision strategies for family planning services in Gua- temala,” Studies in Family Planning 25, no. 4: 232–238. Walsh, Julia A., Chris N. Feifer, Anthony R. Measham, and Paul J. Gertler. 1993. “Mater- nal and perinatal health,” in Dean T. Jamison et al. (eds.), Disease Control Priorities in Developing Countries. New York: Oxford University Press. Warner, Kenneth E. and Bryan R. Luce. 1982. Cost-Benefit and Cost-Effectiveness Analysis in Health Care. Ann Arbor: Health Administration Press. Westoff, Charles F. 1997. “Unmet need revisited.” Mimeo., Office of Population Research, Princeton University. World Bank. 1993a. World Development Report 1993. New York: Oxford University Press. ———. 1993b. Indonesia Public Expenditures, Prices and the Poor. Indonesia Resident Mission, Country Department III, East Asia and Pacific Region. Washington, DC: The World Bank. ———. 1995. Viet Nam: Poverty Assessment and Strategy. Report No. 13442-VN. East Asia and Pacific Region. Washington, DC: The World Bank. ———. 1997. “Project appraisal document on a proposed loan in an amount of $42.5 million to the Republic of Indonesia for a safe motherhood project: A partnership and family approach,” Report No. 16624-IND, East Asia and Pacific Region. Washington, DC: The World Bank. Zaidan, George. 1969. “Population growth and economic development,” Studies in Family Planning 42: 1–6. ———. 1971. The Costs and Benefits of Family Planning Programs. Baltimore: Johns Hopkins University Press.

Click to print article Click Clickto return to return to Table to of Table Contents of Contents Click to return to Table of Contents

Immigration Policy Prior to the 1930s: Labor Markets, Policy Interactions, and Globalization Backlash

ASHLEY S. TIMMER JEFFREY G. WILLIAMSON

AFTER THE 1870S, New World doors gradually closed to immigrants. Al- though much of the literature has focused on the drastic policy changes in the period just after World War I, the doors did not suddenly slam shut on American immigrants when the United States Congress overrode President Woodrow Wilson’s veto of the immigrant literacy test in February 1917 or when it passed the Emergency Quota Act of May 1921. A half-century before the Literacy Act, the United States started imposing restrictions on what had been free immigration, and the United States was not the only country becoming less receptive to immigrants. Argentina, Australia, Bra- zil, and Canada enacted new measures, although the timing varied, and the policies often took the form of a large drop in or even disappearance of immigrant subsidies rather than outright exclusion of immigrants. Immi- gration policy varied considerably across these five countries over the period from 1860 to 1930, the conventional portrayal of one big policy switch around World War I to the contrary. What was true of immigration policy was also true of trade policy. Globalization proceeded in fits and starts after 1846 when Britain repealed the Corn Laws and started a liberal trend toward free trade. It took the form of mass migrations, a trade surge, and international capital flows at (relative) levels never reached before or since.1 The liberal trend did not last long, however, in the face of a globalization backlash. Tariffs started to rise on the European continent. Restrictions on immigration and trade started to rise in the New World. With the end of World War I, the world economy plunged into a dark age of de-globalization and policy antago- nism toward factor and goods mobility.

POPULATION AND DEVELOPMENT REVIEW 24(4):739–771 (DECEMBER 1998)739

Click to print article Click toClick return to returnto Table to of Table Contents of Contents 740 I MMIGRATION POLICY PRIOR TO THE 1930S

What explains the globalization backlash? A number of candidates have been nominated in the case of immigration policy. Economists have suggested that immigrants crowded out native unskilled workers and contributed to rising inequality in labor-scarce economies, as did free trade. The policy reaction may also have reflected the greater voting power in the hands of those hurt most—the working poor. Possible noneconomic factors include increasing racism, xenophobia, and widening ethnicity gaps between the population stock and the current immigrant flow. The increase in immigrant flows, “lower-quality” immigrants, and the threat of even lower- quality immigrants may have provided further impetus to close the doors. There have been few attempts to introduce these factors into explicit models of immigration policy formation.2 This article uses what little theory exists to identify the fundamental factors that underlie changes in immigration policy and to clarify the differences between market and nonmarket influences. In addition, the article explores the extent to which policy responded to the impact of immigrants on labor markets, and the extent to which it tried instead to anticipate those impacts by reacting to the quantity and quality of immigrants. Finally, the article assesses the impact of policies abroad on policies at home. Which countries were most sensitive to immigration policies elsewhere in the New World, and to what extent did the largest among them, the United States, set the pace for the rest? Trade policy over this period was clearly an interactive game, as countries sought to respond to escalat- ing tariffs elsewhere. But little has been said about how other New World destinations reacted to increasing immigration restrictions by the United States, or, for that matter, how they responded to the push of emigrants out of the United Kingdom. We seek to examine the degree to which one country’s immigration policy may have been influenced by the policies of others.

Measuring immigration policy

Having set a goal to calibrate the determinants of immigration policy, we must construct a measure that can quantify the policy in the New World. Such a measure is necessary if we hope to assess the extent to which globalization backlash was at work and to identify the form that it took. We have designed an index of immigration (and in one case, emigration) policy that will be used to confront a set of competing hypotheses. The index is likely to be subject to the same criticism as those used to measure trade openness (Anderson and Neary 1994; Sachs and Warner 1995). We recognize that a subjective component to the index remains, but we have tried to use a consistent algorithm. The index ranges over a scale of +5 to –5. A positive score denotes a pro-immigration policy, possibly including comprehensive subsidies for pas-

Click to return to Table of Contents A SHLEY S. TIMMER / JEFFREY G. WILLIAMSON 741 sage and for support upon arrival. A negative score denotes anti-immigration policy, possibly including quotas, literacy tests, and legal discrimina- tory treatment after arrival. A zero denotes policy neutrality, or a neutral outcome between conflicting pro- and anti-immigration policies. A policy can have two functions: first, to signal to groups that their interests are being tended to and second, to change the status quo. Clearly, political agents were trying to affect the flow of immigrants and to respond to their constituencies. Policies were not always effective, but the goal here is to capture the intention, or political signal. The following algorithm was the basis for our assigning scores, and we use it consistently across countries and over time:

5 Active worker recruitment abroad with advertising and labor offices, free land or subsidized land purchase, subsidized or assisted passage, temporary lodging, free transport inland from port of arrival, easy naturalization, legal property ownership. 4 Free or subsidized land, immigration treaties or contracts with shipping companies, lodging, worker recruitment, easy naturalization, legal property ownership. 3 Overseas immigration offices, debarkation coordination, land designated for settlement, easy naturalization, legal property ownership. 2 Overseas immigration offices, debarkation coordination, easy naturalization, legal property ownership. 1 Modest advertising, easy naturalization, legal property ownership. 0 Open doors, no encouragement, no discouragement. Or, a balance of pro-immigration and anti-immigrant policies. –1 Regulations on shipping companies and/or contracts for assisted passage. –2 Class restrictions on immigration (no paupers, potential wards of the state, criminals) or selective source-country bans (e.g., no Asians). –3 The above restrictions plus laws for registration, deportation provisions, laws restricting property ownership, unenforced selectivity laws (such as literacy tests). –4 Restrictive quotas, enforced literacy tests, or other measures designed to reduce immigration volume significantly. –5 Closed (or only slightly ajar) doors, enforced.

Several complications repeatedly arose in assigning scores to policy. Canada, Australia, and the United States all enacted legislation against Asian immigration, even while encouraging (or at least not discouraging) immigration from Europe. At some points, all three countries had a set of policies that sent a mixed message to both potential immigrants and constitu-

Click to return to Table of Contents 742 I MMIGRATION POLICY PRIOR TO THE 1930S ents. Whenever we found a mix of pro-immigration and anti-immigration policies, we simply added up the positive and negative attributes to get an overall score. Since a source-country ban on immigration was generally scored –2, and subsidy and recruitment programs were generally scored +3, Canada received a net score of +1 around the turn of the century. Simi- larly, in the early twentieth century, Australia recruited and subsidized immigration but also required a dictation test on demand, and we scored this mix a 0 for several years. Because of the mix of policies, we allowed half- steps in the scoring. Appendix B uses United States experience to illustrate the scoring, but the full details for all countries can be found elsewhere (Timmer and Williamson 1996: Appendix C). To assess the impact of Em- pire settlement plans, we also scored the emigration policy of the United Kingdom, using a parallel algorithm. These policy indexes, plotted in Figure 1, confirm that whereas immigration in the 1860s was generally unrestricted, the doors to the New World were effectively closed by 1930. But in the intervening decades, the trends were less clear. Argentina started on a path of increasing openness and pro-immigration subsidies, but reversed the policy in the 1880s, and the index dropped from +4.5 to –2.5 by the 1920s. Brazil’s index underwent a similar path, although the anti-immigrant legislation all came in a rush at the end of the period. Australia’s index fell from +3 in the mid-1860s to –1 shortly after the turn of the century, and to –2 in 1930, but exhibited short episodes of more open policies, especially in the 1920s. Canada’s index be- haved similarly. The United States index fell from 0 in the early 1860s to –5 by 1930, and it was the only country never to have a major policy reversal over the period. Thus, while the United States exhibited a steady drift away from free immigration, the others closed their doors in fits and starts. Although there are some cases of substantial short-term variance, as in Australia between 1890 and 1930, strong policy persistence is the norm. Policy was slow to change, sometimes constant over a decade or more, even though intensive political debate often accompanied the apparent qui- escence. The best examples of this stability are Brazil over the three decades from 1890 to 1920, a period that ended in 1921 when immigration restrictions were imposed, and the United States from 1888 to 1916, a period that ended with the override of President Wilson’s veto of the legislation introducing an immigrant literacy test in 1917. The literature offers several explanations of the evolution of immigration policy from the middle of the nineteenth century to the Great Depression.

Models of immigration policy

Formal models of immigration policy are few, but there is a general consensus that immigration policy has always been sensitive to labor market

Click to return to Table of Contents A SHLEY S. TIMMER / JEFFREY G. WILLIAMSON 743

FIGURE 1 POLICY: An immigration policy index Argentina Australia 5 5 4 4 3 3 2 2 1 1 0 0 -1 -1 –2 –2 -3 -3 –4 –4 -5 -5 1850 1870 1890 1910 1930 1850 1870 1890 1910 1930

Brazil Canada 5 5 4 4 3 3 2 2 1 1 0 0 -1 -1 –2 –2 -3 -3 –4 –4 -5 -5 1850 1870 1890 1910 1930 1850 1870 1890 1910 1930

United States United Kingdom 2 5 4 3 0 2 1 –2 0 -1 –2 –4 -3 –4 -5 1850 1870 1890 1910 1930 1850 1870 1890 1910 1930

NOTE: The index ranges over a scale of +5 to –5. A positive score denotes a pro-immigration policy, a negative score an anti-immigration policy. See text for further explanation. conditions,3 and that immigration itself has always been sensitive to wage and unemployment rate differentials between countries. For example, Claudia Goldin (1994) notes that in the United States in the late 1890s,

Click to return to Table of Contents 744 I MMIGRATION POLICY PRIOR TO THE 1930S during a time of economic recession and high unemployment, there was a new push for immigration restrictions. At that time, however, the rate of immigration slowed markedly, reaching a low in 1897, the same year that the first vote on immigration restriction was taken in the House of Repre- sentatives. Similarly, Australian inflows dropped sharply in the recession of the 1890s when attitudes inimical to immigrant subsidies hardened (Pope and Withers 1994). These observations suggest that the impetus to restrict immigration was far more sensitive to labor market conditions than to immigration levels. To complicate matters, the ethnic composition of immigrants was clearly a factor in the politics of restriction. Australia maintained a strict policy aimed at keeping the country one of British and Irish descent, while avoiding persons of “yellow” skin (Pope and Withers 1994). The United States completely banned immigrants from China in 1882 and immigrants from all of Asia in 1917 (Green 1995). Increasing demands for restriction in the 1880s and 1900s paralleled an increase in the relative numbers of immigrants from southern, central, and eastern Europe, the so-called new immigrants. It is difficult to sort out whether these policies were a result of racism and xenophobia or whether ethnic origin merely served to signal, however imperfectly, the human capital content or “quality” of the immigrants (Foreman-Peck 1992).4 If countries were sensitive to the source of immigrants, this further suggests that there might be competition among them for those of higher quality, or competition to keep out those of lower quality. Did Argentina, for example, have to subsidize immigration in order to attract better immigrants? Were they successful in doing so? As James Foreman-Peck (1992) notes, the two central questions for any model of immigration policy formation are: Who gains and who loses? Who decides the policy? Consensus is clear regarding the first question. Wage earners—unskilled workers in particular—lose with immigration, as the labor pool swells and wages sag. Owners of other factors of production—land, capital, and perhaps even skills—gain from the more abundant unskilled labor supply that makes these other factors more productive. We hasten to add two caveats. While most attempts to measure the impact of mass migration on wages prior to 1914 have found that wages were down- wardly sensitive to immigration (Williamson 1974; Taylor and Williamson 1997; Green 1994; Goldin 1994; Hatton and Williamson 1995; Williamson 1996), a study of Australia found that wages actually increased with immigration, if only marginally (Pope and Withers 1994). The Australian result could be explained if immigrants augmented labor demand enough to offset their impact on increased labor supply (for example, by working previously unsettled land or by inducing an accumulation response as capital from the home country chased after labor). If labor demand keeps pace with labor supply, then native labor is not hurt by immigration.

Click to return to Table of Contents A SHLEY S. TIMMER / JEFFREY G. WILLIAMSON 745

The second caveat concerns disequilibriums in the labor market, when the impact of immigration on wages is unclear. The issue of unemployment has not really been examined in the context of immigration policy, but suppose wages are sticky downward and unrelated to the size of the unemployment pool, perhaps for efficiency reasons or “fairness.”5 In such a situation, immigration will not have any effect on wages, but it will add to the numbers unemployed. No one benefits from immigration: capitalists do not gain by a fall in wages, and the number of unemployed increases. Eventually, both sides might unite in favor of immigration restrictions. Goldin (1994) suggests that this aligning of interests occurred in the United States during the 1890s. Note, however, that the impact of an economic downturn on native unemployment should have been partially muted by immigrant behavior, as recent (but now jobless) immigrants returned home—that is, immigrants themselves did voluntarily what a policy of immigrant restriction would have done.6 This they did in great numbers, but not great enough to make the “guestworker effect” operate with much quantitative muscle, even during that critical depression decade of the 1890s (Hatton and Williamson 1995); while the US unemployment rate was reduced a bit by return migration in the 1890s, the reduction was a modest share of the total unemployment rate. These two caveats aside, most discussions of the politics of immigration assume that the interests of capital and labor are divided. As such, the immigration literature should be closely aligned with theories of long-run interest groups in trade policy. The parallels between trade and immigration policy are discussed at length in our previous article (Timmer and Williamson 1995) so here we focus only on immigration policy. In addition to the capital–labor divide, Foreman-Peck (1992) argues that land ownership might have mattered, especially in the late nineteenth century when agriculture was still a large sector of the economy.7 He takes the following approach. Assume that individuals receive their incomes from one of the following three sources: wages, profits, or land rents. Depend- ing on the franchise, the government maximizes a weighted objective function that includes rents, profits, and wages of native labor. The critical question is whether immigrant and native labor are complements or substitutes in production: if they are substitutes, then immigration hurts wages of the natives. Estimating a production function, Foreman-Peck concludes that they were substitutes in the late-nineteenth-century US economy. Thus, the larger the weight on labor interests, the more restrictive the immigration policy. The reverse is true as the political system attaches larger weights to the interest of capital or land. Foreman-Peck allows for the possibility of two types of immigrants: skilled and unskilled. It might be that skilled immigrant labor was a comple- ment to domestic labor, whereas unskilled immigrant labor was a substi-

Click to return to Table of Contents 746 I MMIGRATION POLICY PRIOR TO THE 1930S tute. We would then expect to see a policy that encouraged the immigration of skilled and discouraged the immigration of unskilled workers. Foreman- Peck argues that this concern about unskilled labor, and not racism or xenophobia, was responsible for policies in the Americas that restricted Asian immigration and for policies in South Africa that restricted African immigration. Although Foreman-Peck does not implement a formal empirical test, his discussion of Argentina, Britain, South Africa, and the United States indicates that some of the historical facts are consistent with his theory. For example, landed interests were largely in control of immigration policy in Argentina and the government offered generous subsidies to attract farm laborers from the Mediterranean Basin. In contrast, the United States had a more universal voting franchise, rejected subsidies, and gradually closed the door as the frontier itself was closed (by 1890, or so said the Census Commissioner at that time). Goldin (1994) takes a different approach. Following a long tradition in American historiography that has focused on sectional interests, she looks at regional splits and rural–urban differences. Although she does not model the relationship formally, she assumes that individual US Senators and Rep- resentatives advocate policies that favor their constituents, in proportion to the numbers represented by each urban, rural, and regional interest group. The passage of the literacy test, which was first attempted in 1897 and was finally successful in 1917, seems to have been the result of two (often opposing) forces: demographic changes and changes of heart. The changes of heart were many. Goldin suggests that capitalists were for the first time aligned with labor in opposing immigration during the recessionary years of the 1890s when unemployment was high. Later, during times of full employment and rising wages, capital shifted back to its more traditional pro-immigration stance, but the South adopted an anti-immigration stance, a change of heart probably motivated by the urge to protect its relative population share and voting clout. Finally, the northern Midwest, fairly pro-immigration in the 1890s, underwent an anti-immigration switch following World War I. Goldin argues that this was mostly a change of heart by older immigrant groups, pushed to patriotism by the war. The political impact of the change in North–South demographic composition was offset by the changing composition of the cities. Goldin finds that the probability that a legislator would vote for immigration restrictions was negatively related to the proportion of foreign-born in the district and was also negatively related to the level of urbanization. This relationship suggests that efforts of what we might now call family reunification were operating in the cities, and since cities were on the rise pro-immigration interests increasingly made themselves heard. More important than either of these nonmarket influences, however, was the impact of increasing immigration on wages and the subsequent

Click to return to Table of Contents A SHLEY S. TIMMER / JEFFREY G. WILLIAMSON 747 effect on votes. Especially after the turn of the century, Goldin finds a significant negative impact of immigration on wages, a result consistent with other historical studies (Hatton and Williamson 1998). The change in real wages is, in turn, a significant explanatory variable in accounting for the Congressional vote to override the presidential veto of the literacy test in 1915. The higher the growth in wages, the less likely was the Representa- tive to vote for an override (and thus for restriction). These two findings of Goldin’s research—that wages influenced US immigration policy and that immigrants influenced wages in American labor markets—are useful in our comparative assessment of immigration policy in the New World. However, we only require that politicians and their constituents believed that immigration retarded wage advance. It appears that they did. William Shughart, Robert Tollison, and Mwangi Kimenyi (1986) look at shifting degrees of enforcement of immigration restrictions. Workers want high wages, and they pressure politicians to enforce immigration restrictions. Capitalists and landowners want lower wages, and they try to reduce enforcement. Their model predicts that as the economy goes through business cycles, the ideal policy mix shifts, resulting in changes in the degree of enforcement against immigration. The authors test the model using data from the United States from 1900 to 1982, and the results are supportive. Even taking into account official changes in immigration policy, the size of the enforcement budget, and the political party in the White House, the degree of enforcement is significantly, and negatively, related to real GNP. Unemployment and the real wage were also significant explanatory variables, but not so consistently as real GNP. Had the authors looked at US policy toward indentured labor contracts prior to 1900, they would have seen the same correlation: harsh policy during slumps; soft policy during booms. The three studies discussed above are the only ones to offer empirical support for theories of immigration policy.8 All three address the role of labor markets, but they limit their attention to absolute gains and losses resulting from immigration, ignoring the relative effects. Recently, a renewed interest in distributional questions has developed among those studying the consequences of migration. Immigrants can create more inequality in the country of destination and less inequality in the country of origin. The empirical literature on this issue has grown voluminous in a short time, perhaps because the consequences of immigration have gained prominence on the American political scene. The debate began over the impact of immigration in the United States (Borjas 1994), expanded to consider Euro- pean immigration (Freeman 1995), and spilled over into the issue of out- migration from developing countries (Wood 1994). The distributional impact of migration has even been confirmed for the late nineteenth cen-

Click to return to Table of Contents 748 I MMIGRATION POLICY PRIOR TO THE 1930S tury, with the demonstration that inequality increased in the receiving countries and decreased in the sending countries (Williamson 1997). There is not yet consensus (or much theory) on how these distributional consequences will affect immigration policy. Consider the impact of immigration on future economic growth. It is true that a labor-scarce country will do better to allow immigration than to allow the export of capital, thus becoming more populous rather than less so (Cheng and Wong 1990). But if immigration induces falling wages or greater inequality, and if, as a consequence, the median voter becomes too poor, then citizens might vote for distortionary redistributive policies that can slow growth. Inequality may also lead to political instability, which can slow growth. While all of these assertions may sound plausible, economists have yet to identify un- ambiguously the impact of inequality on growth.9 There are other models of income distribution and policy formation that do not depend on a link to economic optimization. For example, citizens might vote for restriction on immigration simply because they dislike increased inequality and the lower living standards of their unskilled neighbors (Luttmer 1997). Or, changes in income distribution might tip the balance of political power among competing interest groups, leading to changes in immigration policy (Timmer 1996). Jess Benhabib (1996) suggests that the distribution of capital and labor among voters will, from the perspective of the median voter, affect the skill mix of an ideal immigrant. A relatively capital-rich median voter will prefer a less-skilled immigrant; a relatively labor-rich voter prefers a capital-rich, skilled immigrant.

A menu of hypotheses

This brief review of the literature offers several promising hypotheses that we organize here around a set of explanatory variables. Details of the variables and their sources can be found in Appendix A. First, immigration policy might respond to either the quantity or the quality of immigration, or both. The size of the immigrant flow as a share of the native labor force is one obvious variable, although the experience of the 1890s has already suggested that labor market conditions might have mattered far more than size of flow. The quality of the immigrants is another candidate, measured in comparison with the native labor force. The vast majority of immigrants came from and entered unskilled jobs. Some had good health, high levels of literacy, numeracy, on-the-job training, and considerable exposure to work discipline. Other immigrants did not. Quality and quantity were highly correlated prior to World War I: the switch of emigrant source from higher-wage to lower-wage areas of Europe correlated with the rise in immigration rates. It is likely that these two effects reinforced each other in their impact on policy. A variable that combines

Click to return to Table of Contents A SHLEY S. TIMMER / JEFFREY G. WILLIAMSON 749 the rising quantity of immigration with the falling quality might do better than the two measures of quantity and quality in competition with each other. Second, immigration policy might respond to labor market conditions. This possibility can be sharpened by distinguishing between short-run timing and long-run fundamentals of policy. Unemployment, wage growth, and other macro indicators should serve to isolate the role of business cycles, trade crises, world price shocks, and other short-run events that might affect the timing of changes in immigration policy. In addition, the use of lagged dependent variables should help to ascertain how slowly policy responds to long-run labor market fundamentals. We expect labor market fundamentals to be captured by unskilled real wages—a measure of absolute performance—or by unskilled wages relative to income of the average citizen—a measure of relative performance. The latter is a measure that gauges the unskilled worker’s economic performance against that of the average: it is a measure of inequality that politicians and voters could easily see and understand. The validity of these variables does not require that immigration was the key force driving the living standards of the working poor in the New World. It requires only that politicians and voters believed that immigration was a powerful influence on living standards. Whether it was the absolute or the relative performance that mattered is an empirical issue, but Figure 2 suggests that the inequality variable is likely to do well, especially in the cases of Argentina, Canada, and the United States. Third, a country’s immigration policy may have been influenced by the immigration policies of other countries, either directly or indirectly. If the country anticipates the influence of immigration policies abroad on immigration inflows at home, the effect is direct. Since the labor market in the United States was so large relative to the rest of the New World, and since so many European emigrants went there,10 it is unlikely that the United States paid much attention to immigration policies being introduced elsewhere. Australia may have been more concerned with British Empire settlement policy than with United States policy. Argentina and Brazil, meanwhile, must have paid close attention to United States policy since they could reasonably expect the marginal European emigrant (for example, southern Italian emigrants) to be pulled from or pushed toward Latin America in response to less or more restrictive policy in the United States. Authorities might have moderated those changes by mimicking United States policy before being confronted with the actual migrant response. The same might have been true of Canada, which, in spite of British Em- pire settlement policy, had to accommodate a long porous border with its big neighbor to the south. Fourth, nonmarket forces probably remain after these market forces have been allowed to have their impact. After controlling for immigrant

Click to return to Table of Contents 750 I MMIGRATION POLICY PRIOR TO THE 1930S

FIGURE 2 The inequality and immigration policy correlation

Argentina Australia 120 4 130 3 120 4 100 2 110 2 80 1 100 0 90 0 60 –1 80 –2 40 –2 70

–4 20 –3 60 1850 1870 1890 1910 1930 1850 1870 1890 1910 1930

Brazil Canada 250 4 160

4 200 2 140

2 150 0 120

0 100 –2 100

–2 50 –4 80

–4 0 60 1850 1870 1890 1910 1930 1850 1870 1890 1910 1930

United States 2 160 Income inequality index (1900 = 100, right axis) 140 Immigration policy index 0 (left axis; see Figure 1) 120 –2 100

–4 80

60 1850 1870 1890 1910 1930

NOTE: Higher inequality index indicates less inequality.

quality, did racism have an independent influence? Did differences in ethnicity matter? Did the political response to market events change as the working poor found their political power increasing?

Click to return to Table of Contents A SHLEY S. TIMMER / JEFFREY G. WILLIAMSON 751

Empirical tests

Since our interest is in long-run fundamentals, we smooth the policy index exponentially, and call the converted index polism.11 The converted index should do better in isolating the underlying fundamentals affecting policy in contrast with the timing of policy change. We estimate policy equations to identify the impact of market and nonmarket forces. Was policy driven by labor market conditions or by ethnic concerns? Was the reaction to a rising immigrant flood everywhere the same in the New World? With the policy equations estimated, we then use them to identify the sources of policy change.

Time series results

Overall, the empirical findings are quite satisfying. Table 1 reports results using a lagged dependent variable (that is, a lagged policy index for the country in question), while Table 2 reports results where the policies of other countries replace the lagged dependent variable. There are no variables in the tables that measure political environment. We cannot find evidence that changing political institutions and fran- chises systematically affected the degree and direction of policy change. We tried two measures of political openness: an index of democratic characteristics and a measure of competitiveness in political participation, both constructed by Ted Gurr (1990). Although these variables might be significant in explaining policy differences across countries, they are not important in explaining policy shifts within countries.12 This result is probably due to the fact that significant political change was minor in our time series. The most consistent effect emerging from Table 1 is that immigration policy is slow to change. The lagged dependent variable is highly significant in all countries. This is especially true of Brazil and the United States, but the result is driven by the 1888-1916 period in the United States and by the 1890–1920 period in Brazil. These two episodes of persistence were followed by a major switch in policy, from open to closed. Big policy switches often required long periods of debate. This was not always true, however, as can be seen by the major switch in Argentina’s policy over only five years, 1889–94. We have introduced variables with differing lags. Labor market and immigration variables were usually lagged two periods—an indicator of legislative delay—while economic conditions were taken as current. These measures of current macroeconomic conditions—growth in real GDP per capita (ypcgrr) and unemployment (unemp2)—did not prove consistently useful in accounting for policy change: macroeconomic conditions mattered in Australia, but not in Canada and the United States, and they took the wrong sign in Brazil. Thus, Australia offers the only evidence that these

Click to return to Table of Contents 752 I MMIGRATION POLICY PRIOR TO THE 1930S

TABLE 1 Explaining immigration policy using lagged dependent variables: Ordinary least squares; dependent variable is immigration policy, POLISM Argentina Brazil Australia Canada United States

Labor market effects Nominal wages/ GDP per capita 0.003 WTOY(–1) (0.547) Nominal wages/ GDP per capita 0.005* WTOY(–2) (1.773) Nominal wages/ GDP per capita 0.015** 0.005* WTOY(–4) (2.460) (1.973) Nominal wages 0.004* WAGEN(–2) (1.903) Real wages 0.007* WAGER(–2) (1.698) Change in real wage growth 1.407* D(WGRR(–2)) (1.677) Economic conditions Growth in real GDP per capita –1.044* 3.452** YPCGRR (–1.913) (2.565) Unemployment rate –0.034*** UNEMP2 (–3.104) Immigrant effects % foreign population –4.740** FORPOP(–2) (–2.425) Change in % foreign population –18.405* D(FORPOP(–2)) (–1.741) Average wages at immigrant origin 0.031** IMWAGE(–2) (2.468) Change in immigrant wages 0.028** 0.015** D(IMWAGE(–2)) (2.379) (2.005) Brazilian relative wages 0.011** BRWTOY(–1) (2.291) Brazilian real wage growth 0.862** BRWGRR (2.341) Lagged dependent variable POLISM(–1) 0.677*** 0.953*** 0.761*** 0.874*** 0.957*** (4.413) (24.592) (9.842) (17.582) (29.839) Constant –0.327 –1.116* –0.341 –4.018*** –0.704** (–0.530) (–1.686) (–1.088) (–2.958) (–2.138) No. of observations 54 68 70 57 70 Mean dependent variable 0.362 2.385 1.485 0.050 –1.649 R-squared 0.971 0.924 0.809 0.909 0.972 Adjusted R-squared 0.967 0.919 0.791 0.904 0.970 Log likelihood –27.430 –63.164 –52.495 –55.548 –6.249 Durbin–Watson 1.691 1.589 1.948 2.098 1.386 F-statistic 316.527 192.143 44.570 176.309 555.669

(t-statistics in parentheses; White-corrected standard errors) (*** significant at the .01 level; ** at the .05 level; * at the .1 level) NOTE: POLISM is the POLICY variable smoothed using exponential weights selected by TSP software.

Click to return to Table of Contents A SHLEY S. TIMMER / JEFFREY G. WILLIAMSON 753 macroeconomic conditions were critical in shaping the timing of policy change. Did labor market conditions have a consistent influence on immigration policy? And if so, was it the absolute or relative income performance of unskilled workers that mattered? It appears to have been both. The change in real wage growth of the urban unskilled (dwgrr) mattered most in the United States, nominal wages (wagen) mattered most in Australia, while real wage levels (wager) mattered most in Brazil.13 But apart from the lagged dependent variable, the most significant influence on policy is the ratio of the unskilled wage to per capita income, or of income near the bottom of the distribution to the average (wtoy). This measure of unskilled labor’s relative economic position stands up as an important influence on policy in the United States, Canada, and Brazil, regardless of what else is included in the equation—including real wage growth, real wage levels, unemployment, and attributes of the immigrants. The variable is not significant for Argentina and Australia. But for the other countries, high unskilled wages relative to average income correlate with more open immigration policies, and the correlation is significant; greater relative scarcity of unskilled labor encouraged less restrictive policy; declining relative scarcity of unskilled labor encouraged more restrictive policy. So far, we have looked at the indirect impact of immigration on policy by exploring labor market performance.14 Perhaps the size and character of the current and expected future immigrant flow precipitated policy change, the latter serving to anticipate the labor market impact. Two variables measure these direct immigration effects. First, the quality, or human capital content, of the immigrants is proxied by the real wage of unskilled urban workers in the source countries (imwage). Changes in that proxy were important for Australia. Second, we use measures of both the current flow— the immigration rate (imrate)—and the cumulative stock—the share of foreign-born within the total population (forpop). The immigration rate never proved to be a helpful explanatory variable, but the share of foreign-born did matter for Argentina and Australia. Higher immigrant quality or rising immigrant quality tended to precipitate more open immigrant policy in Australia, Canada, and the United States. More to the point, low and falling immigrant quality precipitated immigrant restriction, even after controlling for other forces. To some extent, therefore, policy in these countries anticipated the impact of low-quality immigrants on unskilled wages and moved to eliminate such immigration. In addition, Argentina seems to have looked to the north across the Rio de la Plata to watch labor market conditions in Brazil, acting as if they knew those conditions could divert immigrants to or from Argentina’s borders, either by immigrant responses to these relative labor market conditions or by their responses to likely policy changes in Brazil. Thus, rising relative unskilled wages and rising

Click to return to Table of Contents 754 I MMIGRATION POLICY PRIOR TO THE 1930S absolute wages in Brazil tended to produce more open policy in Argentina. This result is consistent with the estimated policy spillovers reported in Table 2, results that we discuss below. We also explored the impact of a variable that measured the difference in ethnic composition between the current immigration flow and the population stock (gap), but it was never significant. The literature had led us to expect that a rising gap between the ethnic origins of the population stock and the new immigrants would erode commitments to free immigration. In Table 2 Brazil offers some weak support for this view, but the effect does not appear elsewhere. To what extent was a change in a country’s policy a reaction to policy changes abroad? The results appear in Table 2, where the lagged dependent variable is replaced by migration policy changes abroad. As expected, the United States—the New World immigration leader—was not responsive to competitors’ policies. Nor, for that matter, was Canada, a surprising result that seems to confirm Canadian success in shielding its labor market from the eastern and southern European exodus to North America. For the other countries, policy abroad mattered greatly. For Argentina, it was the combined impact of Australian, Canadian, and Brazilian policy that mattered, more restrictive policy abroad inducing more restrictive policy at home. Brazil tended to mimic the policies followed in Argentina and the United States, although it also exhibited that puzzling inverse response to policy change found in Australia and Canada.15 Australia, in turn, was more likely to favor open immigration policies when the United Kingdom offered more generous subsidies to its emigrants, and also when the United States was more open. While the size of the immigrant flow did not have any consistent impact on New World policy up to 1930, its low and declining quality certainly did, provoking restriction. Racism or xenophobia does not seem to have been at work. Rather, immigrant quality, labor market conditions, and policies abroad—especially those set by the economic leaders, Britain and the United States—mattered most. New World countries acted in an effort to defend the economic interests of domestic unskilled labor.

Policy spillovers on immigrant destination by quantity and quality

Table 3 elaborates on the issues already raised in Table 2, namely, how policies adopted in one part of the New World influenced immigrant flows to other parts of the New World. The dependent variable in Table 3 is a given country’s share of that year’s five-country immigration flow. As we might have expected from Table 2, policies abroad hardly mattered at all for the United States. That is, the US share of immigration did not depend

Click to return to Table of Contents A SHLEY S. TIMMER / JEFFREY G. WILLIAMSON 755

TABLE 2 Explaining immigration policy using cross-country policy responses: Ordinary least squares; dependent variable is immigration policy, POLISM Argentina Brazil Australia Canada United States

Labor market effects Real wages –0.034*** 0.059*** WAGER(–2) (–4.246) (10.260) Nominal wages 0.031*** WAGEN(–2) (5.618) Nominal wages/GDP per capita 0.017* –0.024*** WTOY(–2) (1.941) (–3.940) Nominal wages/GDP per capita 0.051*** 0.031*** WTOY(–4) (3.447) (3.810) Real wage growth 4.730 WGRR (1.390) Economic conditions Unemployment –0.061*** –0.045*** UNEMP2 (–4.738) (–2.885) Growth in GDP per capita 2.720* YPCGRR (1.797) Immigrant effects % foreign population –19.447*** FORPOP(–2) (–6.476) Immigrant–native ethnicity gap –4.146 GAP(–2) (–1.626) Skill-weighted immigration rate –0.649* THREAT(–2) (–1.687) Policy spillovers Argentine policy 1.038*** 0.002 0.213* ARPOLISM(–2) (8.301) (0.017) (1.711) Australian policy 0.480*** –0.960*** –0.022 –0.050 AUPOLISM(–2) (4.773) (–7.583) (–0.105) (–0.327) Brazilian policy 0.401*** 0.068 BRPOLISM(–2) (4.153) (0.654) Canadian policy 0.328*** –0.635*** –0.002 0.120 CAPOLISM(–2) (5.369) (–9.728) (–0.032) (1.055) US policy 0.531*** 0.681*** –0.120 USPOLISM(–2) (3.743) (5.614) (–0.360) British policy 0.186* –0.395 –0.107 UKPOLISM(–2) (1.848) (–0.891) (–0.475) Constant 4.262*** 1.499 –1.828*** –3.649* –5.285*** (6.484) (1.626) (-3.714) (-1.916) (-4.963) No. of observations 54 68 69 57 69 Mean dependent variable 0.278 2.385 1.480 0.050 –1.673 R-squared 0.913 0.815 0.626 0.769 0.635 Adjusted R-squared 0.901 0.794 0.590 0.731 0.593 Log likelihood –57.300 –93.498 –75.422 –82.013 –94.225 Durbin–Watson 0.991 0.943 0.877 0.878 0.242 F-statistic 81.735 37.790 17.309 20.029 15.129

(t-statistics in parentheses; White-corrected standard errors) (*** significant at the .01 level; ** at the .05 level; * at the .1 level)

Click to return to Table of Contents 756 I MMIGRATION POLICY PRIOR TO THE 1930S

TABLE 3 Policy spillovers on immigrant destination: Ordinary least squares; dependent variable is the share of immigrants arriving at destination (IMMPCT) Argentina Brazil Canada Australia United States

Policy spillovers Argentine policy 0.011 ARPOLISM (1.497) Australian policy –0.007* 0.005 AUPOLISM (–1.726) (0.390) Brazilian policy –0.003 –0.005** 0.004 BRPOLISM (–1.413) (–2.387) (0.491) Canadian policy –0.004* 0.004 CAPOLISM (–1.984) (0.516) US policy –0.010*** –0.005 –0.004 –0.003 USPOLISM (–2.991) (–1.422) (–1.150) (–1.493) Lagged dependent variable Immigrant share 0.537*** 0.711*** 0.744*** 0.638*** 0.632*** IMMPCT(–1) (5.443) (8.391) (8.125) (7.256) (6.469) Constant 0.033*** 0.017* 0.043*** 0.031*** 0.207*** (3.557) (1.808) (3.184) (2.911) (2.959) No. of observations 69 69 69 69 69 Mean dependent variable 0.102 0.086 0.108 0.065 0.639 R-squared 0.562 0.556 0.643 0.725 0.597 Adjusted R-squared 0.549 0.542 0.620 0.708 0.565 Log likelihood 134.302 119.532 140.170 166.176 78.201 Durbin–Watson 1.886 2.396 1.613 1.607 1.846 F-statistic 42.406 41.314 28.767 42.138 18.660

(t-statistics in parentheses) (*** significant at the .01 level; ** at the .05 level; * at the .1 level)

on the policies elsewhere. But for the other destinations, more open immigration policies abroad reduced one’s share of arrivals. For example, Australia’s openness decreased flows to Canada, Brazil’s pro-immigrant subsidies reduced flows to Australia, and Argentina saw an increased share of the immigrant pie as the United States closed its doors. The impact of policy spillovers is significant but small, due to the stickiness of both policy and immigration patterns. Table 3 explores the impact of policy abroad on the distribution of immigrant destinations, while Table 4 explores the impact on immigrant quality. The quality proxy is a weighted average of (urban) unskilled wages prevailing in the sending regions at various points in time. Table 4 shows that the lagged dependent variable and policy choices in other countries account for more than 80 percent of the variance in the quality measure. When US policy became more restrictive, immigrant quality rose everywhere else. Immigrants favored the United States, so that when the United

Click to return to Table of Contents A SHLEY S. TIMMER / JEFFREY G. WILLIAMSON 757

TABLE 4 Policy spillovers on immigrant quality: Ordinary least squares; dependent variable is the quality of immigrants (IMWAGE) Argentina Brazil Canada Australia United States

Policy spillovers Argentine policy 0.256 0.691 ARPOLICY (0.657) (1.645) Brazilian policy 0.435* 0.008 –1.114*** BRPOLICY (1.814) (0.030) (–2.955) Canadian policy –0.891** –1.125*** 0.135 –0.667** CAPOLICY (–2.977) (–2.948) (0.344) (–2.351) US policy –1.729*** –1.632** –3.123*** –1.251** USPOLICY (–3.071) (–2.536) (–3.734) (–2.007) Lagged dependent variable Average wages at immigrant origin 0.564*** 0.655*** 0.673*** 0.845*** 0.790*** IMWAGE(–1) (6.672) (7.679) (8.047) (14.548) (10.963) Constant 17.917*** 14.024*** 18.674*** 10.512*** 17.194*** (5.428) (3.901) (3.699) (2.645) (3.070) No. of observations 71 -69 71 71 71 Mean dependent variable 46.907 48.932 72.799 77.987 68.313 R-squared 0.891 0.888 0.867 0.883 0.853 Adjusted R-squared 0.885 0.881 0.859 0.878 0.847 Log likelihood –186.894 –191.023 –207.621 –213.240 –201.135 Durbin–Watson 1.754 1.881 1.573 1.683 2.082

(t-statistics in parentheses) (*** significant at the .01 level; ** at the .05 level; * at the .1 level)

States became more restrictive, the other countries got rejects of higher quality than they would have received otherwise. The same was true of restrictive Canadian policy, for Argentina, Brazil, and the United States. Even the United States found its immigrant quality falling as Brazil subsidized immigration more heavily: that is, as Brazil increased subsidies, the United States lost some high-quality immigrants. Similarly, the United States seemed to benefit from more restrictive Canadian policy (often taking the form of lower subsidies), just as Canada benefited from more restrictive US policy. Interestingly, Argentina may have benefited from aggressive subsidy programs in Brazil. Perhaps higher-quality immigrants, arriving in Brazil under subsidy plans, found conditions unsatisfactory and moved over the border to Argentina.

Quantifying the sources of policy change

Using the estimates from Table 1, Table 5 reports how much each variable contributed to closing the immigrant door. We identified for each country

Click to return to Table of Contents –0.646 32.3% +0.292 –11.7% –4.024 67.1% –1.298 64.9% –0.312 12.5% Argentina Australia Brazil Canada United States WTOY(–1)WTOY(–2) –0.196 4.4% –1.676 25.8% WTOY(–4) WAGER(–2)WAGEN(–2) +0.115 –2.6% –2.330 35.8% D(WGRR(–2)) YPCGRRUNEMP2 –0.737 16.4% –0.457 10.2% –0.357 5.5% FORPOP(–2)D(FORPOP(–2))IMWAGE(–2) –1.182 26.3% –0.004 0.09% –0.681 11.4% D(IMWAGE(–2))BRWTOY(–1)BRWGRR –3.292 73.2% –0.115 +0.111 2.6% –2.5% –1.719 86.0% –1.020 40.8% TABLE 5 Decomposing the sources of policy change PolicyTotal change –4.5 4.5 to 0 100% –4.5 2.5 to –2 100% –6.5 4.5 to –2 100% –6 1.5 to –4.5 100% –2 1 to –1 100% –2.5 –1 to –3.5 100% Time period 1888–98 1926–30 1917–27 1899–1919 1865–85 1885–1917 Attributable to: Labor market effects Nominal wages/GDP per capita Nominal wages/GDP per capita Nominal wages/GDP per capita Real wages Nominal wages Change in real wage growth Growth in GDP per capita Unemployment Immigrant effects Economic conditions % foreign population Change in foreign population Wages at immigrant origin Change in wages at immigrant origin Brazilian relative wages Growth in Brazilian real wages Residual +0.285 –6.3% –3.528 78.4% –2.137 32.9% –1.295 21.6% +1.663 –83.2% –1.46 58.4%

Click to return to Table of Contents A SHLEY S. TIMMER / JEFFREY G. WILLIAMSON 759 a period of major change toward more restrictive immigration policy. How much of the change was attributable to general economic conditions, to indirect labor market effects, to direct immigrant effects, and to other factors?16 When Brazil’s door slammed shut in the 1920s, over 60 percent of the 6.5-point drop in the policy index was attributable to deteriorating labor market conditions, a good share of which was rising inequality. Al- though the residual is large (over 30 percent), labor market forces still account for nearly two-thirds of this major policy switch from an open immigration policy with generous subsidies in 1917, to a restrictive policy in 1927. Canada offers even stronger evidence in support of the view that labor markets mattered. During the Prairie Boom from 1899 to 1919, the policy index dropped 6 points. Two-thirds of this drop can be attributed to rising inequality over those two decades (67 percent), and another tenth or so (11 percent) to diminished immigrant quality. The residual is only 22 percent. Between 1888 and 1898, the policy index for Argentina fell by 4.5 points. Indirect labor market effects at home apparently made only a modest contribution to this big policy change (4 percent). However, it could be argued that Argentina anticipated the likely labor market effects at home by watching labor market developments in Brazil. Rising inequality and deteriorating wage growth in Brazil account for three-quarters of Argentina’s policy switch. Increasing foreign presence in Argentina accounts for an additional quarter of the policy switch (26 percent). Between 1865 and 1885, the immigration policy index for the United States dropped by 2 points. Almost all of that drop can be attributed to labor market effects and deteriorating income conditions of the unskilled. Direct immigrant effects mattered almost as much, captured here by declining quality. We have no explanation for the offsetting residual. In contrast with the powerful labor market effects apparent between 1865 and 1885, almost none of the 2.5-point drop between 1885 and 1917 can be assigned to labor market conditions (1 percent). Thus, Goldin (1994) was right in attributing the passage of the immigrant literacy test largely to nonmarket factors. That is, the residual is very large during this period, confirming the views of American historians who stress nonmarket forces. Note, however, that deteriorating immigrant quality accounts for four-tenths of the move to restriction in the United States during the period (41 percent). The estimated equations do not explain nearly as much of the Aus- tralian switch to more restrictive policy during the late 1920s. The Austra- lian residual is by far the largest in Table 5 (78 percent). We can offer no explanation for the finding, except to argue that many of the variables may have already been affecting the political scene even though policy remained unchanged prior to 1926. The time period is the shortest in the table.

Click to return to Table of Contents 760 I MMIGRATION POLICY PRIOR TO THE 1930S

Conclusions

Our results point to long-run fundamentals driving immigration policy that are very different from the short-term influences on timing about which so much has been written. We find little support for the conventional wis- dom that current macroeconomic conditions, as measured by growth and unemployment, had a consistent influence on policy change. Although there is some evidence that policy was sensitive to the “quality” of immigrants— especially in the non-Latin countries, there is no evidence of the influence of racism or xenophobia, once underlying economic variables are taken into account. Income distribution trends seem to have been especially important for the United States and Canada, both of which tried to protect the economic position of their unskilled workers. Labor became relatively more abundant when immigrants poured in; and governments sought to stop any absolute decline in the wages of the domestic unskilled with whom the immigrants competed, and often even a decline in their wages relative to the average income recipient. The greater the perceived threat to these wages from more immigrants or from lower-quality immigrants, the more restrictive policy became. Meanwhile, Australia was paying attention to unemployment, growth, and nominal wages, and may have reacted to protect the relative position of workers vis-à-vis landed interests (Timmer and Williamson 1996). Immigration policy seems to have been influenced indirectly by labor market conditions, and directly by immigration forces that, if left to run their course, would have had their impact on labor market conditions. The switch to more restrictive policies was less the result of rising immigrant presence and more the result of falling immigrant quality. But countries did not act in isolation. Domestic policy was correlated with policy elsewhere, and rationally so: except for the United States, countries saw both the quantity and quality of immigration respond to the policies of others, so it is hardly surprising that the door-closing was reactive. (Likewise, it is not surprising that the United States did not react to policy changes by others.) These policy correlations may well have to do more with the change in immigration flows than with any preemptive policy measure. These results offer lessons for contemporary debates about immigration. The parallels are clear. Inequality has been on the rise in the Euro- pean economies since the early 1970s, manifested especially by a rising income gap between unskilled and skilled workers, just as it was in the New World economies in the late nineteenth century. We should therefore not be surprised by the renewed interest, in both the United States and Europe, in reducing the migrant flow. Labor-scarce economies have been sensitive in the past to inequality trends in their midst, using restric-

Click to return to Table of Contents A SHLEY S. TIMMER / JEFFREY G. WILLIAMSON 761 tive immigration policy to offset, or at least to dampen, those trends. If history repeats itself, policies will become increasingly anti-immigrant, at least as long as the relative position of unskilled workers lags behind that of other economic groups.

Appendix A: Independent variables

Economic variables Variables are lowercased and italicized when used in text. Sources can be found in Timmer and Williamson (1996: Appendix A).

Population (POP) Reported as actual estimated population. Nominal GDP (GDPN) All series have been converted to indexes, with 1900=100. Real GDP (GDPR) All series have been converted to indexes, with 1900=100. Nominal wages (WAGEN) All series have been converted to indexes, with 1900=100. Unless otherwise noted, the series are wage rates for urban unskilled workers. Real wages (WAGER) All series have been converted to indexes, with 1900=100. Unless otherwise noted, the series are real wage rates for urban unskilled workers. Land values (LANDV) Nominal estimates. Missing years are estimated by linear interpolation. Export (X) and import Current-dollar estimates of merchandise export (M) values and import values.

Growth in real wages (WGRR) Calculated as (WAGERt+1– WAGERt)/(WAGERt) Wages relative to income Calculated as WAGEN/GDPN/POP, indexed to (WTOY) 1900=100. Wages relative to land values (WTOR) WAGEN/LANDV, indexed to 1901=100.

Per capita growth in real GDP Calculated as (GDPRt+1/POPt+1– GDPRt/POPt)/ (YPCGRR) (GDPRt/POPt) Unemployment (UNEMP) Estimated by regressing GDPN on time and time squared, and taking the negative of the residuals. Unemployment (UNEMP2) Estimated by regressing GDPR on time and time squared, and taking the negative of the residuals. Trade share of GDP (XMTOY) Calculated as the total nominal value of exports plus imports, divided by nominal GDP.

Immigration variables Most of the immigration data were assembled from: Ferenczi and Willcox (1929, 1930); more detailed sources defending our revision of these data are found in Timmer and Williamson (1996: Appendix A).

Click to return to Table of Contents 762 I MMIGRATION POLICY PRIOR TO THE 1930S

The following regional geographic groupings were used: Southern Europe Greece, Italy, Portugal, Spain Northern Europe Belgium, Denmark, Finland, France, Germany, Netherlands, Norway, Sweden, Switzerland United Kingdom England, Ireland, Scotland, Wales Eastern Europe Albania, Austria, Bulgaria, Czechoslovakia, Es- tonia, Hungary, Latvia, Lithuania, Poland, Ro- mania, Russia, Turkey, Yugoslavia Asia China, Hong Kong, India, Japan, and others in present-day East Asia, South Asia, Southeast Asia, and the Pacific Islands

Immigration rate (IMRATE): Calculated as total immigration divided by total population. Average wages at origin (IMWAGE): Measures the average quality of the immigrant, at least as implied by the unskilled wages prevailing in sending countries. For each country, immigration flows were grouped into regions of origin, and the percentage of immigration from each region was calculated. For each region, an annual series of wages was constructed using Williamson’s (1995) internationally comparable series, which are purchasing-power-parity adjusted: United Kingdom uses the wage series for Great Britain; Northern Europe uses the series for the Nether- lands; Southern Europe uses the wages for Portugal from 1850 to 1870, and, from 1870 to 1930, Italian wages were used but scaled such that the 1870 level of purchasing power matched that of Portugal for the same year (a correction of less than 10 percent); Eastern Europe and “other miscellaneous origin” wages were estimated to be two-thirds of those in Southern Europe; Asian wages were estimated to be half the level of Southern Europe. The variable simply calculates a weighted average of these wages, using the percentage of immigration from each region as the weight. The following groupings were used: Australia United Kingdom, Northern Europe, Southern Europe, Eastern Europe, Asia, and other Argentina United Kingdom, Northern Europe, Southern Europe, Eastern Europe, and other Brazil Northern Europe (includes in this case United Kingdom), Southern Europe, Eastern Europe, and other Canada United Kingdom, Northern Europe, United States (assigned UK wages), Eastern Europe, and other United States United Kingdom, Northern Europe, Southern Europe, Eastern Europe, Asia, and other

Immigrant wages relative to destination (IMWREL): Like IMWAGE, this variable also captures immigrant quality, but in this case relative to the receiving region. It was calculated in much the same way as IMWAGE, except that, in addition, it measures wages in regions of emigration relative to wages in the country of destination. Wage threat from immigration (THREAT): This variable was calculated to measure the extent to which immigration reflected “unfair competition from cheap foreign

Click to return to Table of Contents A SHLEY S. TIMMER / JEFFREY G. WILLIAMSON 763 labor,” that is, a threat to unskilled resident labor. Calculated to interact immigration rates with relative immigrant quality: THREAT = (100 – IMWREL)*IMRATE. Low IMWREL and high IMRATE implies big threat and large positive THREAT. Percent foreign population (FORPOP): For most countries, the foreign-born population is counted every ten years in the census. Using immigration data cited above, and in some cases emigration data, the between-census years are estimated. These estimates are divided by the total population estimates to calculate the percent who are foreign. Difference in regional stocks and flows (GAP): Using the annual composition of immigration (grouped as in FORPOP) and the annual composition of the foreign population (as estimated for FORPOP), an index was constructed to measure a shift in the composition of immigration relative to the current foreign-born population. For each year and for each group the difference between the percentage of immigrants and the percentage of foreign born was squared, and all groups except “other” were then summed. The index has a minimum value of zero, if the immigration flow looks just like the current foreign population. The theoretical maximum value is 1.

Appendix B: On constructing the dependent variable: The example of the United States

The text described how the index of immigration policy is constructed, the index or score (POLICY) ranging from +5, a policy of generous subsidy and non-discrimination upon arrival, to –5, a policy of serious immigration restriction, effectively enforced. This appendix illustrates the index for the United States between 1860 and 1930. The dependent variable for other countries is described at length in Timmer and Williamson (1996, Appendix C). The value of POLICY is reported only for years when it changed.

United States immigration policy, 1860–193017

Pre-1860 Prior to 1840, most policy was set by individual states. Some restricted the entrance of paupers and criminals, or imposed head taxes to pay for immigrant services. Naturalization was allowed after five years of residence. From 1847 to 1849, the first effective legislation regulating passenger ships was enacted, requiring 14 square feet of clear deck space per passenger and adequate ventilation and food supplies. In 1849, the Supreme Court ruled that the state policies of head taxes and bonding were unconstitutional, leaving no funds to pay for lodging and health services provided to immigrants. In 1855, the individual passenger acts were consolidated and recodified to strengthen the health and safety regulations. Also in 1855, wives and foreign-born children of citizens were granted automatic citizenship. 1860 Passenger Acts are amended to protect female passengers from “se- duction by ship personnel.” POLICY=0 1862 Congress bans trade by US vessels in coolie, or indentured, labor.

Click to return to Table of Contents 764 I MMIGRATION POLICY PRIOR TO THE 1930S

1864 Commission of Immigration Office is established with a budget of $20,000/year for publishing and distributing recruiting literature. Congress legalizes indentured labor contracts of less than one year for payment of passage. POLICY=+1.0 1865 Congress fine-tunes the steamship regulations. 1866 Congress issues a formal protest to European governments against the deportation of criminals to the United States. 1868 Congress repeals the labor-contract provision of the 1864 act. POLICY=0 1869 Laws against the coolie trade are strengthened, notably making it illegal to transport individuals under fraudulent claims to induce emigration. 1870 Responding to concern that there were insufficient safeguards in the naturalization process, Congress tightens the regulations and puts checks into place. The act extends the right of naturalization to those of African descent. 1871 Recognizing deficiencies in the law, Congress reworks the passenger acts, without substantive change. 1875 The Immigration Act establishes the notion of “excludable” classes. The Act prohibits the importation of Chinese women for “immoral purposes” (prostitution) and bringing in persons without their con- sent; makes contracting to supply coolie labor a felony; designates criminals as an excludable class, but specifies that this does not include political offenses or those who received pardons in return for leaving their country of origin. POLICY=–0.5 1876 Congress requires a “declaration of intent” prior to naturalization (i.e., filling out paperwork.) 1880 The United States negotiates a treaty with China, recognizing the right of the United States to regulate, limit, or suspend Chinese immigration, but not the right to prohibit it. 1882 The Passenger Acts are completely revised, detailing the required deck space, food portions, water, and ventilation. New classes are added to the list of excludables: paupers, convicts, persons suffering from “mental alienation,” lunatics, and idiots. A head tax of $0.50/immigrant is imposed to defray the costs of administration. Congress establishes the first legal terms for deportation, by legislating that convicts will be returned to their country of origin. Chinese immigration is suspended for ten years, with a provision to deport illegal Chinese residents. Congress instructs the courts that they are to disallow citizenship for the Chinese. POLICY=–1.0 1884 Congress amends the Chinese immigration suspension law to require evidence from legal entrants of belonging to an allowed group (mer- chants and travelers). Congress clarifies that the law applies to all

Click to return to Table of Contents A SHLEY S. TIMMER / JEFFREY G. WILLIAMSON 765

Chinese, regardless of country of origin. Carriers between the United States and Mexico or Canada are exempted from the head tax, while the tax is imposed on those who come by land as well as by ship. 1885 The Alien Contract Labor Act makes it illegal to prepay an individual’s voyage in return for labor services; voids all existing contracts made prior to immigration; establishes penalties for violators. The Act ex- empts diplomats and other foreigners temporarily in the United States who bring over staff, specialty labor, domestic servants, and certain professional groups. 1887 The Contract Labor Law clarifies the enforcement mechanism of the 1885 Act, and provides that prohibited workers would be sent home. Congress passes a law banning any noncitizen from owning real estate and prohibiting more than 20 percent foreign-held ownership of a corporation, unless the individuals had properly declared their intent to become citizens. POLICY=–1.5 1888 The Chinese Exclusion Act suspends all Chinese immigration for 20 years (with student/diplomatic exemptions) and establishes the rules of deportation and fines for violators. For the first time, the law allows for the imprisonment of those who are in the United States unlawfully. (The suspension part of the Act was later found null after failure to ratify the treaty, although the 1882 ban remains in effect.) Congress makes it illegal for Chinese residents to return to the United States if they leave (even if here legally), and stops issuing identity certificates, which had functioned as passports. Alien land- ownership laws are amended to allow governments to set up their attachés in Washington, DC. Congress authorizes funds for finding and deporting illegal contract labor. 1891 The Immigration Act adds new groups to the list of excludable classes: those “likely to become public charges,” polygamists, those suffering from contagious and dangerous diseases, and anyone “assisted” in passage. The Act bans all advertising for the purpose of encouraging immigration, except by offices of the states. Also, the Act extends the exemptions from the contract labor law to include professors, professionals, and ministers, while adding to those prohibited contracts with family or friends. 1892 The Chinese Exclusion Act extends the ban on immigration for another ten years, requires legal Chinese to file for a residency certificate within one year, and provides for the deportation of those who do not have their certificates within that year, unless “at least one credible white witness” can attest to their difficulty in obtaining the certificate. 1893 Quarantine act allows the President to restrict or suspend immigration in response to contagious disease threats in foreign countries. Congress reworks some of the procedures to help enforce existing laws. The Chinese Exclusion Act is amended to strengthen its en-

Click to return to Table of Contents 766 I MMIGRATION POLICY PRIOR TO THE 1930S

forcement, and to allow any non-Chinese witness in place of the white witness. 1895 Head tax is raised to $1/immigrant 1898 Congress sets up a commission to examine the effects of immigration on labor and industry, to report back to Congress with advice for handling immigration. 1902 The Chinese Exclusion Act extends the ban for another ten years. Essentially, it is the 1892 law reissued. 1903 The Immigration Act raises the head tax to $2. It also adds to the list of excludable classes: professional beggars, epileptics, the insane, prostitutes, and anarchists or others endorsing the overthrow of foreign governments. The Act also extends the period of deportability to two years from admission. 1904 Immigrants from Newfoundland are exempted from the head tax. Congress extends the ban on Chinese immigration to all US islands and territories. 1907 The Immigration Act raises the head tax to $4, except for arrivals from Mexico, Canada, Newfoundland, and Cuba. It also restricts entry of those who were granted a passport for a different destination. The Act adds more classes to the list of excludables: unaccompanied minors, “induced” immigrants, and the disabled. The Act establishes a financial test, so that each individual must have $25, or $50 per family, the first such requirement on immigrants. Congress sets up another commission to study immigration. POLICY=–2.0 1909 Canada and Mexico are exempted from having to produce mani- fests of their alien arrivals. 1910 The White Slave Traffic Act expands deportation statutes and laws on prostitution offenses to include any alien (i.e., any foreign men involved can be prosecuted as well as women), and to extend the period of deportability indefinitely. 1917 The Immigration Act establishes a literacy test for immigrants, to be given in any language. Failure to demonstrate literacy will be grounds for denial of admission, although certain groups are exempted. Act adds to the classes of excludables those of “constitutional psycho- pathic inferiority,” a jargon phrase that was also used in Canadian legislation. It is interpreted to mean those who will fail to assimilate. The Act also defines a zone in Asia (actually most of Asia) from which individuals would be ineligible for citizenship through naturalization. Immigration is banned for those who would not be eligible for citizenship through naturalization. Thus, all immigration of Asians is effectively banned. The Act also doubles the head tax to $8. POLICY=–3.5 1918 Congress strengthens the ban on anarchists and other political

Click to return to Table of Contents A SHLEY S. TIMMER / JEFFREY G. WILLIAMSON 767

troublemakers, and also agrees to readmit certain aliens who served in the military for the United States or its allies during World War I. 1919 Congress gives the President temporary powers to make any necessary rules/prohibitions on alien entry in order to protect the public safety. (Power expired on 4 March 1921.) POLICY=–4.0 1920 Congress establishes a five-year window of opportunity to allow admission to those who cannot read, if they are going to marry someone who fought in the war, even if he is an alien (war brides). Con- gress passes rules to deal with alien activists. It allows the deportation of those “interned as dangerous but not actually convicted of any crime.” It also extends the definition of anarchist to include those associated with antigovernment groups, publications, or organizations affiliated with the publications. POLICY=–3.5 1921 Quotas are established to restrict the quantity of immigration from any one country to 3 percent of its population in the United States in 1910, for one year. The ban on all Asian immigration remains in effect, while all immigration from the Western Hemisphere is free from restriction. To keep Canada and Mexico from being through- ways to the United States, immigrants from the Western Hemisphere have to have been in those countries for one year before qualifying for quota-free admission. This law, the Emergency Quota Act, was originally a temporary measure, expired in 1922. POLICY=–4.5 1922 Act extends the 1921 Act until 1924, and extends the Western Hemi- sphere residency period to five years. Establishes a $200 fine for bringing an illegal immigrant, and allows certain aliens brought in over quota to remain. 1924 1921 Act is amended to use quotas of 2 percent of a country’s population, using 1890 as the base year (thus further restricting the “new” immigrants). Establishes that, as of July 1927, the quota will be 150,000 total, in the same proportion as the “national origin” of the US population in 1920, excluding from the count immigrants brought against their will (i.e., former slaves do not count toward Africa’s quota). The Act establishes that wives and children under 18 have non-quota status, as do natives of the Western Hemisphere, ministers, professors, and students. Quota preference is given to children of citizens under 21, parents, spouses, and those trained in agriculture. 1926 Congress admits wives and children under 18, and professors, who were in the United States prior to 1924. The use of a “national origins” system is postponed until 1928. 1928 “National origins” system is postponed until 1929. Women who were US citizens but who gave up such status by marrying a foreigner are admitted if they are unmarried. Congress establishes that one-half of the quotas will be reserved for the preferred classes—wives and children, parents, agricultural workers. Clarifies that American Indi-

Click to return to Table of Contents 768 I MMIGRATION POLICY PRIOR TO THE 1930S

ans may travel freely across borders without immigration restrictions, as long as they are not part of a tribe by adoption. 1929 The Deportations Act makes it a felony to return to the United States if deported and a felony or misdemeanor to enter the country at an unauthorized point. Also establishes that those punishable will first be imprisoned, then deported after serving their sentence. National Origins Act takes effect 1 July. POLICY=–5.0

Appendix C: Decomposing policy changes

We use the estimated equations from Table 1 to construct the decomposition in Table 5. After the change in each of the right-hand side variables is calculated, it is multiplied by the estimated coefficient. Then we calculate their multiplicative impact through the lagged dependent variable. Consider the following example. We have a six year period, 1925–30. Each variable contributes contemporaneously, but also will have its share in the lagged dependent variable. Suppose we have the following equation:

POLICY = C0 + C1* POLICY (–1) + C2 * WTOY(–2) The change in policy is the difference in the index from 1925 to 1930. Then we calculate how much of that change is due to changes in WTOY(–2) from 1925 to 1930 as the sum of all the following components:

A = {WTOY(1924) – WTOY(1923)}*C2 ,

B = {WTOY(1925) – WTOY(1924)}*C2 ,A*C1 ,

C = {WTOY(1926) – WTOY(1925)}*C2 ,B*C1 ,A*C1*C1 ,

D = {WTOY(1927) – WTOY(1926)}*C2 ,C*C1 ,B*C1*C1 ,A*C1*C1*C1 ,

E = {WTOY(1928) – WTOY(1927)}*C2 ,D*C1 ,C*C1*C1 ,B*C1*C1*C1 ,A*C1*C1*C1*C1 Note that this method does not consider the impact of previous changes to WTOY that are still playing themselves out slowly through the lagged dependent variable. It is not clear whether this means we are underestimating the effects, since the equations themselves omit variables that may have been significant for certain time periods, but were not statistically significant in the regressions using the entire time series.

Notes

This is a much-revised version of “Racism, the National Science Foundation (grants SES xenophobia or markets? The political 92-23002 and SBR 9505656), the technical economy of immigration policy prior to the help of Timothy Hatton and Alan Taylor, and Thirties,” NBER Working Paper No. 5867, the research assistance of Spyros Poulios. In National Bureau of Economic Research, addition, the authors have benefited by com- Cambridge, MA (December 1996). The au- ments from William Collins, James Fore- thors gratefully acknowledge the support of man-Peck, Timothy Hatton, Dani Rodrik,

Click to return to Table of Contents A SHLEY S. TIMMER / JEFFREY G. WILLIAMSON 769

Peter Timmer, and the participants in the that reason, Benhabib does not test the model Harvard Economic History lunch seminar, empirically. the Harvard Economic History Workshop, 9 See, however, Perotti (1996) for a the Harvard–MIT Research Training Group comprehensive review of the competing hy- in Positive Political Economy, and the MIT potheses. Seminar on Migration. 10 About 60 percent of the total emigra- 1 Liberalization also helped induce eco- tion out of Europe was to the United States nomic convergence within the greater Atlan- (Hatton and Williamson 1998: Ch. 2), and tic economy (Williamson 1996). about 70 percent of the total emigration to 2 The exceptions are surveyed in our our five-country New World sample was to previous article (Timmer and Williamson the United States. 1995). 11 The smoothing function uses expo- 3 After World War II, a focus on human nentially decreasing weights, as selected by rights developed; most Western countries TSP software. changed their immigration policies to pro- 12 In Timmer and Williamson (1996), vide special consideration for political and we constructed a panel data set and used the economic refugees. Prior to the 1930s, such political variables to help explain policy lev- classifications did not exist. els. However, only fixed-effect estimation 4 The world labor market was by 1890 was possible since the economic data are pre- almost completely segmented into what sented in a form where they are indexed to economists today would call “North” and 1900. “South” (Lewis 1978; Taylor 1994; Hatton 13 These results do not, of course, speak and Williamson 1994b), and these new im- to the issue of whether immigration had an migrant flows were from the “South.” impact on wages. Indeed, we know that it 5 This, it turns out, is a reasonable as- did (Taylor and Williamson 1997; Hatton and sumption by the 1890s, at least for United Williamson 1998). However, policy changes States manufacturing (Hanes 1993, 1996). usually did not have a large enough impact 6 Immigrants did it even better, of on immigration to matter much for wages. course. A policy of immigrant exclusion 14 We also constructed a variable that at- would have done no better than to reduce tempted to measure the threat to native wages, the net inflow to zero. Voluntary return by dividing the immigration rate by the aver- migration drove up out-migration rates to lev- age wages in the countries of origin (THREAT). els high enough to make net inflows negative. Thus, the variable increases with the volume 7 The same is true in many developing of immigration and with declining immigrant countries today, where agriculture is a fifth, skills. It was found not significant. a quarter, or even a third of the economy. 15 This inverse response may well be In such countries, rural wage employment due to collinearity among the policy indexes. is important and landed interests are pow- 16 We do not measure the impact of past erful. performance on the lagged dependent variable, 8 Jess Benhabib (1996) takes the me- although we do multiply through the changes dian-voter approach, allowing individuals to in the explanatory variables as they play out earn both labor and capital income in the slowly within the period. Appendix C details spirit of the growth model of Alesina and the methodology of the calculations. Rodrik (1994); voters determine the amount 17 See E. P. Hutchinson, Legislative His- of capital that immigrants must bring with tory of American Immigration Policy, 1798–1965 them in order to be admitted. The model, an (Philadelphia: University of Pennsylvania attempt to look at the dynamics of policy Press, 1981). implications, gets very complicated. Perhaps for

Click to return to Table of Contents 770 I MMIGRATION POLICY PRIOR TO THE 1930S

References

Alesina, Alberto and Dani Rodrik. 1994. “Distributive politics and economic growth,” Quar- terly Journal of Economics 109, no. 2: 465–490. Anderson, James and J. Peter Neary. 1994. “Measuring the restrictiveness of trade policy,” The World Bank Economic Review 8, no. 2: 151–169. Benhabib, Jess. 1996. “On the political economy of immigration,” European Economic Review 40, no. 9: 1737–1743. Black, Duncan. 1948. “On the rationale of group decision making,” Journal of Political Economy 56, no. 1: 23–34. Borjas, George. 1994. “The economics of immigration,” Journal of Economic Literature 32, no. 4: 1667–1717. Cheng, Leonard K. and Kar-Yiu Wong. 1990. “On the strategic choice between capital and labor mobility,” Journal of International Economics 28, nos. 3/4: 291–314. Collins, William J., Kevin H. O’Rourke, and Jeffrey G. Williamson. 1997. “Trade and factor mobility: Complements or substitutes in history?” Paper presented at the Council on Eco- nomic Policy Reform Conference on Trade and Factor Mobility, Venice, 23–26 January. Ferenczi, Imre and Walter Willcox. 1929. International Migrations: Volume I—Statistics. New York: National Bureau of Economic Research. ———. 1930. International Migrations: Volume II—Interpretations. New York: National Bureau of Economic Research. Flam, Henry and M. June Flanders. 1991. Heckscher-Ohlin Trade Theory. Cambridge: MIT Press. Foreman-Peck, James. 1992. “A political economy model of international migration, 1815– 1914,” The Manchester School 60, no. 4: 359–376. Freeman, Gary P. 1992. “Migration policy and politics in the receiving states,” International Migration Review 26, no. 4: 1144–1167. Freeman, Richard. 1995. “Are your wages set in Beijing?” Journal of Economic Perspectives 9, no. 3: 15–32. Goldin, Claudia. 1994. “The political economy of immigration restriction in the U.S., 1890 to 1921,” in Claudia Goldin and Gary Libecap (eds.), The Regulated Economy: A Histori- cal Approach to Political Economy. Chicago: University of Chicago Press. Green, Alan G. 1994. “International migration and the evolution of prairie labor markets in Canada, 1900–1930,” in Hatton and Williamson 1994a. ———. 1995. “A comparison of Canadian and US immigration policy in the twentieth century,” in Don J. DeVoretz (ed.), Diminishing Returns: The Economics of Canada’s Recent Immigration Policy. Toronto: C. D. Howe Institute. Gurr, Ted R. 1990. Polity II: Political Structures and Regime Change, 1800–1986, Inter-university Consortium for Political and Social Research, data set #9263. Ann Arbor: ICPSR. Hanes, Christopher. 1993. “The development of nominal wage rigidity in the late nineteenth century,” American Economic Review 83, no. 4: 732–756. ———. 1996. “Changes in the cyclical behavior of real wage rates, 1870–1990,” Journal of Economic History 56, no. 4: 837–861. Hatton, Timothy J. and Jeffrey G. Williamson (eds.). 1994a. Migration and the International Labor Market, 1850–1939. London: Routledge. ———. 1994b. “Late-comers to mass emigration: The Latin experience,” in Hatton and Williamson 1994a. ———. 1995. “The impact of immigration on American labor markets prior to the quotas,” NBER Working Paper No. 5185. National Bureau of Economic Research, Cambridge, MA. ———. 1998. The Age of Mass Migration. New York: Oxford University Press. Hillman, Arye L. 1989. The Political Economy of Protection. New York: Harwood Academic Publishers.

Click to return to Table of Contents

ASHLEY S. TIMMER / JEFFREY G. WILLIAMSON771

Lewis, W. Arthur. 1978. The Evolution of the International Economic Order. Princeton: Princeton University Press. Luttmer, Erzo. 1997. “Group loyalty and the taste for redistribution,” mimeo., Harvard Uni- versity. Markusen, James R. 1983. “Factor movements and commodity trade as complements,” Jour- nal of International Economics 14, nos. 3/4: 341–356. Perotti, Roberto. 1992. “Income distribution, politics, and growth,” American Economic Re- view 82, no. 2: 311–316. ———. 1993. “Political equilibrium, income distribution, and growth,” Review of Economic Studies 60, no. 205: 755–776. ———. 1996. “Growth, income distribution, and democracy: What the data say,” Journal of Economic Growth 1, no. 2: 149–187. Pope, David and Glenn Withers. 1994. “Wage effects of immigration in late-nineteenth- century Australia,” in Hatton and Williamson 1994a. Sachs, Jeffrey D. and Andrew Warner. 1995. “Economic reform and the process of global integration,” Brookings Papers on Economic Activity 1. Shughart, William, Robert Tollison, and Mwangi Kimenyi. 1986. “The political economy of immigration restrictions,” Yale Journal on Regulation 51, no. 4. Stolper, Wolfgang and Paul A. Samuelson. 1941. “Protection and real wages,” Review of Economic Studies 9, no. 1: 58–73. Taylor, Alan M. 1994. “Mass migration to distant southern shores,” in Hatton and Williamson 1994a. Taylor, Alan M. and Jeffrey G. Williamson. 1997. “Convergence in the age of mass migration,” European Review of Economic History 1, no. 1: 27–63. Timmer, Ashley S. 1996. “Using politics to keep up with the Joneses: United States immigration policy and relative incomes,” paper presented to the Research Training Group in Positive Political Economy, Harvard University, April. Timmer, Ashley S. and Jeffrey G. Williamson. 1995. “The political economy of migration policies,” mimeo., Harvard University. ———. 1996. “Racism, xenophobia or markets? The political economy of immigration policy prior to the Thirties,” NBER Working Paper No. 5867, National Bureau of Economic Research, Cambridge, MA. Venables, Anthony J. 1997. “Trade liberalization and factor mobility: An overview,” paper presented at the Council on Economic Policy Reform Conference on Trade and Factor Mobil- ity, Venice, 23–26 January. Verdier, Thierry. 1994. “Models of political economy of growth: A short survey,” European Economic Review 38, nos. 3/4: 757–763. Williamson, Jeffrey G. 1974. “Migration to the New World: Long term influences and impact,” Explorations in Economic History 11, no. 4: 357–389. ———. 1990. Coping with City Growth During the British Industrial Revolution. Cambridge: Cam- bridge University Press. ———. 1995. “The evolution of global labor markets since 1830: Background evidence and hypotheses,” Explorations in Economic History 32, no. 2: 141–196. ———. 1996. “Globalization, convergence and history,” Journal of Economic History 56, no. 2: 277–306. ———. 1997. “Globalization and inequality, past and present,” The World Bank Research Ob- server 12, no. 2: 117–135. Wood, Adrian. 1994. North-South Trade, Employment and Inequality: Changing Fortunes in a Skill-Driven World. Oxford: Clarendon Press. Wong, Kar-Yiu. 1983. “On choosing among trade in goods and international capital and labor mobility: A theoretical analysis,” Journal of International Economics 14, nos. 3/4: 223–250.

Click to print article ClickClick to return to return to Table to of Table Contents of Contents Click to return to Table of Contents

The Onset of Fertility Transition in Pakistan

ZEBA A. SATHAR JOHN B. CASTERLINE

FERTILITY IN PAKISTAN has shown a stubborn resistance to change. Because of sharp declines in mortality following World War II, the population of Pakistan was growing at the rate of 2.7 percent per annum around 1960. In response to concern about rapid growth, a national policy of slowing population growth was articulated in the 1960s, with a program of family planning services as the main tool (Government of Pakistan 1965). During its first two decades, however, the program appeared to have had scant impact on fertility: the total fertility rate (TFR) continued to hover between six and seven births per woman throughout the 1970s and 1980s, and the population growth rate approached 3 percent per annum (Sathar 1993). Repeated assertions that a decline in marital fertility was underway proved baseless.1 In this article we present empirical evidence from multiple and independent studies carried out in the past eight years demonstrating that the decline of marital fertility has finally begun in Pakistan. The decline is gentle but nevertheless represents a genuine break from the past, most notably because of the increasing use of modern contraception for the purpose of limiting family size. The slight declines in overall fertility prior to 1990 were attributable almost entirely to increases in age at first marriage (United Nations 1993). With a population of 130.5 million in 1998 (Population Census Or- ganization 1998), Pakistan is the world’s seventh most populous country. According to UN projections, it will become the fourth most populous by the year 2050. It is one of only eight countries as of the mid-1990s with a population in excess of 25 million in combination with a TFR in excess of five births per woman (United Nations 1996). Pakistan stands apart from its populous neighbors in South Asia, all of which (with the exception of Nepal) experienced substantial declines in fertility prior to 1990 and therefore show markedly lower fertility in the mid-1990s (estimated TFRs in

POPULATION AND DEVELOPMENT REVIEW 24(4):773–796 (DECEMBER 1998)773

Click to print article Click Clickto return to returnto Table to of Table Contents of Contents 774 FERTILITY TRANSITION IN PAKISTAN

1995 are 2.2 for Sri Lanka, 3.4 for India and Bangladesh, and 5.4 for Ne- pal).2 These comparisons raise questions about which factors have precipitated and sustained fertility transition in the region, in particular the contribution of family planning programs (Shah and Cleland 1993). Some analysts credit the rapid and, by many criteria, early fertility decline in Bang- ladesh to the effectiveness of the family planning program (Cleland et al. 1994). Pakistan’s program, by comparison, has repeatedly been criticized for poor management and the low quality of the services offered to the majority of its clients (Robinson, Shah, and Shah 1981; Rukanuddin and Hardee-Cleaveland 1992; United Nations n.d.). An alternative, and not necessarily contradictory, argument is that the motivation to limit fertility did not crystallize in Pakistan before the 1990s, and therefore responsibility for the late transition cannot be placed entirely on an indecisive population policy and inadequate family planning services (Sathar 1993). In this article we review the evidence suggesting that important demographic changes are underway; describe the large-scale social and economic changes that have motivated the recent changes in reproductive behavior; and examine the more-direct causes of these changes and the constraints on further changes. We conclude by speculating about the prospects for further declines in fertility in Pakistan.

Status of the fertility transition

Numerous fertility surveys indicate that the TFR in Pakistan remained above 6.0 births per woman throughout the 1980s (see Table 1). Significantly, all estimates for the 1990s for the first time fall below 6.0 births per woman. The direct estimate of the TFR for 1986–91 from the 1990–91 Pakistan De- mographic and Health Survey (PDHS) is 5.4. This is almost certainly an underestimate, judging from the evidence from reinterviews conducted with a subsample of PDHS respondents. The reinterview data suggest that roughly 10 percent of births were omitted (Curtis and Arnold 1994). Adjusting for this through the Gompertz relational model, Juarez and Sathar (forthcoming) estimate that the TFR was probably around 6.1 for the period 1986– 91, a figure that is consistent with the Pakistan Demographic Survey estimate for 1992 of 5.8 (Federal Bureau of Statistics 1997). The estimate from the 1994–95 Pakistan Contraceptive Prevalence Survey (PCPS) is slightly lower, at 5.6 for the period 1994–95 (Population Council 1998b). Collec- tively, these estimates imply a modest decline of 0.3 births during the four- year period 1991 through 1994, which if extrapolated implies a TFR of 5.4 in 1997.3 The Pakistan Fertility and Family Planning Survey (PFFPS) of 1996–97 (NIPS/LSHTM 1998) provides a direct estimate of 5.3 for the period 1992–96, which suggests a slightly more rapid decline during the 1990s than implied by the previous surveys.4 Finally, the 1998 census, whose

Click to return to Table of Contents Z EBA A. SATHAR / JOHN B. CASTERLINE 775

TABLE 1 Estimates of total fertility rates, by decade, source, and type of estimate, Pakistan, 1960s to 1990s TFR Period Decade and source covered Direct Indirect

1960s Population Growth Estimation 1962–65 8.0a Experiment (1962–65) 1962–65} 6.1a Pakistan Fertility Survey (1975) 1965–69 7.1 National Impact Survey (1968–69) 1968–69 5.0 Population Growth Survey (1968–71) 1968–71 6.0 1970s Pakistan Fertility Survey (1975) 1970–74 6.3 Pakistan Labour Force and Migration Survey (1979) 1970–74 7.1 Pakistan Labour Force and Migration Survey (1979) 1975–79 6.5 Population Growth Survey (1976–79) 1976–79 6.9 1980s Pakistan Contraceptive Prevalence Survey (1984–85) 1984–85 6.0 Pakistan Demographic Survey (1984–88) 1984–88 6.9 Pakistan Demographic and Health Survey (1990–91) 1986–91 5.4 Pakistan Demographic and Health Survey (1990–91) 1987–91 6.1b 1990s Pakistan Demographic Survey (1992) 1992 5.8 Pakistan Contraceptive Prevalence Survey (1994–95) 1994–95 5.6 Pakistan Fertility and Family Planning Survey (1996–97) 1992–96 5.3 aTwo different methods of estimation were applied to the PGE data for this period. bThis rate was calculated by Juarez and Sathar (forthcoming) using the Gompertz relational model to adjust for misplacement errors and applies to 0–4 years before the survey. SOURCES: Sathar (1993): Table 1; Population Council (1998b): Table 7.6; NIPS/LSHTM (1998): Table 3.6.

provisional results were released in July 1998, indicates that the average population growth rate for the period 1981–98 was 2.6 percent per annum, a decline from previous intercensal rates and consistent with a decline in fertility in the 1990s. All demographic analyses point to a decline in fertility in the 1990s. Over the past four decades there has been no other period when demographers could point with confidence to a decline of similar magnitude and duration (i.e., beyond year-to-year fluctuation). Demographic estimation in Pakistan is a treacherous business, however, owing to the inaccurate

Click to return to Table of Contents 776 FERTILITY TRANSITION IN PAKISTAN reporting of ages and dates of events. This is a familiar problem for those who have followed demographic research on Pakistan: from the first demographic surveys carried out in the 1960s up to the present, systematic response errors have frustrated efforts to construct an accurate picture of levels and trends in fertility and its proximate determinants (Shah, Pullum, and Irfan 1986; Hashmi and Ahmed 1992). The more persuasive evidence that fertility decline is underway is provided by data on contraceptive practice. As indicated in Table 2, the fraction of currently married women reporting use of a method of contraception at the time of the survey remained below 10 percent through the 1980s. Surveys in the 1990s, in contrast, reveal a marked increase in the contraceptive prevalence rate, from 12 percent in 1990–91 (PDHS) to 18 percent in 1994–95 (PCPS) to 24 percent in 1996–97 (PFFPS), a rise of roughly 2 percentage points per annum. This rise in the proportion of married women using contraception is the most compelling corroborating evidence that marital fertility is declining. The method-specific trends in Table 3 reveal that increased use of modern contraceptive methods—female sterilization and the IUD in particular—explains most of this overall trend (7.9 points out of the 12.1 percentage point increase from 1990–91 to 1996–97). This finding should not overshadow the substantial contribution of withdrawal, which accounts for almost one-third of the increase, more than any other single method. This is an unusual feature that distinguishes the early stages of Pakistan’s marital fertility transition from other Asian transitions in the second half of the twentieth century (Population Council 1998a; Freedman 1995). Urban–rural differentials in fertility provide further insight into the nature of the fertility transition. This differential widened in the 1980s (see Table 4), as marital fertility decline was initiated in the major cities, most notably Karachi (NIPS/IRD 1992). The differential has widened in recent

TABLE 2 Estimates of use of contraception among currently married women (percent), by source, Pakistan, 1960s to 1990s Source Ever use Current use

National Impact Survey (1968–69) 12.1 5.5 Pakistan Fertility Survey (1975) 10.5 5.2 Pakistan Contraceptive Prevalence Survey (1984–85) 11.8 9.1 Pakistan Demographic and Health Survey (1990–91) 20.7 11.8 Pakistan Contraceptive Prevalence Survey (1994–95) 28.0 17.8 Pakistan Fertility and Family Planning Survey (1996–97) 36.4 23.9

Click to return to Table of Contents Z EBA A. SATHAR / JOHN B. CASTERLINE 777

TABLE 3 Estimates of current use of contraception by method among currently married women (percent), by source, Pakistan, 1990s Source PDHS PCPS PFFPS Method (1990–91) (1994–95) (1996–97)

Any method 11.8 17.8 23.9 Any modern method 9.0 12.6 16.9 Condom 2.7 3.7 4.2 Pill 0.7 0.7 1.6 Injectable 0.8 1.0 1.4 IUD 1.3 2.1 3.4 Female sterilization 3.5 5.0 6.0 Other 0.0 0.1 0.3 Any traditional method 2.8 5.2 7.0 Rhythm 1.3 1.0 1.9 Withdrawal 1.2 4.2 4.6 Other 0.3 — 0.5

NOTE: For names of surveys see Table 2. years, with the result that the urban–rural difference in the TFR exceeds 1.5 births as of the mid-1990s. A corresponding urban–rural difference in contraceptive prevalence is also evident (36 percent and 18 percent in urban and rural areas respectively, according to the 1996–97 PFFPS). A closer examination of patterns over the past two decades reveals that a pronounced

TABLE 4 Estimates of total fertility rates, by place of residence and source, Pakistan, 1970s to 1990s Source Period Urban Rural

Pakistan Fertility Survey (1975) 1970–74 6.2 6.4 Pakistan Labour Force and Migration Survey (1979) 1975–79 6.2 6.6 Pakistan Contraceptive Prevalence Survey (1984–85) 1984–85 5.5/6.1a 6.2 Pakistan Demographic Survey (1992) 1992 6.2 7.3 Pakistan Demographic and Health Survey (1990–91) 1986–91 4.7/5.2a 5.6 Pakistan Contraceptive Prevalence Survey (1994–95) 1994–95 4.5 6.3 Pakistan Fertility and Family Planning Survey (1996–97) 1992–96 4.2 5.7 aThe PCPS and the PDHS divided urban areas into major urban areas and “other” urban areas; TFRs were higher in the latter category.

Click to return to Table of Contents 778 FERTILITY TRANSITION IN PAKISTAN increase in the practice of fertility control began among the urban upper strata (as delineated, for example, by female schooling) in the early 1980s and subsequently spread to other urban strata (Sathar and Mason 1993). All the available data indicate that contraceptive prevalence did not begin to increase in rural areas until the 1990s and therefore has not yet had a noticeable impact on fertility. The increase is found in each of the four provinces of Pakistan: as expected, Punjab has the highest prevalence levels, but the Northwest Frontier province experienced the most rapid rise in contraceptive use in the early 1990s. The upshot is a transition that in its early stages is following a conventional pattern in which reproductive intentions and behavior change first among the upper social strata in the cities, with other urban residents and rural areas lagging behind, roughly six or seven years in this instance. To the extent it persists, the urban–rural differential has implications for national fertility trends because Pakistan has been rapidly urbanizing (as compared, for example, to other countries in South Asia). According to the 1998 census, the urban population constitutes 33 percent of the total population (Population Census Organization 1998). We place great weight on trends in contraceptive behavior because they represent deliberate efforts by couples to break from the reproductive regime that has characterized Pakistan in recent decades. A balanced assessment of this historic transformation must also consider the other major proximate determinants of fertility: nuptiality, postpartum practices, and induced abortion. Information on nuptiality is relatively abundant and shows a steady increase in the average age at first marriage of females throughout the five decades since Independence: singulate mean ages at first marriage calculated from census and survey data rose from 16.9 years in 1951 to 21.6 years in 1990–91 (PDHS) and 22.1 in 1996–97 (PFFPS) (NIPS/LSHTM 1998). The rise has been somewhat sharper in urban areas (Sathar 1997). Thus nuptiality change appears to have preceded the onset of marital fertility decline in Pakistan, as in other Asian countries in the twentieth century (Casterline 1994). The magnitude of the recent nuptiality change, however, is too slight to account for much of the estimated fertility decline in the 1990s. Recent trends in postpartum behavior are more difficult to discern because of gaps and incomparabilities in survey measurement over the past two decades. The available evidence suggests that the overall length of breastfeeding (unsupplemented and supplemented) has remained around 19 to 20 months from the 1970s to the present (WFS 1984; NIPS/IRD 1992) and therefore has not contributed to the recent fertility decline. There are no reliable estimates of the incidence of induced abortion.5 But judging from qualitative investigations, induced abortion is a far more common method of birth control in Pakistan in the 1990s than had been previously assumed. Whether recourse to induced abortion has increased in the 1990s is even more difficult to ascertain, although this view is com-

Click to return to Table of Contents Z EBA A. SATHAR / JOHN B. CASTERLINE 779 monly volunteered by medical practitioners in Pakistan. Like withdrawal, induced abortion is a strategy for limiting fertility that falls outside the pur- view of the family planning program. In summary, a review of the evidence on trends in the proximate determinants of fertility provides no basis for questioning either the validity of the estimated fertility decline in the 1990s or the assertion that the decline is mainly the result of increased practice of contraception within marriage.

Determinants of recent changes in fertility

We now turn to the question of which factors have led to this historic development. The context is laid out in Table 5, which presents trends in basic demographic, economic, and social indicators. If one accepts the classic hypothesis that fertility transition is a response to declines in mortality and improvements in social and economic conditions, circumstances in Pa- kistan in the past four decades support conflicting expectations. By many criteria the country’s fertility transition is several decades overdue: mortality fell steadily beginning in the 1950s, resulting in very rapid population growth by the 1960s that continued through the 1980s. Urbanization, income, and literacy/schooling have increased steadily throughout these decades. Exposure to the mass media, as indicated by the prevalence of radios and televisions, increased rapidly beginning in the 1970s.

TABLE 5 Selected demographic, economic, and social indicators, Pakistan, 1960–90 1960 1970 1980 1990

Population size (millions) 45.9 60.4 82.1 112 Population growth (percent per year) 2.7 2.8 2.9 3.0 Population density per square km of agricultural land 200 248 324 — Life expectancy at birth, both sexes (years) 43 46 50 56 Infant mortality rate (per 1000 births) 162 143 126 103 GDP per capita (1987 US$) 135 206 251 349 Percent urban 22 25 28 32 Percent literate Total 15 21 24 35 Female — 11 — 21 Primary school enrollment ratio Total 30 40 39 44 Female 13 22 27 30 Radios per 1000 population 6 17 67 87 Televisions per 1000 population — 2 10 17

SOURCES: World Bank 1983, 1992; United Nations 1990, 1997.

Click to return to Table of Contents 780 FERTILITY TRANSITION IN PAKISTAN

Offsetting these trends, however, were other conditions that acted against fertility reduction. The social environment has not been conducive to fertility change. By most measures, the status of Pakistani women is very low, even in comparison to other settings in South Asia (United Na- tions 1993). Pakistani women lack autonomy and have restricted mobility and minimal discretionary access to economic resources (Sathar et al. 1988; Sathar and Kazi 1997). Virtually all social indicators reveal large gender inequities. Despite improvement, levels of literacy and school attainment remain very low among females (Table 5). The gender system gives dispro- portionate power to Pakistani men in decisionmaking about reproduction. Further buttressing the existing reproductive regime has been the feudal structure of Pakistani society in which political and economic power has remained heavily concentrated in a small number of families, especially in rural Sind and southern Punjab. These families have resolutely sustained pronatalist institutions and social values and posed obstacles to the implementation of effective social-sector programs. The religious milieu, which has consistently maintained doubts about the moral acceptability of the practice of fertility control, constituted an additional stumbling block. Nor did macroeconomic trends provide a rationale for fertility reduction prior to the 1990s. The economy performed strongly from the 1960s through the 1980s, with significant income growth resulting from expansion in the agricultural sector and from substantial remittances from Pakistani migrant workers in the Middle East, among other factors. Gross domestic product (GDP) per capita grew at an average annual rate of 2.7 percent during the period 1970–85 (Hasan 1997). Income growth was, however, uneven and was not accompanied by proportionate growth in investments in such social services as schooling, health, and housing (Hasan 1997). In short, whatever the economic costs to Pakistani couples of large numbers of births, these costs seemed affordable in a period of income growth and, more to the point, were perceived as lower than the costs of challenging deeply held cultural precepts about reproduction and women’s roles. This contrasts with the historical process in East Asia, where economic development fueled an upsurge in desires to restrict fertility (Greenhalgh 1988; Asian Development Bank 1997). Accompanying this set of cultural, social, and economic factors that propped up the existing fertility regime in Pakistan was weak and inconsistent political support for policies aimed at reducing fertility (Rosen and Conly 1996). The end result of all these forces was deep ambivalence at the household level about restricting fertility (Shah and Cleland 1993). A final factor, which perhaps should be viewed as an expression of the combined force of all the other factors, was the ineffectiveness of the family planning program, which left the vast majority of Pakistani couples without convenient access to adequate services (Rukanuddin and Hardee-Cleaveland 1992).

Click to return to Table of Contents Z EBA A. SATHAR / JOHN B. CASTERLINE 781

Each of these factors sustaining high fertility has weakened in the 1990s. The steady economic growth of the 1970s and 1980s slowed considerably in the 1990s. Annual growth in GDP per capita declined to 1.2 percent in 1985– 95 (Hasan 1997). The popular view is that the country is in the midst of a deep economic crisis—aggravated, according to many observers, including economists, by the structural adjustment and related conditionalities imposed by the IMF and the World Bank as a remedy (Hussain et al. 1997; Khan and Aftab 1997).6 The economic crisis of the 1990s has compounded the stresses created by the long-term trend in rural areas toward land fragmentation and the more recent declines in returns to many agricultural crops (Choudhry 1997). Symptomatic of labor absorption problems in rural areas has been a growing stream of migrants from rural to urban areas in search of off-farm employment, as a result of which the population has become increasingly urbanized (see Table 5). The outcome of these macroeconomic forces has been an increase in unemployment, a decline in real wages for unskilled labor, and consequent increases in poverty and income inequality (Hussain et al. 1997). In retrospect, the relatively healthy economy of the 1970s and 1980s sustained resistance to social change, a resistance that the economic crisis of the 1990s has finally undermined. Associated with the sense of economic stress, and in some respects accentuating it, have been changes in life styles and in parental aspirations for children. These changes have led to a pronounced increase in the cost of each child. Consumer aspirations have outstripped changes in purchasing power in real terms. The greater concentration of the population in urban areas and the expanded exposure to the mass media (itself a result in part of urbanization) are the proximate causes of these changes. These trends are shown in Table 5 for the period 1960–90. The trends continue in the 1990s. For example, in the six-year period between the PDHS and the PFFPS the fraction of households with televisions increased from 27 to 38 percent (NIPS/IRD 1992; NIPS/LSHTM 1998). This has occurred despite the travails of the economy, revealing the rapidly increasing attachment to modern consumer durables that itself must contribute to increases in the perceived costs of childbearing. A much more modern Pakistani life style is now transmitted through the media (particularly in advertising) to all cor- ners of the country, including rural areas that were previously isolated. This in turn has had a direct bearing on how parents want to raise their children, with implications for standards of hygiene and health, tastes in clothing and consumer durables, and schooling aspirations. As Pakistanis themselves note in qualitative interviews (Population Council 1997d), all these changes make children more costly; this has driven down the demand for children and increased the perceived costs of unwanted births. Changes in social structure, particularly in kinship relationships, are also underway. In Pakistan, as elsewhere in the northern half of the In-

Click to return to Table of Contents 782 FERTILITY TRANSITION IN PAKISTAN dian subcontinent, the traditional kinship structure places comparatively little emphasis on the conjugal bond, which is viewed as a threat to highly valued consanguineous ties, intergenerational ties in particular (Vatuk 1987; Das Gupta 1996). This system also has the dual effect of restricting the decisionmaking power of couples of childbearing age and, perhaps more importantly, marginalizing women (especially young married women). Al- though key elements of this system have been changing, these changes do not include expanded decisionmaking power for women. The evidence suggests that gender inequalities have been relatively constant in Pakistan, especially in rural areas. The vast majority of women continue to experience restrictions on spatial movement outside the home and limited decisionmaking power within the household, and do not communicate effectively with their husbands (Sathar and Kazi 1997). We find no evidence that the dominant authority of men over their wives—regarding childbearing and other matters—has diminished as of the mid-1990s. It is not an increase in the autonomy of women that seems to have been decisive during the past decade but rather an increase in the autonomy of couples of childbearing age. Over the past three decades, kinship relations and household structures have evolved in a manner that has eroded the power of elders and other relatives, and as a result decisionmaking about family matters has become more nucleated. One indicator of this change is an increase in husband–wife discussion of fertility desires, reported by roughly one-third of respondents in the 1990–91 PDHS and by almost double that fraction in the 1994–95 PCPS (NIPS/IRD 1992; Popula- tion Council 1998b). A second indicator of this change is the increase in the proportion of nuclear-family households.7 About one-half of the households in Pakistan are now nuclear (Population Council 1998b). When, as in Pakistan, a major cause of the growing predominance of nuclear households is rural-to-urban migration, in which couples with or without children move to towns and cities and set up separate households, it is much less likely that the new households are adjacent to households of extended kin. Reinforcing the effect of migration in freeing couples from the daily influence of extended kin is the fact that migration in Pakistan selects posi- tively for literacy and schooling, among other reasons because of the job skills demanded by the nonagricultural labor markets in towns and cities (Kazi and Sathar 1996).8 Among the effects of the increasing independence of the conjugal unit has been greater consensus between spouses in their assessments of the costs and benefits of childbearing and related domestic issues. The economic and other costs of children are now perceived as more exclusively the responsibility of the children’s parents, with the consequence that Pakistani men have become much more cognizant of the costs of childrearing. This development is reflected in recent survey and qualitative data that show

Click to return to Table of Contents Z EBA A. SATHAR / JOHN B. CASTERLINE 783 both men and women acknowledging the advantages of restricting fertility, with women stressing health concerns related to childbearing and men the economic costs of rearing many children (Population Council 1997c). It is reflected, too, in the large contribution of male methods (withdrawal, condoms) to the recent increase in contraceptive prevalence (Table 3). These broad economic and social changes in Pakistan are the fundamental causes of the observed changes in the 1990s in fertility and its proximate determinants—contraception in particular—that were summarized above. These fundamental causes have acted through a set of factors that are the more-direct causes of the observed changes. We now turn to these more-direct causes, classifying them into three categories: —Changes in the demand for children. —Changes in family planning services, for example their density and quality. —Changes in other factors constraining contraceptive use, that is, the social, psychic, and cultural “costs of contraception.”

Demand for children

The earliest demographic surveys in Pakistan in the 1960s did not reveal a demand for large families. The mean ideal family size reported in the 1968– 69 National Impact Survey was 4.4 children and in the 1975 Pakistan Fer- tility Survey 4.1 children. As a result, the prevailing assumption for the past three decades has been that there is a substantial latent demand for fertility regulation in Pakistan (Sirageldin, Norris, and Hardee 1976; Shah and Cleland 1993). As in an earlier article (Sathar 1993), we take issue with this conventional interpretation. We feel that in the past latent demand for fertility regulation was in fact weaker than survey data have suggested, for four reasons: the limited autonomy of women, which for multiple and reinforcing reasons has worked against the adoption of fertility regulation practices (Sathar et al. 1988); the strong preference for male progeny; the relatively healthy state of the economy; and a political and cultural climate that looked unfavorably on the deliberate exercise of fertility control. During the 1990s, fertility demand has changed in two fundamental respects: the desired number of children has declined; and the attachment to fertility preferences has strengthened, that is, preferences have “crystallized.” We consider both developments. One basic fertility demand parameter is ideal family size, captured by responses to a question such as “If you could choose exactly the number of children to have in your whole life, how many would that be?” The available evidence suggests a sizable downward shift in ideal family size in the first half of the 1990s, from 4.1 children in the 1990–91 PDHS to 3.6 in the 1994–95 PCPS. More valid measurement of fertility preferences is prob-

Click to return to Table of Contents 784 FERTILITY TRANSITION IN PAKISTAN ably provided by responses to the question “Do you want another child?” The parity-specific percentages desiring no more children increase between 1990–91 and 1994–95 (see Table 6). The trend for the 20 years from the mid-1970s to the mid-1990s is curvilinear: a gradual decline in antinatalism from the 1975 PFS to the 1990–91 PDHS is followed by an increase in the first half of the 1990s. Ideal family size and the data on fertility preferences shown in Table 6 relate to the number of children desired, but they say nothing about the strength of attachment to those desires. The strength of preferences can vary independently of the number of children desired, as a function for example of the perceived magnitude of the costs and benefits of an additional child (Bongaarts 1993). The strength of preferences is difficult to measure in demographic surveys. National surveys in Pakistan do not contain items that permit an assessment of trends in the strength of fertility preferences; hence we can only rely on what we have concluded from empirical data (quantitative and qualitative) collected in Pakistan over the past two decades. These data suggest to us that the degree of conviction about stated preferences is much deeper in the 1990s than was the case earlier: there is a heightened consciousness of these preferences, the preferences are in sharper focus (i.e., more explicitly articulated), and the determination to realize them is stronger, in comparison to previous decades. These changes in the demand for children are a response to the economic adversity and other social changes described above. Qualitative interviews conducted in the mid-1990s reveal the link between a growing sense of economic stress and an increased desire to restrict fertility (Popu-

TABLE 6 Fertility preferences by number of living children: Percentage of currently married women who want no more children, by source, Pakistan, 1970s to 1990s Number of PFS PCPS PDHS PCPS living children (1975) (1984–85) (1990–91)a (1994–95)

0 2 1 2 0 1 7 4 412 2 30171725 3 48363651 4 69585268 5 78756378 6 90837183 7+ 94 90 74 90 Total 49 43 40 52 a Survey combines the responses “want no more” and “sterilized.” NOTE: For names of surveys see Table 1. SOURCES: PFS (1975): Population Planning Council of Pakistan 1976: Table 3.1.1; PCPS (1984–85): Population Welfare Division, Government of Pakistan 1986: Table X.1; PDHS (1990–91): National Institute of Population Studies 1992: Table 8.1; PCPS (1994–95): Ministry of Population Welfare 1995: Figure 5.1.

Click to return to Table of Contents Z EBA A. SATHAR / JOHN B. CASTERLINE 785 lation Council 1997c). In these unstructured interviews, Punjabi men and women were asked about the advantages and disadvantages of having many children and why they and/or their neighbors were motivated to practice family planning. Again and again, both men and women emphasized the costs of children and, especially, how much more difficult it has become to shoulder these costs in the economic circumstances of the 1990s.9 While not clearly articulated by the respondents, one senses that the changes in life style aspirations and in childrearing aspirations that we noted above have also contributed to the conviction that the household is falling short economically. In inquiries carried out in earlier decades, the view that children have become unaffordable was not stated as often or as emphatically (NIPS n.d.). The diffusion of this conviction throughout the population appears to be a phenomenon of the 1990s. Despite the growing sense that children are unaffordable, ideal family size remains in excess of three children. According to the parity-specific percentages of women wanting no more births (Table 6), a small minority of women are content with two children or fewer: in the 1994–95 PCPS, only 25 percent of the women expressed a desire to stop at two surviving children, and 12 percent a desire to stop with one child. A far larger fraction of women must be prepared to limit their childbearing to two births or less if replacement-level fertility is to be achieved. Westoff and Bankole (1995) have demonstrated this through simulations in which various frac- tions of those women who express a desire to avoid pregnancy but are not using contraception are assumed to be using a method. Under two simulations using 1990–91 PDHS data, contraceptive prevalence would increase from 12 percent to 27 percent and 38 percent. These would result in declines in the TFR from 5.4 births per woman (the PDHS direct estimate) to 4.4 and 3.6, respectively, a range that roughly brackets the most recent estimates of the average ideal number of children. These would be substantial declines in relation to present fertility levels but would leave fertility far above replacement nevertheless. Son preference partially accounts for the persistence of the desire for relatively numerous births. Many women with three or more surviving children want to have more children in the hope of having a son (Mahmood 1996). Analyses of survey data collected in the 1990s consistently show that son preference affects both fertility preferences and contraceptive use. In particular, even among those women indicating a desire to postpone the next pregnancy or to avoid having another pregnancy altogether, the likelihood of using contraception increases with the number of surviving sons (Population Council 1997a; Sathar and Kazi 1997). Two sons remain the basic minimum in the minds of most couples, and there is no evidence that this view is waning significantly even in urban areas. This powerful and highly conscious son preference is to be expected given the many social and economic structures that devalue females and the reliance in a

Click to return to Table of Contents 786 FERTILITY TRANSITION IN PAKISTAN patrilineal system on sons for support in old age (Sathar and Kazi 1997). Other forces are responsible for a continuing preference for three or more children. Despite substantial declines in the past four decades, infant and child mortality levels remain relatively high (Table 5), and insecurity about child loss continues to be an oft-cited reason in unstructured interviews for having a larger than ideal family size (Population Council 1997c). Fur- thermore, a larger family size increases the likelihood that one or more children might enjoy success in employment or marriage, thereby improving the standing of the family. Particularly in the large agricultural sector, children’s labor continues to be valued. Most Pakistani households view children as the most stable form of social and economic security (Sathar and Kazi 1997).

Family planning services

A second class of explanations for the onset of fertility decline in Pakistan emphasizes improvements in family planning services, in terms of both density and quality. From its inception through the 1980s, the national family planning program fell short on both counts (Robinson, Shah, and Shah 1981; Rukanuddin and Hardee-Cleaveland 1992). There are solid grounds for believing that the provision of family planning services has improved in the 1990s, extending a trend that began in the 1980s (Ross et al. 1992). Several new schemes that were launched in the Eighth Five-Year Plan (1993–98) have begun to take root (Rosen and Conly 1996). Two innova- tions are noteworthy. The social marketing of contraceptives, begun in 1986, has gained momentum in the 1990s. Despite some setbacks, social marketing has probably improved accessibility in urban areas (Davies and Agha 1997; Agha and Davies 1998). A Village-Based Family Planning Workers program, launched on a pilot basis in 1992, now provides family planning services and advice in about 8,000 villages (Population Council 1997b). Other initiatives may have contributed to the improvement of family planning services in the 1990s. A re-invigorated primary health care program was launched by the government in 1994, with another cadre of community-based female workers posted in both urban slums and rural areas. Thirty thousand such workers are believed to have been deployed with the responsibility to provide primary health care, including family planning services, to the communities from which they were recruited (Minis- try of Health 1996). Furthermore, Ministry of Health outlets, including rural health centers and basic health units, now have a stronger mandate to provide family planning services. Finally and perhaps most significantly, radio and television campaigns promoting family planning have picked up momentum, in terms of both their frequency and the explicitness of the messages. The percentage of women who reported that they had heard or seen a family planning message more than

Click to return to Table of Contents Z EBA A. SATHAR / JOHN B. CASTERLINE 787 doubled between the 1990–91 PDHS and the 1994–95 PCPS, exceeding 60 percent in the latter survey (NIPS/IRD 1992; Ministry of Population Wel- fare and Population Council 1995). The radio and television messages le- gitimize family planning practice and offer concrete information about how contraceptives can be obtained. A recent analysis comparing the impact of information, education, and communication programs in Bangladesh, India, and Pakistan shows substantial improvements in the reach of family planning messages in the Pakistani media and in their association with contraceptive adoption (Westoff and Bankole 1998). Undoubtedly, then, family planning program activities have improved in the 1990s and probably deserve some credit for the recent increase in the rate of contraceptive prevalence and the resulting decline in fertility. But the important enhancements in services were implemented only in the mid-1990s and hence cannot be given credit for the sharp increase in contraceptive prevalence in the first half of the decade. In our judgment, the contribution of improvement in family planning services has been mi- nor as compared to the contribution of the other major factors (changes in the demand for children, reduction in other costs of contraception). As a telling indication of this, a substantial fraction of the increase in the rate of contraceptive prevalence in the 1990s has involved use of withdrawal: 3.4 points out of the 12.1 percentage point increase between 1990–91 and 1996–97 (see Table 3). Withdrawal is a means of avoiding pregnancy that is free of any dependence on contraceptive supplies or services and, indeed, is often regarded as a sign that fertility decline is more the result of increasing motivation to avoid pregnancy than of improvements in the means to do so (Santow 1995). Family planning services remain deficient in many respects. For example, basic coverage of the population is far from complete in both urban and rural areas, and especially in the latter. Various criteria classify far less than half of the population as having convenient access to any type of services. One study concludes that family planning clinics (the Family Wel- fare Centres) are accessible to only 10 percent of the population, with only 5 percent living within easy walking distance (Rosen and Conly 1996). Data from the 1994–95 PCPS show an average traveling time to a family planning service delivery point of 60 minutes in rural areas (as compared to 15 minutes in urban areas). The nearest facility is three or more miles away for 61 percent of never users in rural areas and for 13 percent of never users in urban areas. Clearly clinics are far closer on average for women in urban areas, but deficiency of access is a serious obstacle nevertheless because of the many restrictions on women moving about outside their dwell- ings. The upshot is that only a small minority of women report having had contact with a family planning worker (Ministry of Population Welfare and Population Council 1995), and this is essentially unchanged since the late 1970s (Syed 1979). The social marketing and the Village-Based Family Plan-

Click to return to Table of Contents 788 FERTILITY TRANSITION IN PAKISTAN ning Workers schemes hold much promise of significantly improving access to supplies and services, although rural areas remain largely untouched by social marketing (Davies and Agha 1997). Beyond access, there are serious weaknesses in the quality of family planning services. The situation analyses that have been carried out are unambiguous on this point (e.g. Cernada et al. 1993). Outlets rarely offer a choice of methods and informed advice about alternative methods, often do not have sufficient supplies on hand, and typically lack an adequate number of trained staff. Pakistani couples concur with this evaluation: in unstructured interviews, men and women (especially rural residents) repeatedly describe providers as poorly trained and unmotivated and the contraceptive supplies as defective and unreliable (Population Council 1997c). The majority of women, especially in rural areas, seek family planning services in large hospitals; other types of outlets, even where available, are rarely visited (Population Council 1998b). This harsh evaluation of the family planning program is based on evidence from careful local studies. Like Freedman (1995), we consider Ross et al.’s (1992) rating of the program effort in Pakistan as “moderate” too favorable, probably reflecting undue weight given to policy rhetoric and formal bureaucratic structure and too little weight given to the extent and quality of the services actually provided to Pakistani couples.

Other costs of contraception

Our third explanation for recent changes is that other forces that hinder contraceptive use have weakened. Many types of evidence suggest that the costs of using contraception in Pakistan go well beyond the financial and time costs of obtaining contraceptive supplies and services of acceptable quality. We refer to social, psychic, health, and cultural factors that, together with financial and time costs, constitute what Easterlin (1975) defines as the “costs of contraception.” Measurement of these factors is not straightforward, hence their contribution is difficult to assess empirically. A reading of the literature on fertility in Pakistan since the 1960s and evidence obtained through in-depth qualitative and quantitative research conducted in Punjab province in 1996 suggest that three obstacles to fertility control behavior weakened significantly in the 1990s: the view that family planning is proscribed by moral and religious codes (Shah and Shah 1984); the view that extended-family members, and parents in particular, have a rightful voice in the fertility decisions of a husband and wife; and the view that contraceptive practice threatens men’s authority over women. For contraceptive practice to become widespread, these views must be undermined through a process that legitimizes couples’ capacity to take deliberate actions to avoid pregnancy (Shah and Cleland 1993).

Click to return to Table of Contents Z EBA A. SATHAR / JOHN B. CASTERLINE 789

Such a process has been evident in Pakistan in the 1990s, driven by social and economic changes. Among the most important of these have been the changes in family structure described above, which have eroded the power of elders and other extended kin, especially among migrants to urban areas (Kazi and Sathar 1996). This erosion of power undermines the second of the obstacles just identified. We have argued that one consequence of this change in family structure is greater concurrence, and most likely better communication, between spouses about the costs of children and a heightened sense that these costs fall on both husband and wife. For example, in data collected in 1996 Punjabi women report little hesitance in approaching their husbands about using contraception (Population Coun- cil 1997a). This has meant, in turn, that family planning practice does not threaten gender relations to the extent it appears to have done in the past; in the 1990s, to use contraception is not to act against the interests of the husband. Moreover, couples that opt for those contraceptive methods— withdrawal, condoms—that are male-controlled (see Table 3) eliminate the threat contraception might pose to patriarchal interests. Finally, the real- ization that rearing children is costly, and that these costs are increasing, counters the conviction that use of contraception is immoral. In unstructured interviews, male respondents in particular cite the moral duty to provide for children’s needs, a duty that is difficult to fulfill with a large family (Population Council 1997c). These factors have led to greater legitimization of contraception, reinforced by the convincing antinatalist pronouncements of government officials since Prime Minister Nawaz Sharif’s public endorse- ment of family planning in 1991 (repeated in 1997 and 1998), by the intensi- fication of media campaigns espousing family planning, and by the greater visibility of family planning program activities at local levels. What has been underway in Pakistan—first in urban areas, more slowly in rural areas (probably beginning in the Punjab)—is a social dynamic in which family planning has acquired the acceptability that it lacked in the past. There is a shift of values away from traditional to modern, and a shift in the locus of decisionmaking from extended kin to the conjugal couple. The dynamic has its own momentum: as some Pakistani couples begin to practice contraception and this becomes known, the same choices become more acceptable to other couples. Couples may also acquire useful information from each other, about where to obtain services and how to cope with deleterious side effects of contraception on health, for example. In this fashion the social, psychic, and cultural barriers to contraception begin to crumble, very much resembling a “contagion” process. This dynamic may well characterize most fertility transitions at some stage (Mont- gomery and Casterline 1996). It provides one of the more plausible and parsimonious explanations for the sudden increase in contraceptive prevalence in Pakistan in the 1990s after several decades during which surveys

Click to return to Table of Contents 790 FERTILITY TRANSITION IN PAKISTAN repeatedly suggested that latent demand for fertility control far exceeded contraceptive practice. While certain costs of contraception may be falling, others remain as serious obstacles to adopting and continuing use of contraception. As a result, despite the recent sharp increase in contraceptive prevalence, “unmet need for contraception” (i.e., non-use of contraception by women at risk of pregnancy who do not wish to become pregnant) remains high—on the order of one-third of currently married women (Westoff and Bankole 1995; Population Council 1998b; NIPS/LSHTM 1998). Indeed, comparison of estimates from the three national surveys conducted in the 1990s suggests that unmet need may even be on the increase, a common pattern in the early stages of fertility transition as declines in demand for children out- pace increases in contraceptive use (Bongaarts 1997). In a recent in-depth study conducted in the Punjab on factors underlying unmet need, three factors emerge as the most decisive obstacles to family planning: a woman’s perceptions that her husband disapproves; fear of detrimental health side effects of contraception; and concerns about the social, cultural, and religious acceptability of birth planning (Population Council 1997a).10 These and other considerations make many women feel that the decision to use contraception is both unacceptable on social and cultural grounds and against their best interests, even if they wish to avoid pregnancy. In tran- scripts from unstructured interviews, the anxiety that women feel about the decision whether or not to use contraception is almost palpable. Con- flicting forces encourage and discourage family planning practice. The resulting ambivalence makes it difficult for many women to use contraception, especially given the effort required to obtain a method in the many communities where family planning services are distant and/or of poor quality. These considerations are compounded by the limited autonomy of women in this and other spheres of life, which as noted above does not appear to be changing appreciably. The experience of East Asia in the period 1950–80 demonstrated that fertility can decline to relatively low levels despite the low status of women: there are no grounds for asserting that increased female autonomy is a precondition for fertility transition. Nevertheless, the lack of female autonomy almost certainly hinders the transition from a high to low fertility regime: the expression of women’s desire for fewer children is muted, and the range of contraceptive choices is restricted (Jejeebhoy 1995). In addition, many women who wish to postpone or terminate childbearing believe that their husbands desire more children and oppose contraceptive practice. In fact, empirical data collected in the mid-1990s indicate that men have more positive views about contraception (and particular contraceptive methods) and are less pronatalist than their wives assume (Population Council 1997a). However, given the limited communication between spouses and the unequal balance of power in gender relations, these misperceptions by wives are not readily corrected

Click to return to Table of Contents Z EBA A. SATHAR / JOHN B. CASTERLINE 791 and are a stronger deterrent than would be the case in many other settings. The result is that Pakistani women often take the safe route of not implementing their fertility preferences, in deference to the presumed wishes of their husbands. It is hardly surprising that men have played a decisive role in the developments of the 1990s, as reflected, for example, in the relatively large contribution of increased use of withdrawal and condoms.

Conclusions

Evidence is accumulating that a transition in marital fertility behavior has begun in Pakistan in the 1990s. The evidence is of several types and comes from multiple and independent sources. In searching for the forces underlying this development, we have emphasized changes in the demand for children—a crystallization of existing desires for smaller families, combined with a decline in family size desires (itself reflecting broader economic and social changes)—as well as changes in the social, psychic, and cultural costs of contraception. We give less credit to improvements in family planning services, although they have made a contribution. In examining the obstacles to further changes in fertility, in contrast, we have stressed the immediate need for improvements in family planning services and for further reductions in other costs of contraception. Taking this assessment of the constraints on further change as a starting point, we can sketch likely scenarios for the next decade or so, and we can identify the types of policies that might promote faster fertility decline. Because a large fraction of currently married women has unmet need for contraception, in theory a substantial increase in contraceptive practice and a corresponding decline in fertility could occur without any change in fertility preferences (Shah and Cleland 1993).11 This is an important conclusion from the standpoint of near-term policy formulation and program development. But from the standpoint of the long-term goal of achieving replacement-level fertility in Pakistan, the existing level of demand for children must be regarded as the most serious constraint. As noted above, Westoff and Bankole (1995) estimate that, were existing unmet need to be satisfied, the TFR would remain in the range 3.6–4.4 births per woman, a level that implies continued rapid population growth. A first point is that the downward trend in the demand for children apparent in the early 1990s is likely to continue to exert downward pressure on fertility. As a natural outcome of a growing awareness that marital fertility can be deliberately regulated, the strength of attachment to desires for moderate or small numbers of children (i.e. fewer than five) will grow, while simultaneously the fraction of couples desiring large numbers of children will decline, in response to the trends in economic aspirations and perceived costs of children described above. Pakistan is increasingly urbanized, and childrearing and consumer aspirations that outstrip the ability to

Click to return to Table of Contents 792 FERTILITY TRANSITION IN PAKISTAN satisfy them, and hence have antinatalist implications, are spreading to lower socioeconomic strata and rural areas. Thus we expect that a growing fraction of Pakistani couples will wish to limit their number of children to fewer than four. Eventually this will undermine the marked differentials in reproductive behavior according to place of residence (urban or rural) and level of schooling that have characterized the past two decades. Are these changes in demand likely to be matched by rises in contraceptive prevalence and by corresponding declines in fertility? In the long run the answer must be yes. In the short run, however, the answer is dependent on the extent of changes in family planning service delivery. A particular need is greater coverage in rural areas, where the limited mobility of women makes even modest travel distances to family planning services and supplies an imposing barrier to the adoption and continued use of contraception. One strategy would be to depend mainly on an extensive network of private services in urban areas (here we second Rukanuddin and Hardee-Cleaveland 1992 and Rosen and Conly 1996 in calling for greater reliance on the private sector, both nongovernmental organizations and medical practitioners), with public resources freed up for more extensive deployment in rural areas. Efforts must also be made to upgrade the quality of services. Interviews in the 1996 study in Punjab reveal the dim view of public-sector services held by many women and men. In short, given the heightened motivation to restrict fertility, the key in the short run to maintaining and perhaps even accelerating fertility transition in Pa- kistan will be improvements in family planning services. The onset of the transition has been triggered by changes in fertility demand and by reductions in the social, psychic, and cultural costs of contraception. None of these costs has been entirely eliminated. If family planning services remain inadequate, then we expect that unmet need for family planning will remain at its current high level or even increase. Under the most desirable scenario the ongoing declines in demand for children would be accompanied by radical improvements in the supply and quality of family planning services. This scenario offers the conditions for a rapid fertility decline in Pakistan over the next decade or so.

Notes Major funding for this research was provided Amin, John Bongaarts, Shahrukh Khan, and by the Rockefeller Foundation through its Cynthia Lloyd. grant to the Population Council, “Unmet Need 1 On the basis of a detailed analysis of the for Family Planning: Its Nature and Causes.” birth histories collected in the 1975 Pakistan Additional funding was provided by the United Fertility Survey, Alam (1984) concluded that Nations Population Fund. Helpful comments fertility decline had begun in the early 1970s. on an earlier draft were provided by Sajeda This conclusion was later refuted by Retherford

Click to return to Table of Contents Z EBA A. SATHAR / JOHN B. CASTERLINE 793 et al. (1987), who demonstrated that the im- PFS and the 1990–91 PDHS. The comparison pression of decline could almost certainly be is not exact, because of definitional differences attributed to reporting errors. Similarly, a de- in the household structure variables in the two cline in marital fertility in the late 1980s evi- data sets. Our comparison probably understates dent in direct estimates from the Pakistan the amount of change. We find that the per- Demographic and Health Survey of 1990–91 centage of households that are nuclear has (NIPS/IRD 1992) disappears when the data are been relatively stable in rural areas but has in- subjected to more rigorous analyses that ad- creased by ten percentage points in urban ar- just for problems of data quality (Curtis and eas. This is consistent with our argument be- Arnold 1994; Juarez and Sathar forthcoming). low that urban Pakistani couples in particular 2 A far lower TFR for Nepal is proposed are experiencing new-found independence in by Dangol, Retherford, and Thapa (1997). On reproductive decisionmaking, as reflected in the basis of analysis of survey data from the the much greater recent change in fertility in 1970s to the early 1990s, they conclude that urban areas. the TFR as of 1991 was 4.8. According to their 8 There is good evidence that household analysis, the marital fertility transition in Ne- structure is associated with women’s au- pal was progressing relatively rapidly as of the tonomy. Data collected in the Punjab in 1993 late 1980s. show that women living in nuclear households 3 This is consistent with the United Na- have greater autonomy: 62 percent and 22 per- tions (1996) estimate for the TFR in 1995 of cent of women living in nuclear households 5.5 births per woman. make decisions regarding schooling of children 4 The PFFPS estimates will need to be and major household purchases, respectively, scrutinized. In Pakistan, fertility estimates de- compared to 46 percent and 12 percent of rived from birth histories typically are down- women living in extended households (Sathar wardly biased for the period immediately pre- and Kazi 1997). The same study finds that ceding the survey. Even so, it seems likely that women living in nuclear households have as of 1998 the TFR falls in the range of 5.0– greater mobility; for example, 40 percent of 5.5—rather high by international and regional these women are able to go to a health center standards but one birth less than the level that alone, compared to 19 percent of those living characterized the previous 30 years. in extended households. 5 In the 1994–95 PCPS, 5.9 percent of 9 Ironically, the economic adversity cre- women aged 35 and older reported that their ated by the combination of stagnant wages and most recent pregnancy ended in an abortion, rising costs of basic goods has made contracep- spontaneous or induced. Less than 1 percent tives less affordable for most households. This of the women reported an induced abortion. circumstance is especially relevant for meth- Recognizable spontaneous abortions should ods such as injectables, oral contraceptives, and be on the order of 15 to 20 percent among condoms that are supplied primarily by the pri- women in this age group. vate sector. Agha (1998) shows strong cross- sectional associations between the use of some 6 The rupee has weakened substantially of these methods and household income. since the early 1980s, driving up the local price of essential items that are import-de- 10 A similar set of factors is identified in pendent such as cooking oil, kerosene, and research by Hashmi, Alam, and Sheraz fertilizer. The consumer price index trebled (1993) on reasons for non-use. between 1982 and 1995 (Hasan 1997), and 11 Our 1996 research in Punjab province according to authoritative (but unofficial) (Population Council 1997a) revealed that estimates the rate of inflation in the costs of weak attachment to stated fertility prefer- basic goods in the 1990s has exceeded 20 ences is among the reasons for unmet need percent per annum. for contraception. It follows that increased 7 A number of independent sources con- contraceptive practice could also result from firm this assertion. We have compared the dis- further crystallization of preferences and tribution of households by type in the 1975 stronger convictions about those preferences.

Click to return to Table of Contents 794 FERTILITY TRANSITION IN PAKISTAN

References

Agha, Sohail. 1998. “Is low income a constraint to contraceptive use among the Pakistani poor?” PSI Research Division Working Paper No. 15. Washington, DC: Population Ser- vices International. Agha, Sohail and John Davies. 1998. “Contraceptive social marketing in Pakistan: Assess- ing the impact of the 1991 condom price increases on sales and consumption,” PSI Research Division Working Paper No. 14. Washington, DC: Population Services Inter- national. Alam, Iqbal. 1984. “Fertility levels and trends,” in Iqbal Alam and Betzy Dinesen (eds.), Fertility in Pakistan: A Review of Findings from the Pakistan Fertility Survey, pp. 65–80. Voorburg, Netherlands: International Statistical Institute. Asian Development Bank. 1997. Emerging Asia: Changes and Challenges. Manila. Bongaarts, John. 1993. “The supply-demand framework for the determinants of fertility: An alternative implementation,” Population Studies 47, no. 3: 437–456. ———. 1997. “Trends in unwanted childbearing in the developing world,” Studies in Family Planning 28, no. 4: 267–277. Casterline, John B. 1994. “Fertility transition in Asia,” in Therese Locoh and Veronique Hertrich (eds.), The Onset of Fertility Transition in Sub-Saharan Africa, pp. 69–86. Liège: Derouaux Ordina. Cernada, George P., A. K. Ubaidur Rob, Safia I. Ameen, and Muhammed Shafiq Ahmad. 1993. “A situation analysis of family welfare centres in Pakistan,” Working Paper No. 4, Population Council/Islamabad. Choudhry, Ghaffar. 1997. “Pakistan’s agricultural development since independence: Intertemporal trends and explanations,” paper presented at the 13th Annual General Meeting of the Pakistan Society for Development Economists, Islamabad. Cleland, John, James F. Phillips, Sajeda Amin, and G. M. Kamal. 1994. The Determinants of Reproductive Change in Bangladesh: Success in a Challenging Environment. Washington, DC: World Bank. Curtis, Sian L. and Fred Arnold. 1994. “An evaluation of the Pakistan DHS Survey based on the reinterview survey,” Demographic and Health Surveys Occasional Papers, No. 1. Calverton, MD: Macro International. Dangol, Bishnu Das, Robert D. Retherford, and Shyam Thapa. 1997. “Declining fertility in Nepal,” Asia-Pacific Population Journal 12, no. 1: 33–54. Das Gupta, Monica. 1996. “Life course perspectives on women’s autonomy and health outcomes,” Health Transition Review, Supplement to Volume 6: 213–232. Davies, John and Sohail Agha. 1997. “Ten years of contraceptive social marketing in Paki- stan: An assessment of management, outputs, effects, costs and cost-efficiency, 1987– 1996,” PSI Research Division Working Paper No. 7. Washington, DC: Population Ser- vices International. Easterlin, Richard. 1975. “An economic framework for fertility analysis,” Studies in Family Planning 6, no. 3: 54–63. Federal Bureau of Statistics. 1997. Pakistan Demographic Survey 1992. Karachi: Statistics Divi- sion. Freedman, Ronald. 1995. “Asia’s recent fertility decline and prospects for future demographic change,” Asia-Pacific Population Research Reports No. 1. Honolulu: East-West Center, Pro- gram on Population. Government of Pakistan. 1965. Third Five Year Plan. Karachi: Planning Commission. Greenhalgh, Susan. 1988. “Fertility as mobility: Sinic transitions,” Population and Develop- ment Review 14, no. 4: 629–674. Hasan, P. 1997. “Learning from the past: A fifty year perspective of Pakistan’s development,” paper presented at the 13th Annual General Meeting of the Pakistan Society for Development Economists, Islamabad.

Click to return to Table of Contents Z EBA A. SATHAR / JOHN B. CASTERLINE 795

Hashmi, Sultan S. and Tauseef Ahmed. 1992. “Shy/silent users of contraception: Further evidence,” Pakistan Population Review 3, no. 1: 19–40. Hashmi, Sultan S., Khushnood Alam, and Aysha Sheraz. 1993. Non-Users and Unmet Need for Contraception. Islamabad: National Institute of Population Studies. Hussain, Akmal, et al. 1997. Overcoming Poverty. Report of the Task Force on Poverty Eradi- cation, May, Islamabad. Jejeebhoy, Shireen J. 1995. Women’s Education, Autonomy, and Reproductive Behaviour: Experi- ence from Developing Countries. Oxford: Clarendon Press. Juarez, Fatima and Zeba Sathar. Forthcoming. “Emerging evidence of fertility change in Pakistan,” in Basia Zaba and Alaka Basu (eds.), Brass Tacks. Oxford: Oxford University Press. Kazi, Shahnaz and Zeba Sathar. 1996. “Gender and development: Searching for explanations for fertility change in rural Pakistan,” paper presented at the IUSSP Seminar on Comparative Perspectives on Fertility Transition in South Asia, Rawalpindi, Decem- ber. Khan, Shahrukh Rafi and Safiya Aftab. 1997. “Structural adjustment, labour, land, and the poor in Pakistan,” The Lahore Journal of Economics 2, no. 1: 1–18. Mahmood, Naushin. 1996. “Gender differences in fertility desires and son preference in Pakistan: Implications for reproductive behaviour,” paper presented at the IUSSP Semi- nar on Comparative Perspectives on Fertility Transition in South Asia, Rawalpindi, December. Ministry of Health. 1996. The Prime Minister’s Programme for Family Planning and Primary Health Care, Second Evaluation. Islamabad. Ministry of Population Welfare and the Population Council. 1995. Pakistan Contraceptive Preva- lence Survey 1994–95. Islamabad: Population Council. Montgomery, Mark R. and John B. Casterline. 1996. “Social learning, social influence, and new models of fertility,” in John B. Casterline, Ronald D. Lee, and Karen A. Foote (eds.), Fertility in the United States: New Patterns, New Theories, pp. 151–175. Population and Development Review, Supplement to Volume 22. National Institute of Population Studies (NIPS) and IRD/Macro International Inc. 1992. Pa- kistan Demographic and Health Survey 1990/1991. Columbia, MD: IRD/Macro Interna- tional. National Institute of Population Studies (NIPS)/London School of Hygiene and Tropical Medicine (LSHTM). 1998. Pakistan Fertility and Family Planning Survey 1996–1997. Islamabad: National Institute of Population Studies. National Institute of Population Studies (NIPS). n.d. The Economic Value of Children. Islamabad. Population Census Organization. 1998. Provisional Results of Fifth Population and Housing Census, Islamabad. Population Council. 1997a. The Gap Between Reproductive Intentions and Behaviour: A Study of Punjabi Men and Women. Islamabad. ———. 1997b. “Initial performance and impact of VBFPWs in four districts in Punjab,” Research Report No. 5. Islamabad. ———. 1997c. “Exploring barriers to contraception in Pakistan: Findings from qualitative research.” Islamabad. ———. 1997d. The Village Relations Study. Population Council Briefing Paper. Islamabad. ———. 1998a. “A qualitative investigation into the use of withdrawal,” Research Report No. 6. Islamabad. ———. 1998b. Pakistan Contraceptive Prevalence Survey 1994–95: Final Report. Islamabad. Population Planning Council of Pakistan. 1976. Pakistan Fertility Survey: First Report. Islamabad. Population Welfare Division, Government of Pakistan. 1986. Pakistan Contraceptive Preva- lence Survey 1984–85. Islamabad. Retherford, Robert, G. M. Mirza, M. Irfan, and Iqbal Alam. 1987. “Fertility trends in Paki- stan: The decline that wasn’t,” Asia and Pacific Forum 1, no. 1: 3–10.

Click to return to Table of Contents

796 FERTILITY TRANSITION IN PAKISTAN

Robinson, Warren C., Makhdoom A. Shah, and Nasra M. Shah. 1981. “The family planning program in Pakistan: What went wrong?” International Family Planning Perspectives 7, no. 3: 85–92. Rosen, James E. and Shanti R. Conly. 1996. Pakistan’s Population Program: The Challenge Ahead. Washington, DC: Population Action International (Country Study Series No. 3). Ross, John A., W. Parker Mauldin, Steven R. Green, and E. Romana Cooke. 1992. Family Planning and Child Survival Programs as Assessed in 1991. New York: Population Council. Rukanuddin, Abdul Razzaque and Karen Hardee-Cleaveland. 1992. “Can family planning succeed in Pakistan?” International Family Planning Perspectives 18, no. 3: 109–115. Santow, Gigi. 1995. “Coitus interruptus and the control of natural fertility,” Population Studies 49, no. 1: 19–44. Sathar, Zeba A. 1993. “The much awaited fertility decline in Pakistan: Wishful thinking or reality?” International Family Planning Perspectives 19, no. 4: 142–146. ———. 1997. “La croissance démographique au Pakistan: Une maîtrise délicate,” in J.-C. Chasteland and J.-C. Chesnais (eds.), La population du monde: enjeux et problèmes. Paris: Presses universitaires de France. Sathar, Zeba, Nigel Crook, Christine Callum, and Shahnaz Kazi. 1988. “Women’s status and fertility change in Pakistan,” Population and Development Review 14, no. 3: 415– 432. Sathar, Zeba A. and Shahnaz Kazi. 1997. Women’s Autonomy, Livelihood, and Fertility: A Study of Rural Punjab. Islamabad: Pakistan Institute of Development Economics. Sathar, Zeba A. and Karen Mason. 1993. “How female education affects reproductive behavior in urban Pakistan,” Asian and Pacific Population Forum 6(4). Shah, Iqbal H. and John G. Cleland. 1993. “High fertility in Bangladesh, Nepal, and Paki- stan: Motives versus means,” in Richard Leete and Iqbal Alam (eds.), The Revolution in Asian Fertility: Dimensions, Causes, and Implications, pp. 175–207. Oxford: Clarendon Press. Shah, Iqbal H., Thomas Pullum, and M. Irfan. 1986. “Fertility in Pakistan during the 1970’s,” Journal of Biosocial Science 18, no. 2: 215–229. Shah, Nasra M. and Makhdoom A. Shah. 1984. “From non-use to use: Prospects of contraceptive adoption,” in Iqbal Alam and Betzy Dinesen (eds.), Fertility in Pakistan: A Re- view of Findings from the Pakistan Fertility Survey, pp. 149–162. Voorburg, Netherlands: International Statistical Institute. Sirageldin, Ismail, Douglas Norris, and J. Gilbert Hardee. 1976. “Family planning in Paki- stan: An analysis of some factors constraining use,” Studies in Family Planning 7, no. 5: 144–154. Syed, Sabiha H. 1979. “Communications channels and family planning in Pakistan,” Studies in Family Planning 10, no. 2: 53–60. United Nations. 1990. Human Development Report 1990. New York: Oxford University Press. ———. 1993. Women’s Status and Fertility in Pakistan: Recent Evidence. New York: United Na- tions, Department for Economic and Social Information and Policy Analysis. ———. 1996. World Population Prospects: The 1996 Revision. New York: United Nations, De- partment for Economic and Social Information and Policy Analysis. ———. 1997. Human Development Report 1997. New York: Oxford University Press. Vatuk, Sylvia. 1987. “Authority, power, and autonomy in the life cycle of the North Indian woman,” in P. Hockings (ed.), Dimensions of Social Life. New York: de Gruyter Press. Westoff, Charles F. and Akinrinola Bankole. 1995. “Unmet need: 1990–1994,” Demographic and Health Surveys Comparative Studies No. 16. Calverton, MD: Macro International. ———. 1998. “Mass media and reproductive behavior in Pakistan, India, and Bangladesh.” Manuscript: Office of Population Research, Princeton University. World Bank. 1983. World Tables: Third Edition: Baltimore: Johns Hopkins University Press. ———. 1992. World Development Report 1992. New York: Oxford University Press. World Fertility Survey (WFS). 1984. World Fertility Survey: Major Findings and Implications. Voorburg, Netherlands: International Statistical Institute.

Click to print article Click toto returnreturn to to T ableTable of ofContents Contents

NOTES AND COMMENTARY

Population, Carbon Emissions, and Global Warming: Comment

PAUL DE SA

IN A RECENT article in this journal, Frederick Meyerson (1998) examined the connection between population growth and the anthropogenic emissions of greenhouse gases (GHGs) that are thought to be changing the global climate (Intergovernmental Panel on Climate Change 1996). Focusing on the recent Kyoto Protocol (United Nations 1997), which, upon ratifica- tion, will commit 38 developed (“Annex I”) countries to binding limits on their future levels of GHG emissions, Meyerson claimed to show that “[a]t Kyoto, the negotiators appear to have overlooked or ignored the wide variation in projected population change among signatory countries” (p. 115).1 In this comment, I will point out flaws both in the evidence that Meyerson presents to support this charge and in a number of his other statements about the connection between climate change and population growth.

Meyerson’s argument

Meyerson’s only argument supporting his claim that the Kyoto Protocol ignores projected differences in population growth is as follows: a) Under the Protocol, the Annex I countries have agreed to reduce their combined annual GHG emissions, averaged over the period 2008–12, to approximately 95 percent of 1990 levels. This aggregate reduction will be achieved through a binding commitment from each Annex I country to limit its own average annual GHG emissions between 2008 and 2012 to a certain proportion of 1990 levels. These proportions, given in the first column of Meyerson’s Table 2 (p. 120), range from 92 percent for the countries of the European Union to 110 percent for Iceland. b) Using a standard projection (United Nations 1996) to anticipate future population growth, one may convert the Kyoto targets for total na-

POPULATION AND DEVELOPMENT REVIEW 24(4):797–803 (DECEMBER 1998)797

Click to print article ClickClick to return to return to Table to of Table Contents of Contents 798 P OPULATION, CARBON EMISSIONS, AND GLOBAL WARMING tional emissions levels into per capita equivalents. For example, taking 2010 as the reference year, Germany is committed to reducing its total GHG emissions to 92 percent of 1990 levels, while the United Nations forecasts that the German population will increase by 3.9 percent over the period 1990– 2010. Thus Germany’s per capita emissions must fall by approximately 11 percent between 1990 and 2010 to meet its Kyoto obligation. Converting each Annex I country’s national limits into per capita equivalents in a similar fashion (the fifth column of Table 2, p. 120), one finds that the Kyoto commitments translate into a “wide variation in the effective national per capita reduction requirements” (p. 121), ranging from Luxembourg’s 23 percent per capita reduction between 1990 and 2010 to a 10 percent per capita increase over the same period for Latvia. c) These differences in per capita reduction commitments mean that “countries with high population growth rates are in a sense held to a more difficult standard . . . because they must make larger reductions in their per capita emissions rate” (p. 125). This is neither “workable [n]or fair” (p. 126), and demonstrates that the negotiators at Kyoto must have ignored population growth, presumably because they would not otherwise have produced such an inequitable agreement.

Problems with the argument

In this section, I will discuss two problems with Meyerson’s argument. First, his reliance on per capita emissions reductions as a measure for assessing the Kyoto Protocol is unfounded. Second, he neglects the ways in which projections of future population growth did, in fact, influence the outcome of the Kyoto negotiations.

Reductions in per capita emissions as a measure of fairness

As described above, Meyerson’s argument that the different population growth rates of developed countries were ignored at Kyoto rests entirely on his observation that, under the Protocol, individual Annex I countries have, effectively, undertaken to make per capita emissions reductions of differing magnitudes. But why should we be concerned with per capita reductions in the first place? After all, it seems that one could reproduce Meyerson’s argument equally well for any of the other factors that influence national GHG emissions levels. For example, using economic growth projections to turn the Kyoto targets into emissions per unit of GDP, one would again find that countries have agreed to reduction commitments of differing sizes. Is this also unfair, showing that the Kyoto negotiators ignored the differential rates of economic growth projected for the Annex I

Click to return to Table of Contents P AUL DE SA 799 countries? Furthermore, it is easy to show not only that total GHG emissions could increase even as per capita emissions fall, which would obviously defeat the purpose of the Kyoto agreement, but that allocating a fixed emissions total between countries such that per capita reductions are equal- ized in effect introduces an incentive for countries to increase their populations, so as to decrease their emissions reduction requirements.2 The only reason Meyerson gives for focusing on per capita carbon emissions is to claim that their levels “have stabilized or even decreased in the last two decades . . . [so] that emission increases in the developed world are now primarily driven by population growth” (p. 125). But he provides no proof for this assertion, even though it contradicts not only other, more detailed studies that have examined the importance of different sources of GHG emissions increases, such as economic growth, energy pricing, carbon intensity, and so on (e.g. Bongaarts 1992; Panayotou and Sachs 1998), but also the data in Table 1 (p. 119), which show that per capita emissions levels of most Annex I countries have in fact varied significantly over the last 25 years. Moreover, even if one accepts his focus on per capita emissions, Meyerson does not explain what is “unfair” about asking Annex I countries to make per capita reductions of differing sizes. For example, although they have comparable standards of living, Table 1 (p. 119) shows that Japan’s average annual per capita carbon emissions in 1990 amounted to 2.32 metric tons (mt), while those of the United States totaled 5.07 mt. It is, therefore, not immediately obvious why it is so unreasonable that “the United States will be required to reduce its per capita emissions by 20.9 percent from 1990 levels, compared to reductions of only . . . 8.6 percent for Japan” (p. 122). Surely, even if average per capita emissions levels had stabilized, as Meyerson claims, it would still be necessary to explain why these levels cannot be changed—for example, why the average American should have the right to produce over twice the emissions of an average Japanese, rather than an approximately equal amount as one might intu- itively expect (Subak 1993; Masood 1997). Without such an explanation, there is no reason why it is unfair for Annex I countries to cut their per capita emissions by different percentages, nor why historical per capita emissions trends are meaningful indicators of the likelihood of meeting future targets (Figures 5–8, pp. 123–125).3 In fact, without justification, Meyerson himself replaces the standard of equal per capita emissions reductions with one of equal per capita emissions when the discussion turns to developing countries. In the section “Be- yond Kyoto” (pp. 126–127), he assumes that total annual global GHG emissions are limited to 60 percent of 1990 levels—a target aimed at stabilizing atmospheric concentrations of carbon dioxide at twice preindustrial lev- els4—and converts this total into average global per capita equivalents for

Click to return to Table of Contents 800 P OPULATION, CARBON EMISSIONS, AND GLOBAL WARMING the years 1990–2050 by dividing by the total world population forecast for those years (Figure 10, p. 127). Comparing these notional future per capita limits with the current per capita emissions of non–Annex I countries, Meyerson concludes that “the developing world as a whole will already exceed the advisable per capita emission limit by the year 2000, and that major players, like China and Mexico, are already having a detrimental effect on climate, both because of their total carbon emissions and because of high average per capita emissions” (p. 127). This is a puzzling statement, not only because it holds non–Annex I countries to a much more difficult standard than Annex I countries (how can one call, respectively, China’s and Mexico’s 0.7 mt and 1.1 mt 1995 per capita emissions “high,” when the average 1995 Annex I emissions for that year were 3.16 mt per capita?), but also because one of the basic scientific facts about climate change is that all GHG emissions into the atmosphere contribute to climate change, irrespective of whether they come from countries with high or low per capita or total emissions: there is no threshold, above which emissions suddenly become “detrimental.” In the words of Schelling (1997), “the atmosphere is a global common where everybody’s emissions mingle with everybody else’s”—and all of them contribute to climate change. Finally, on the subject of non–Annex I countries, although they will indeed “become an equal and then progressively greater source of emissions within the next few decades” (p. 126), it is not entirely true to claim that they have been “exempted from emission limitations because of their ‘developing’ status” (p. 128). Rather, developing countries did not assume binding commitments at Kyoto at least in part because of the Annex I countries’ historical responsibility for the GHGs currently accumulated in the atmosphere, which are actually causing climate change. With today’s developed countries responsible for 70–80 percent of the anthropogenic contribution to current levels of carbon dioxide in the atmosphere (Subak 1993), it is disingenuous to portray the exemption of the non–Annex I countries from binding emissions limits as some kind of favor, granted spe- cially on account of their developing status. Rather, it reflects the need to balance future commitments against past responsibilities.

How differential rates of population growth are incorporated into climate-change policy

Meyerson’s argument is further weakened by his failure to discuss the ways in which projections of future population growth were in fact incorporated into the discussions leading up to the Kyoto Conference. For example, on a prosaic level, most of the National Communications submitted by the Annex I countries to the UN’s Kyoto Secretariat in preparation for the conference include information about trends in population size, growth rate,

Click to return to Table of Contents P AUL DE SA 801 and demographic characteristics (e.g. Government of Japan 1997; Govern- ment of the United Kingdom 1997).5 Thus, although the Protocol itself was the outcome of a contentious political process, there is clear evidence that the negotiators were aware of the relevance of population issues. Moreover, it is important to recognize that much of the debate surrounding the Kyoto targets was influenced by projections of future levels of national GHG emissions. These projections, which implicitly include assumptions about differential rates of population growth (as well as about economic growth, technological progress, and so forth), are based on modeling the causal processes that underlie the production of GHG emissions, processes that Meyerson ignores in his simplistic focus on the correlation between emissions and population growth. For example, one standard method of projecting future GHG emissions is to recognize that they largely arise from fossil fuels, burned to meet the demand for energy in various sectors of the economy. To represent this process, total emissions levels can be disaggregated to reflect the sectors from which they arise:

Total Emissions = Emissions from industry + Emissions from transportation + Emissions from electricity generation + Emissions from other sources.

Each factor on the right hand side can then be analyzed according to the causal relationship between energy use and emissions, so that, for example:

Emissions from industry = Industrial energy use x Average emissions per unit of energy used by industry and similarly for the other three terms. For each sector, the factors that influence the “energy use” term—such as levels of energy efficiency and increased demand due to economic or population growth—can be examined, as can those that influence the “emissions per unit of energy used” term—such as carbon intensity, technological change, and the types of fuel used. Using different projections for these various factors, alternative scenarios for future levels of GHG emissions at a local, national, or global scale can be produced to forecast reductions in GHG emissions that might result from different policy measures. For example, assuming no action to reduce GHG emissions (a so-called business-as-usual scenario), and rates of population growth similar to those forecast by the UN, one projection (Eu- ropean Commission 1996) forecasts that carbon dioxide emissions from the United States will increase by 18 percent between 1990 and 2010, while those from Japan will increase by 20 percent over the same period. The Kyoto targets should then be considered as reductions from these business-as-usual projections for national emissions, so that, continuing with

Click to return to Table of Contents 802 P OPULATION, CARBON EMISSIONS, AND GLOBAL WARMING the European Commission projection, Japan’s 6 percent Kyoto reduction commitment is in effect a 24 percent emissions reduction compared to what would otherwise have occurred, and the US reduction commitment of 7 percent is similarly equivalent to a 25 percent reduction. The point is that Annex I countries based their Kyoto negotiating positions, at least to some extent, on what levels of emissions reductions these projections showed to be feasible, and that such projections implicitly include differential rates of population growth, as can be easily verified by examining the assumptions used to produce them (e.g. Energy Informa- tion Administration 1996; European Commission 1996). It is therefore inaccurate to claim that differential population growth rates were simply ignored at Kyoto: through the use of these type of projections, they were implicitly included in exactly the same way as other factors that influence national GHG emissions levels.

Conclusion

I have tried to show that Meyerson’s argument that differential rates of population growth were ignored at Kyoto is seriously weakened by his un- justified assertion that equality of per capita emissions reductions should be the criterion for fairness, and by his neglect of the ways in which projections of future population growth were incorporated into the Kyoto negotiations, both directly in documents and indirectly through projections of future GHG emissions levels on which the Kyoto commitments are, at least partly, based. Let me close, however, by stressing that neither of these criticisms should be taken as implying that the Kyoto Protocol is a perfect solution, or that population matters—like many other issues—did not deserve greater attention during the Kyoto negotiations. Indeed, Meyerson is undoubtedly correct to highlight the need to understand more adequately the relationship between population and climate change as fully, and as quickly, as possible.

Notes

1 All page references are to Meyerson “capable of meeting its Kyoto obligations with (1998). relative ease” (p. 123), as Meyerson asserts 2 On the other hand, Kyoto-type caps solely on the basis of Figure 8 (p. 125). on total emissions provide an incentive to 4 Meyerson mistakenly refers to “twice reduce population growth as one method of the pre-anthropogenic carbon dioxide lev- reducing total emissions. els” (p. 127). 3 For example, as is clear from the analy- 5 All of the National Communications sis in European Commission (1997), it is by are available at the United Nations web site: no means obvious that the European Union is www.unfccc.org.

Click to return to Table of Contents

PAUL DE SA803

References

Bongaarts, John. 1992. “Population growth and global warming,” Population and Develop- ment Review 18: 299–319. Energy Information Administration. 1996. Assumptions for the International Energy Outlook 1996. Washington, DC: US Department of Energy. European Commission. 1996. European Energy to 2020: A Scenario Approach. Luxembourg: Office for Official Publications of the European Communities. ———. 1997. Climate Change—The EU Approach for Kyoto. Brussels: Commission of the European Communities. Government of Japan. 1997. Japan’s Second National Communication Under the UN Framework Convention on Climate Change. Tokyo: Ministry of Foreign Affairs. Government of the United Kingdom. 1997. Climate Change: The United Kingdom Programme. London: Her Majesty’s Stationery Office. Intergovernmental Panel on Climate Change. 1996. Climate Change 1995: Impacts, Adapta- tions, and Mitigation of Climate Change, ed. R. Moss, R. Watson, and M. Zinyowera. Cam- bridge: Cambridge University Press. Masood, Ehsan. 1997. “Equity is the key criterion for developing nations,” Nature 390: 216– 217. Meyerson, Frederick A. B. 1998. “Population, carbon emissions, and global warming: The forgotten relationship at Kyoto,” Population and Development Review 24: 115–130. Panayotou, Theodore and J. Sachs. 1998. Private communication. Schelling, Thomas C. 1997. “The cost of combating global warming,” Foreign Affairs 76: 8– 14. Subak, Susan. 1993. “Assessing emissions: Five approaches compared,” in The Global Green- house Regime: Who Pays?, ed. P. Hayes and K. Smith. New York: United Nations Uni- versity Press. United Nations. 1996. World Population Prospects: The 1996 Revision. New York. ———. 1997. Conference of the Parties, Third Session. Kyoto Protocol to the United Nations Framework Convention on Climate Change. Kyoto, Japan.

Click to print article ClickClick to return to return to Table to of Table Contents of Contents

Toward a Per Capita–Based Climate Treaty: Reply

FREDERICK A. B. MEYERSON

REASONABLE PEOPLE can disagree on the best approach for greenhouse gas reduction, but few would dispute that the Kyoto Protocol is in serious trouble as a mechanism for effective, long-term climate policy. In the wake of the November 1998 Buenos Aires Conference of the Parties, a functioning global emissions reduction system is still an unrealized goal, and what will happen after 2012 is even less clear. It is evident that future population size and growth rates will have a major impact on the quantity and geographical source of emissions, as well as a bearing on the policy instru- ments necessary to effect reductions and mitigate the potentially serious economic, social, and environmental effects of climate change. In this issue of the Review, Paul de Sa makes a number of interesting comments on my article, “Population, carbon emissions, and global warming: The forgotten relationship at Kyoto” (Meyerson 1998b). Some of his critique constitutes valid discussion of a complex scientific and policy problem; other points may be more the product of misunderstanding. His comments, in conjunction with my earlier article, create a valuable springboard for a more serious discussion about the relationship between population and carbon emissions, and the possibilities for a treaty that recognizes and integrates this essential nexus. De Sa’s two basic criticisms of my analysis of the Kyoto Protocol are that 1) using per capita emission reductions as a measure for assessing the Kyoto Protocol is “unfounded,” and 2) my article neglects the ways in which projections of future population growth influenced the outcome at Kyoto. He comments only briefly on the central conclusions of my article, that the fixed national cap system employed by Kyoto is unworkable in the long run because of projected demographic changes among Annex 1 countries, and that this issue is of even greater importance for the developing countries, which necessarily must be active participants in any future climate treaty. I will briefly address the two issues de Sa has raised and then discuss the larger issue—how to devise a climate policy approach that can

804POPULATION AND DEVELOPMENT REVIEW 24(4):804–810 (DECEMBER 1998)

Click to print article Click toClick return to returnto Table to of Table Contents of Contents F REDERICK A. B. MEYERSON 805 effectively respond to the multiple, interrelated challenges of emissions reduction, demographic change, and equity.

A per capita standard for carbon emissions

Assessing the effect of the Kyoto Protocol in terms of per capita emissions runs to the heart of two central questions—is the treaty equitable? and is it practical in terms of the immediate and long-term demands it places on each country? If the answer to either question is no, then the likelihood for the success of the climate treaty is small. Countries, individually or collectively, will find a way to weaken, forestall, or back out of their commitments. The atmosphere, including its ability to effectively absorb greenhouse gas emissions, is a global resource, much like the oceans. There is now broad scientific consensus that anthropogenically induced climate change poses a serious threat to human economic and social welfare (Intergovern- mental Panel on Climate Change 1995; O’Neill, MacKellar, and Lutz in press). If climate change is to be minimized, the atmosphere in the sense of a sink for greenhouse gas emissions must be considered a limited resource. One can make a reasonable argument that the right to use this resource should be akin to an international human right. It also seems appropriate, in that case, to develop rights and responsibilities with respect to that global resource on the basis of the per capita average necessary to stabilize the atmosphere.1 Viewing the Kyoto Protocol from this perspective, problems are apparent with regard both to its initial formulation of national caps and to the long- term effects of those fixed caps attributable to demographic change. Among the Annex 1 countries, emissions in 1990 ranged from 1.15 metric tons of carbon per capita (mt/capita) for Portugal to 6.88mt/capita for Luxembourg (Meyerson 1998b). Among the major economic powers, US emissions in 1990 were 5.07mt/capita, more than twice those of Japan (2.32 mt/capita) and of most European countries. Leaving aside the issue of demographic change, the Kyoto Protocol from the outset locks in an uneven set of rights and responsibilities whose basis is a particular historical level of emissions. It is difficult to view as equitable an agreement that creates virtually the same responsibilities for the government and citizens of Portugal (an 8 percent reduction from 1990 emission levels by 2010) as for the United States (a 7 percent reduction), when the United States is producing more than four times as much greenhouse gas per capita (Meyerson 1998b; United Nations 1997). The credibility of the Kyoto agreement is stretched even thinner when one considers that a number of key developing countries have per capita emissions that exceed those of many Annex 1 countries, including South Africa (2.0mt in 1995) and the Republic of Korea (2.3mt) (Carbon Dioxide Information Analysis Center 1998). Other major non–An-

Click to return to Table of Contents 806 T OWARD A PER CAPITA–BASED CLIMATE TREATY nex 1 countries, like Mexico and China, are also poised to cross over this line in the next decade. As discussed in my original article and in de Sa’s comment, the wide variation in demographic projections from 1990 to 2010 among the An- nex 1 countries further illustrates the difficulties with using fixed national emission caps, because of their effects on per capita emissions standards over time. Here perhaps de Sa misunderstands the underlying thrust of my argument as somehow being a defense of high per capita emissions levels by certain countries, particularly the United States. This is a mistaken interpretation; there is no question that countries with high total and per capita emissions should bear the greatest responsibility in any greenhouse gas reduction plan (Meyerson 1998c). However, the major differences in projected population growth and decline between 1990 and 2010 among the Annex 1 countries, ranging from a 23.5 percent population increase for Australia to a 16.5 percent decrease for Latvia, demonstrate the hazards of applying a simple national cap formula based on 1990 emissions (Meyerson 1998b; United Nations 1996). Countries with high population growth as a result of natural increase or immigration, such as Australia, the United States, and Canada, will simply have a more difficult time curbing their emissions. This raises the general question as to whether the Kyoto agreement is equitable from a per capita viewpoint, and whether high-growth countries can practically reach their targets.

The use of demographic projections at Kyoto

De Sa’s second point is that projections of future population growth may have in fact influenced the outcome of the Kyoto negotiations, at least indirectly. This may well be true, since demographic information and projections must have been available to each country’s negotiating team. What is less clear is whether population projections were used to systematically analyze the long-term effects of national caps. While demographic projections may, as he says, have been “implicitly include[d]” under greenhouse gas “business-as-usual” emissions scenarios, this is a far cry from thoughtful analysis. Most likely, the driving forces behind the scenarios were economic, not demographic projections. And as is particularly apparent in this year of international economic tur- moil, it is probably unwise to formulate long-term political and environmental commitments around economic projections for individual countries, whose fortunes can change almost overnight. Demographic projections, while far from infallible, may be relatively more reliable and less variable, at least over the short run. Moreover, in this case, their use more directly addresses the equity issue.

Click to return to Table of Contents F REDERICK A. B. MEYERSON 807

Looking ahead, the newly released 1998 population projections indicate that the divergent demographic trends that make the Kyoto national cap formula problematic are likely to continue (United Nations 1998). For instance, according to the medium variant projection, by 2050 the US population is projected to exceed 349 million, a 37 percent increase from 1990. In contrast, the populations of Germany, Japan, and the Russian Federa- tion are projected to decline by 8 percent, 15 percent, and 18 percent, respectively. Meanwhile, key developing countries such as China, Brazil, Mexico, and India would see their populations rise by 28 percent, 65 percent, 76 percent, and 80 percent respectively between 1990 and 2050. These projected trends lead back to a central question—how to design an equitable, truly global climate treaty that flexibly responds to demographic change.

Developing a climate treaty based on per capita standards

In general terms, how could a per capita–based climate treaty be designed? First, a global greenhouse gas emissions cap and a desired future trajectory for that cap would be determined, on the basis of scientific evidence and technical feasibility. The average global per capita emissions level necessary to achieve this cap would be determined simply by dividing the global population into the cap (Meyerson 1998a; Engelman 1994). This per capita standard would decline as population grows, and/or as the global cap becomes stricter. National caps would then be created for countries whose average per capita emissions exceed the global average (or some agreed-upon percentage of global average). Countries would “graduate” into the national cap system as their per capita emissions rise above the benchmark level (or as the global cap declines).2 The system could and should allow flexibility mechanisms, similar to emissions trading, joint implementation, and the clean development mechanism employed by the Kyoto Protocol, so that countries could effectively buy and sell the “rights” to emit carbon up to the per capita emissions cap. Potentially, the national caps could also be adjusted to reflect differences in climate, geography, natural resources, and other factors, as long as these adjustments were effected under the rubric of the global cap. A per capita approach, of course, has both advantages and disadvantages as compared with the Kyoto fixed national cap system. It could, in theory, create an incentive for population growth, since a country’s cap would be based on its population size (Meyerson 1998c). In practice, this potential “moral hazard” is unlikely to be a significant concern. Pronatalist national policies have almost never been effective, and, in any event, the

Click to return to Table of Contents 808 T OWARD A PER CAPITA–BASED CLIMATE TREATY economic and social costs of additional population growth would far out- weigh the slight advantage derived from an increase in the national carbon emissions cap. On the other hand, a per capita approach would create an equitable system for dealing with immigration—countries that accept net immigration would see an increase in their allowable national cap, instead of being penalized, as they effectively are under the Kyoto Protocol. High-emissions developed countries may well not favor a per capita approach because of the economic burdens, but this is true of virtually any plan that has been proposed. This problem could be at least lessened through a sliding scale or “convergence” toward a unified standard, whereby countries would be expected to reduce their per capita emissions by a certain percentage over time.3 While it is inescapable that countries with high per capita emissions, including the United States, would have to either make major reductions or pay for the right to emit, a convergence plan would allow that choice to be exercised gradually over a reasonable time. In addition, these countries are also the best equipped technologically and finan- cially to meet their obligations. It is possible that some developing countries may not favor this per capita approach, since it creates potential future mandatory national caps for all countries. However, these caps take effect only after a country has “graduated” past a certain level of per capita emissions and, one would assume, has achieved a higher level of economic development. Countries below this cap would acquire valuable rights salable on the international market, which they retain as long as they keep their emissions below the global per capita standard. A per capita–based system has a number of clear advantages. First and foremost, it is equitable, because rights and responsibilities are based on per capita use of the global atmosphere (i.e., through greenhouse gas emissions). It does not lock in a “right” to some particular historical level of emissions, as the Kyoto Protocol does. It should be relatively easy to establish and periodically adjust national caps under a per capita system. Each unit of greenhouse gas emissions, whether emitted or avoided, could have the same value internationally and thereby produce efficient market signals. Issues of compliance would still exist, but to a lesser extent than under the Kyoto Protocol, since many of the inscrutable valuation and accounting methodologies and problems surrounding the clean development mechanism would disappear.4 The per capita system would be global from the outset, and therefore would directly address the global emissions/climate problem, which the Kyoto Protocol largely fails to do (Parry et al. 1998). The per capita approach disaggregates the traditional groupings of developed and developing countries, a step that should decrease the usual politics surrounding developing/developed country issues. It deals flexibly with the demographic

Click to return to Table of Contents F REDERICK A. B. MEYERSON 809 change and population shifts that may increase as a result of climate change or political and economic conditions. It could still allow (and probably enhance) flexibility mechanisms such as emissions trading, but it would also create incentives for countries to keep per capita emissions below the point that triggers national caps, or to reduce their per capita emissions to that level. Under this approach, every reduction in per capita emissions would be valuable economically—it would either reduce the amount by which a country has exceeded its cap or create a tradable right, if the country is below its cap. Note also that there would be a general incentive for the global community to stabilize population, since the per capita standard would be affected by changes in world population size. Finally, a per capita approach can be relatively easily communicated to policymakers and the general public (particularly from an equity perspective), something that unfortunately cannot be said of the Kyoto Proto- col. As a result, the per capita approach would facilitate public perception of individual responsibility for carbon emissions reductions, as well as the critical role that population plays in global environmental change.

Notes

The author thanks Robert Engelman, Daniel on Climate Change Working Group III re- Esty, Laura Ahearn Meyerson, William port (Banuri et al. 1996). Nordhaus, Holly Pearson, Steven Sinding, 2 Variations of a per capita/equity ap- William Smith, and Robert Wyman for their proach have been developed by a few nongov- review and helpful comments. ernmental organizations, e.g., the “contraction 1 General discussion of the climate eq- and convergence” proposal of the Global Com- uity issue and the relationship between mons Institute (http://www.gci.org.uk). population and greenhouse gas emissions 3 See note 2. can be found in two publications of Popula- 4 The Kyoto Protocol need not necessar- tion Action International (Engelman 1994, ily be abandoned; it should be possible to de- 1998). A number of arguments for and ap- sign an orderly transition from the Protocol to proaches to the equitable allocation of emis- a per capita–based climate treaty. sion rights and responsibilities are reviewed in Chapter 3 of the Intergovernmental Panel

References

Banuri, T., K. Goran-Maler, M. Grubb, H. K. Jacobson, and F. Yamin. 1996. “Equity and social considerations,” in J. P. Bruce, H. Lee, and E. F. Haites (eds.), Climate Change 1995: Economic and Social Dimensions of Climate Change. Intergovernmental Panel on Cli- mate Change; Cambridge: Cambridge University Press. Carbon Dioxide Information Analysis Center. 1998. Historical Carbon Dioxide Emissions. Oak Ridge, TN: Oak Ridge National Laboratory. Engelman, R. 1994. Stabilizing the Atmosphere: Population, Consumption and Greenhouse Gases. Washington, DC: Population Action International.

Click to return to Table of Contents

810TOWARD A PER CAPITA–BASED CLIMATE TREATY

———. 1998. Profiles in Carbon: An Update on Population, Consumption and Carbon Dioxide Emis- sions. Washington, DC: Population Action International. Intergovernmental Panel on Climate Change. 1995. IPCC Working Group I Summary for Poli- cymakers. Cambridge: Cambridge University Press. Meyerson, F. A. B. 1998a. “Population growth, carbon emissions and climate change: A crucial policy relationship,” in Global Climate Change: Science, Policy and Mitigation/Adap- tation Strategies. Washington, DC: Air and Waste Management Association. ———. 1998b. “Population, carbon emissions, and global warming: The forgotten relationship at Kyoto,” Population and Development Review 24: 115–130. ———. 1998c. “Population, development and global warming: Averting the tragedy of the climate commons,” Population and Environment 19: 443–463. O’Neill, B. C., F. L. MacKellar, and W. Lutz (in press). Population and Climate Change. Cam- bridge: Cambridge University Press. Parry, M. et al. 1998. “Adapting to the inevitable,” Nature 395: 741. United Nations. 1996. World Populations Prospects: The 1996 Revision. New York. ———. 1997. Conference of the Parties, Third Session. Kyoto Protocol to the United Nations Framework Convention on Climate Change. Kyoto, Japan. ———. 1998. World Populations Prospects: The 1998 Revision. New York.

Click to print article Click to returnreturn to to T ableTable of ofContents Contents

DATA AND PERSPECTIVES

Low-Skilled Immigrants and the Changing American Labor Market

MARÍA E. ENCHAUTEGUI

BETWEEN 1980 AND 1994 the number of immigrants—defined as foreign- born persons whose original citizenship is that of another country—in the United States without a high school diploma almost doubled, from 2.8 million to 5.1 million. These immigrants, who represent 35 percent of US immigrants, now account for roughly 30 percent of the total US labor force between the ages of 18 and 55 who do not have a high school diploma. Many low-skilled immigrants do not speak English, have very low levels of education, reside illegally in the United States, and perform menial low-skilled work. In this note, I examine the economic status and economic prospects of low-skilled immigrants in the changing American labor market. While considerable attention has been paid to the plight of low-skilled workers in the changing labor market (Burtless 1990; Kosters 1991; Murphy and Welch 1996; Holzer 1996), this literature has not paid attention to the foreign- born component among the low-skilled. When immigrants are discussed at all, it is in terms of their negative effects on the economic status of US- born low-skilled workers, leaving aside immigrants’ own economic situation (Bound and Johnson 1991; Acs and Danziger 1993). But with immigrants comprising some 30 percent of all workers without a high school diploma, the plight of US low-skilled workers is necessarily tied to the economic fortunes of low-skilled immigrants. Low-skilled immigrants as a group are often seen as a success story: coming to the United States with little human capital, working hard, and finding jobs despite adversity. But the rhetoric of success masks the poor economic performance of low-skilled immigrants in the last decade. Data examined below show that despite their relatively solid employment position and their over-representation in various low-skilled occupational

POPULATION AND DEVELOPMENT REVIEW 24(4):811–824 (DECEMBER 1998)811

Click to print article Click toClick return to returnto Table to of Table Contents of Contents 812 L OW-SKILLED IMMIGRANTS AND THE AMERICAN LABOR MARKET groups, the poverty rates among immigrants without a high school diploma are increasing faster than those of natives. The wages of low-skilled immigrants are declining relative to those of natives, and their economic future is threatened by declines in jobs in some of their key employment sectors. I refer to low-skilled workers as those who do not possess a high school diploma or who have completed less than 12 years of education. Although education is an imperfect proxy for skills, it is an important determinant of economic success, and lack of education makes navigating the changing labor market very difficult (Blackburn, Bloom, and Freeman 1990). In the labor market literature, education continues to be the main divider between skilled and low-skilled workers. The data in this note come from the 1980 and 1990 US Censuses of Population and the March 1994 Current Population Survey (CPS). The CPS included 10,615 civilians 18 to 55 years old, not enrolled in school, with less than 12 years of education; 3,000 of these were immigrants. Although census and CPS data include illegal immigrants, it is impossible to determine immigrants’ legal status. It is likely that a large segment of the low- skilled immigrant population resides in the country illegally. Occupational patterns and policy implications may vary according to the legal status of immigrants, but this issue will not be explored here.

The growing number of low-skilled immigrants

In 1980 in the United States, 2.8 million foreign-born persons aged 15 to 55 and not enrolled in school did not have high school diplomas. By 1994 the number had jumped to 5.1 million (see Table 1). The striking growth in low-skilled immigrant labor contrasts with the sharp decline in US-born workers without a high school diploma. The number of US-born high school dropouts declined from 20 million in 1980 to 13 million in 1994. The growth in immigrant labor coupled with the decline in native labor in the low educational category resulted in immigrants occupying a rapidly increasing share within the low-skilled workforce. In geographic areas of high immigration— where immigrants arriving within the last ten years account for more than 10 percent of the population—low-skilled immigrants make up roughly three-quarters of all low-skilled workers. If the number of US-born high school dropouts continues to decline and the number of immigrants continues to increase at the same rate as in the recent past, foreign-born workers will soon account for the majority of low-skilled labor in the United States. If the decrease in the native low- skilled workforce were to halt, the immigrant proportion could still rise to 50 percent by the year 2020. While both the number of immigrant men and the number of immigrant women are increasing, the growing immigrant representation in the low-skilled labor market is found more among men. Until 1990, immi-

Click to return to Table of Contents M ARÍA E. ENCHAUTEGUI 813

TABLE 1 Number of 18–55-year-olds without a high school diploma by nativity and percent of the foreign-born age 18–55 in selected categories: United States, 1980–94 Nativity 1980 1990 1994

Population (excluding school enrollees) US-born (millions) 20.2 18.1 13.4 Foreign-born (millions) 2.8 4.5 5.1 Population who worked the previous year Foreign-born (percent) 12.4 21.2 29.2 Foreign-born in high-immigration areas (percent)a 48.1 63.0 73.9 Foreign-born among men (percent) 12.0 21.9 31.7 Foreign-born among women (percent) 13.0 20.3 25.5 aHigh-immigration areas are metropolitan statistical areas where at least 10 percent of the population immigrated between 1980 and 1990. SOURCES: US Census Bureau, 1980 and 1990 censuses, 1 percent Public Use Microdata Sample (PUMS); and March 1994 Current Population Survey. grant men and immigrant women were almost equally represented in the gender-specific low-skilled labor forces. In 1980, 12 percent of the low- skilled male labor supply was immigrant, compared with 13 percent of the low-skilled female labor supply. However, after 1990 the gender gap in immigrant representation in the labor force reversed and began to widen. In 1994, 32 percent of low-skilled men were immigrants, compared with 25 percent of low-skilled women. The widening gender gap in immigrant representation within low- skilled labor could have resulted from the high influx of low-skilled immigrants from Mexico during the 1990s. This type of immigration is heavily male. Another explanation for the considerably higher immigrant representation in the male low-skilled labor force in 1994 is that many of the Special Agricultural Workers (SAWs) who legalized under the Immigration Reform and Control Act established residence in the United States in 1990 and thereafter. The SAWs are predominantly low-skilled and largely male. According to critics of US immigration policy, the large number of low-skilled immigrants is a result of US immigration policy’s emphasis on family-based rather than skills-based visas. Almost 60 percent of the immigrants who entered the country legally in 1994 did so under family reunification provisions. Recent policy changes, however, have sought to increase the number of employment-based admissions. The 1990 Immigration Act allocated 140,000 visas per year to employment-based immigrants (Sorensen et al. 1992). In 1995 the Commission for Immigration Reform recommended the elimination of all low-skilled, employment-based immigration. Although it is not driven by labor skill concerns, the recently enacted policy requiring sponsors of immigrants to have incomes exceeding 125 percent of the poverty level may have consequences for the future skill composition of the US immigrant population. Immigrants at the low-

Click to return to Table of Contents 814 L OW-SKILLED IMMIGRANTS AND THE AMERICAN LABOR MARKET est end of the income distribution are usually the least skilled and are likely to request admission for relatives with very low skills. On the other hand, changes in legal immigration policy may have little effect on the number of low-skilled immigrants. The largest share of low-skilled foreign-born persons is from Mexico, and a large share of these immigrants enters the United States illegally. Despite the efforts of the Im- migration and Naturalization Service, illegal immigration remains high. Refugees, who account for another significant share of legal low-skilled immigration, are granted admission for humanitarian reasons. Not even employment-based visas guarantee highly skilled immigrants since 60 percent of all employment visas do not go to direct beneficiaries but to relatives of direct beneficiaries (Sorensen et al. 1992: 29). As noted above, native low-skilled labor declined by over one-third in the last two decades. A gross indicator of demand for low-skilled labor is the total employment of low-skilled workers, since employment represents the demonstrated potential of the economy for labor absorption. Employment of workers without a high school diploma declined by 4.3 million between 1980 and 1994, while the number of these workers among the US-born population declined by 6.8 million. This difference suggests that the sharp decline in the native low-skilled labor supply may have provided circumstances in which immigrant low-skilled labor could be absorbed. Another factor that might have accommodated the growing number of immigrants is the aging of the US population coupled with low post– baby boom fertility rates. The mean age of the US population rose from 31.9 years in 1970 to 34.5 in 1994. Within 13 years, the baby boom cohorts will begin to reach age 65, accelerating the decrease of the labor supply through retirement. At the other end of the age spectrum, low post– baby boom fertility rates have moderated the number of new entrants. The slowing growth in the size of the population of labor force age allowed for an increase in the number of immigrants entering the United States.

The employment niches of low-skilled immigrants

Two-thirds of low-skilled foreign-born persons do not possess any high school education, compared with one-fifth of native-born persons; 20 percent do not speak English. Despite these shortcomings and as shown in Table 2, in 1994 the unemployment rate among low-skilled immigrants was significantly below that of low-skilled natives. The lower unemployment for immigrants may have been related to the fact that many of these immigrants are not eligible for federal transfer programs and therefore must work in order to generate income. Also, immigrants have stronger employment networks than natives, and networks have been found to be more effective than other methods in placing people in jobs (Holzer 1988).

Click to return to Table of Contents M ARÍA E. ENCHAUTEGUI 815

TABLE 2 Unemployment rates (percent) among low-skilled workers by nativity: United States, 1980–94 Location and nativity 1980 1990 1994

United States 11.3 12.8 15.6 US-born 11.6 13.3 16.7 Foreign-born 9.5 11.0 12.9 Difference 2.1 2.3 3.8 High-immigration areas 10.6 13.4 15.4 US-born 11.9 16.6 20.4 Foreign-born 9.1 11.6 13.7 Difference 2.8 5.0 6.7 Low-immigration areasa 11.3 12.4 15.0 US-born 11.4 12.8 16.0 Foreign-born 8.9 9.1 9.0 Difference 2.5 3.7 7.0 aAreas where less than 5 percent of the population are foreign-born persons who arrived in the United States in the last ten years. SOURCES: US Census Bureau, 1980 and 1990 censuses, 1 percent PUMS; and March 1994 Current Population Survey.

Nevertheless, the high level of immigrants’ employment is achieved in the most menial jobs. Only 22 percent of immigrants without a high school diploma can be found in clerical, sales, and skilled blue-collar jobs, compared with 33 percent of the US-born in this category. The concentration of immigrants at the low end of the occupational distribution is also demonstrated by their preponderance in low-skilled occupational groups. Seventeen percent of all low-skilled workers, versus 26 percent of low-skilled immigrant workers, hold jobs in the ten least-skilled occupational groups. I define low-skilled occupational groups by the proportion of workers in that group who lack a high school diploma. Table 3 shows the ten least- skilled occupational groups according to this definition. These data are taken from the 1990 census, since more recent Current Population Survey data do not have large enough sample sizes for the occupational detail required for these tabulations. Low-skilled immigrants are over-represented in the ten least-skilled occupational groups. While the foreign-born accounted for 21 percent of all low-skilled workers in 1990, the last row in Table 3 shows that they accounted for 33 percent of low-skilled workers in these ten occupational groups. The foreign-born account for the majority of low-skilled farm and precision textile, apparel, and furnishings workers. They are also over-represented among textile, apparel, and furnishings machine operators, cleaning and building workers, private household workers, and in related agricultural occupations. I also ranked occupations by size according to the number of workers without a high school diploma and examined the immigrant representa-

Click to return to Table of Contents 816 L OW-SKILLED IMMIGRANTS AND THE AMERICAN LABOR MARKET

TABLE 3 The ten occupational groups with the highest representation of low-skilled workers and percent of foreign-born among the low-skilled: United States, 1990 Foreign-born Low-skilled among low- Degree of over- workers skilled workers representation Occupational group (percent) (percent) of foreign-borna

Farm workers 55.9 51.7 30.5 Textile, apparel, and furnishings machine operators 45.8 34.1 12.9 Forestry 42.4 9.4 –11.8 Private household services 41.7 42.6 21.4 Woodworking machine operators 40.5 10.1 –11.1 Precision textile, apparel, and furnishings workers 40.3 52.0 30.8 Cleaning/building services 39.8 24.6 3.4 Related agricultural occupations 38.8 36.7 15.5 Fishers/hunters/trappers 37.7 10.9 –10.3 Fabricating machine operators 35.4 21.7 0.5 All ten occupational groups 42.8 32.9 11.7 aDegree of over-representation is the difference in percentage points between the proportion of foreign-born among low-skilled workers in the given occupation and the proportion of foreign-born in the overall low-skilled workforce in 1990 (21.2 percent). NOTE: The occupational categories listed in this table belong to a set of 71 narrowly defined categories derived from the 1980 Standard Occupational Classification. SOURCE: US Census Bureau, 1990 census, 1 percent PUMS. tion in these occupations. This exercise allows one to see to what extent immigrants numerically dominate these occupations. The ten largest employers of high school dropouts and the immigrant representation in these occupations are listed in Table 4. These ten occupational groups employ 61 percent of all high school dropouts in the United States. This list is headed by handlers/helpers/laborers, which employed some 1.4 million low-skilled workers in 1990, followed by food services, with 1.2 million. Immigrants are not over-represented when the ten occupational groups shown in Table 4 are considered together. Rather, they are represented exactly according to their proportion in the low-skilled labor market. Separately, immigrants are under-represented in four of the ten occupational groups with the largest number of low-skilled workers: motor vehicle operators, sales workers, mechanics and repairers, and construction workers. Note also that only two of the least-skilled occupational groups listed in Table 3 in which immigrants are most highly over-represented are among the largest employment sectors for low-skilled workers listed in Table 4. It appears that numerically immigrants tend to dominate small to medium size occupations. The least skilled of all occupations are not simply entry occupations for newly arrived immigrants. Rather, low-skilled immigrants tend to re-

Click to return to Table of Contents M ARÍA E. ENCHAUTEGUI 817

TABLE 4 The ten occupational groups with the largest number of low-skilled workers and percent of foreign-born among the low-skilled: United States, 1990 Foreign-born Low-skilled among low- Degree of over- workers skilled workers representation Occupational group (thousands) (percent) of foreign-borna

Handlers, helpers, and laborers 1,383 21.2 0.0 Food services 1,214 27.3 6.1 Construction workers 1,093 19.7 –1.5 Cleaning/building services 996 24.6 3.4 Motor vehicle operators 882 11.7 –9.5 Machine operators, assorted materials 762 27.8 6.6 Retail sales workers 757 13.8 –7.4 Mechanics and repairers 694 15.0 –6.2 Fabricators/assemblers/ hand workers 646 23.1 1.9 Textile, apparel, and furnishings machine operators 539 34.1 12.9 All ten occupational groups 8,966 21.6 0.4 aDegree of over-representation is the difference in percentage points between the proportion foreign-born among low-skilled workers in the given occupation and the proportion foreign-born in the overall low-skilled workforce in 1990 (21.2 percent). NOTE: The occupational categories listed in this table belong to a set of 71 narrowly defined categories derived from the 1980 Standard Occupational Classification. SOURCE: US Census Bureau, 1990 census, 1 percent PUMS. Tabulations were performed using CIESIN’s Ulysses cross-tabulation engine. main in these occupations even after lengthy residence in the United States. If the least-skilled occupations were a point of entry, recent immigrants would be over-represented in these occupations. But, as shown in Table 5, this is not the case. In 1990, 16 percent of all immigrants without a high school diploma had arrived in the United States within the last three years— exactly the proportion found in the low-skilled occupational groups. In fact, the duration of US residence of immigrants in low-skilled occupations matches that of all low-skilled immigrants. Moreover, there is no clear tendency for declining representation in low-skilled occupations with duration of US residence. For instance, among low-skilled immigrants in low-skilled occupations, 13 percent have from four to five years of residence in the United States. Another 13 percent have from 16 to 20 years of US residence. The niches that low-skilled immigrants occupy in menial, small to medium size occupations are important sources of employment even for immigrants who have lived in the United States for more than a few years. These figures raise the question of how immigrants come to dominate certain occupations. Small ethnographic studies indicate that employment networks are one way in which an occupation or industry may be-

Click to return to Table of Contents 818 L OW-SKILLED IMMIGRANTS AND THE AMERICAN LABOR MARKET

TABLE 5 Total number of low-skilled foreign-born and low-skilled foreign- born in the ten least-skilled occupational groups by year of entry to the United States, 1990 Low-skilled All foreign-born in low-skilled Percent ten least-skilled Percent Year of entry foreign-born of total occupational groups of total

1987–90 764,900 16 161,800 16 1985–86 513,800 11 124,300 13 1982–84 487,800 11 112,000 11 1980–81 652,500 14 150,300 15 1975–79 802,100 17 171,600 17 1970–74 637,200 14 126,400 13 1965–69 369,800 8 73,700 7 1960–64 208,500 4 33,900 3 1950–59 166,900 4 26,700 3 Before 1950 32,600 1 6,900 1 Total 4,636,100 100 987,600 100

SOURCE: US Census Bureau, 1990 census, 1 percent PUMS. come dominated by immigrants and, at times, by specific ethnic groups (Grasmuck and Pessar 1991; Grenier et al. 1992; Newman 1994; Waldinger 1995). Since the cost to the provider of a bad referral is high, only the most reliable workers make it through the job networks. Where such networks of immigrants are established, they provide a means of positive selection that perpetuates the hiring of immigrant labor. Labor contracting also contributes to the concentration of immigrants in certain occupations. As was noted above, immigrants are over-represented in agriculture, apparel, and personal service. Labor contracting in agriculture is longstanding. The bracero program in California and the sug- arcane cutter program in Florida (Portes and Rumbaut 1990) are examples of efforts to contract foreign workers into agricultural labor. The apparel industry has historically employed many immigrant workers (Parsons 1990), but it has also used labor recruitment, especially in the Northeastern United States (Maldonado 1979; Gleassel-Brown 1988). The preponderance of immigrants in certain occupations is reinforced by the locational choices of both immigrants and employers. On the one hand, immigrants locate in areas where demand for their labor is highest. On the other hand, employers of low-skilled labor may set up operations in areas where immigrant labor is abundant. Finally, the possibility that the over-representation of immigrants in some occupations and industries is achieved through discrimination against native- born workers, particularly African Americans, cannot be ruled out. Interviews with employers indicate that they rank immigrant workers above African Americans (Kirshenman and Neckerman 1991; Kasinitz and Rosenberg 1994).

Click to return to Table of Contents M ARÍA E. ENCHAUTEGUI 819

While employers tend to give productivity-related reasons for this ranking, many of the attributes described by employers are consistent with stereotypi- cal views that favor immigrants and handicap African Americans.

Prospects in a changing labor market

In the last decade, low-skilled immigrants lost ground in wages and income, and their future employment prospects are in jeopardy because of declines in a number of the sectors where they are concentrated. Between 1979 and 1989, the annual earnings of low-skilled workers fell by 13 percent nationally. Immigrants fared only slightly worse than natives, largely as a result of the higher cost of living (and, consequently, higher wages) in high-immigration areas, where a large share of low-skilled immigrants reside (Enchautegui 1997a). Table 6 compares immigrant real annual earnings to native earnings in 1979 and 1989 for low-skilled natives and immigrants who worked during 1989. Immigrants in areas of high immigration are having difficulty maintaining stable wages. Nationally in 1989, low-skilled immigrants earned 93 percent of the wages of low-skilled natives, on average. At first glance, the wage gap seems remarkably small, considering that immigrants are more likely to be in farm occupations, many lack English-language fluency, a significant portion has been in the United States for only a few years, and many have much less education than low-skilled natives. The relatively small wage gap, like the apparent similarity in wage declines, however, is largely an artifact of immigrants’ geographic concentration in high-wage areas. In 1989, the mean annual earnings for low-skilled workers in high- immigration areas was $14,047; in low-immigration areas it was 8 percent lower at $12,958 (author’s calculations based on the 1989 PUMS). When wage gaps are analyzed by geographic area, immigrant wages are significantly below those of natives, and sizable increases appear in the immigrant-to-native wage gap. Analyzing areas separately according to immigrant concentration, the lowest wage ratio between low-skilled immigrants and natives is found in areas of high immigration. In these areas, low-skilled immigrants earned 79 percent of the wages of low-skilled natives in 1989. Of all metropolitan areas, the lowest wage ratios were found in two metropolitan statistical areas in California with heavy immigrant concentrations: Anaheim and Los Angeles, where low-skilled immigrants earned 67 percent and 72 percent of the wages of low-skilled natives, respectively. Moreover, the ratio of wages between immigrants and natives dropped by 6 percentage points in Los Angeles and by 11 percentage points in Anaheim between 1979 and 1989. The growing hardship experienced by low-skilled immigrants is also reflected in their poverty rates. Table 7 shows that the poverty rates of low-skilled immigrants aged 18 to 55 have increased sharply in the last 15

Click to return to Table of Contents TABLE 6 Foreign-born-to-US-born wage ratios for workers without a high school diploma, by groups of metropolitan statistical areas (MSAs) and in selected MSAs: United States 1979 and 1989 1980 immigrant Location 1979 1989 sample size

US totals 0.94 0.93 19,368 25 largest MSAs 0.85 0.85 12,276 Outside largest MSAs 0.96 0.91 7,092 High-immigration areas 0.84 0.79 9,465 Medium-immigration areas 0.88 0.84 4,143 Low-immigration areas 1.01 0.93 5,760 Selected MSAs High immigration Los Angeles–Long Beach, California 0.78 0.72 3,471 New York, New York 0.88 0.91 2,654 Oakland, California 1.00 0.92 155 Miami, Florida 0.90 0.94 783 Bergen and Passaic counties, New Jersey 0.90 0.93 234 Jersey City, New Jersey 1.02 0.93 247 McAllen–Edinburg–Mission, Texas 0.92 1.05 153 San Francisco, California 0.93 0.74 1,074 San Jose, California 0.89 0.92 211 Anaheim–Santa Ana–Garden Grove, California 0.78 0.67 483 Medium immigration Chicago, Illinois 0.94 0.91 1,363 Washington, DC 0.80 0.80 175 Boston, Massachusetts 1.02 0.94 302 Riverside–San Bernardino–Ontario, California 0.80 0.85 194 San Diego, California 0.78 0.86 317 Newark, New Jersey 0.96 1.05 349 Low immigration Philadelphia, Pennsylvania-New Jersey 0.89 0.88 189 Detroit, Michigan 1.01 1.10 271 Nassau and Suffolk counties, New York 1.00 0.94 225 Houston, Texas 0.87 0.89 521 Dallas, Texas 0.85 0.86 283 Phoenix, Arizona 0.99 0.79 125

NOTES: The individual cities and counties listed in this table are among the 25 largest metropolitan statistical areas, according to 1990 census numbers, plus five additional MSAs with high immigrant concentrations (Bergen and Passaic counties, NJ; Jersey City, NJ; McAllen–Edinburg–Mission, TX; San Francisco, CA; San Jose, CA). High-immigration areas are MSAs where at least 10 percent of the population immigrated to the United States between 1980 and 1990; medium-immigration areas contain between 5 percent and less than 10 percent recent immigrants; low-immigration areas contain less than 5 percent recent immigrants. Low-immigration MSAs (Atlanta, GA; Dallas, TX; Detroit, MI; Houston, TX; Minneapolis-St. Paul, MN; Nassau and Suffolk counties, NY; Philadelphia, PA-NJ; St. Louis, MO-IL) include nonmetropolitan areas (overall less than 1 percent of the population in these areas immigrated between 1980 and 1990). SOURCE: Author’s tabulations based on 1980 and 1990 censuses, 1 percent PUMS. Tabulations were performed using CIESIN’s Explore cross-tabulation engine.

Click to return to Table of Contents M ARÍA E. ENCHAUTEGUI 821

TABLE 7 Poverty rates (percent) among the foreign-born and among the native population: United States, 1980–94 1980 1990 1994

Foreign-born population All persons 15.6 18.3 22.7 Persons aged 18–55 14.7 17.8 21.3 Persons aged 18–55 without a high school diploma 20.6 24.9 36.3 US-born population All persons 12.2 12.7 13.9 Persons aged 18–55 9.8 10.5 10.9 Persons aged 18–55 without a high school diploma 19.2 24.6 30.8

SOURCES: US Census Bureau, 1980 and 1990 censuses, 1 percent PUMS; and March 1994 Current Population Survey. years. The proportion of low-skilled immigrants living in poverty grew from 21 percent in 1980 to 36 percent in 1990. The poverty rates of the low- skilled native population also grew substantially, although not at the same rate as those of immigrants. More evidence about the poor economic prospects of immigrants re- lates to living arrangements. Overcrowded housing has increased in southern California, mainly among immigrant families (Myers and Lee 1996). Enchautegui (1997b) found that poverty rates in Latino neighborhoods are closely linked to recent immigration flows and that the penalty associated with being a recent immigrant grew substantially between 1979 and 1989. As with natives, the growing impoverishment of immigrants is likely to be related to the generally poor prospects of low-skilled workers in the labor market. To this disadvantage is added the growing number of low- skilled immigrants, since the wages of recent immigrants are sensitive to the competition posed by other immigrants (LaLonde and Topel 1991). The economic prospects of low-skilled workers may deteriorate further as we enter the new century. The occupations that are growing involve contact with the public. Immigrants’ lack of English fluency, very low levels of education, and at times lack of legal documentation hinder their ability to occupy these positions. In his study of labor demand based on interviews with employers in four large cities, Holzer (1996) found that over half of the jobs for non-college graduates require in-person or telephone contact with the public on a daily basis. Frequently, these jobs require English-language skills and some require high school education. Holzer also reports that the expanding occupations require the ability to read para- graphs and roughly two-thirds require daily use of arithmetic. According to projections for the year 2005 by the US Department of Labor (1994), many of the occupations in which immigrants are concen-

Click to return to Table of Contents 822 L OW-SKILLED IMMIGRANTS AND THE AMERICAN LABOR MARKET trated are not expanding. Agricultural jobs of the type immigrants perform are expected to decline by 10 percent because of mechanization. Operator, laborer, and fabricator jobs in which immigrants are most concentrated are expected to decline by anywhere from 10 percent in machine setter jobs to 20 percent in textile operator jobs. A decrease of one-third in private household employment is expected. Finally, cleaning and building services employment, another stronghold of low-skilled immigrants, is expected to increase by 19 percent. But at a 19 percent growth rate, cleaning and building services employment is the slowest growing occupation in the expanding service sector.

Policy implications and discussion

Research has pointed to the difficulties American low-skilled labor experienced in recent decades and to predicted future problems. Undoubtedly this bleak outlook is particularly likely to hold true for immigrants without a high school diploma. Immigrants are often praised for their work ethic, their resilience, and their ability to find jobs. They have carved niches in occupations where low levels of human capital are acceptable. The unemployment rate of low- skilled immigrants is well below that of natives. But work ethic and resilience may not be enough for procuring employment and a decent wage in an economy that demands more highly skilled labor. Low-skilled immigrants’ future ability to find employment is challenged by declining employment in many of the jobs where immigrants are concentrated and by the expanding range of occupations that require contact with the public, established legal status, and basic arithmetic and reading skills. Support networks have served immigrants well in finding employment. But because these networks are also concentrated in declining occupations where very low levels of education and poor command of English do not bar entry, they are unable to open the door to mainstream low-skilled jobs. Surely many immigrants will find employment in the growing service sector, but these jobs will continue to be the least-skilled of all low-skilled jobs. Recently enacted policies affecting immigrants, such as welfare reform’s exclusion of working-age noncitizens from the Food Stamp Program, enforcement of public charge rules, and status verification, may drive low- skilled immigrants more deeply into the fringes of the labor market as they try to make a living in a society that is increasingly inhospitable to them. Welfare reform rules also closed some potential avenues of intervention to improve the poor economic circumstances of immigrants. Legal adult noncitizens cannot receive food stamps, future legal immigrants prior to naturalization cannot benefit from a wide range of federal programs, and states have discretion to exclude such immigrants from some social programs.

Click to return to Table of Contents M ARÍA E. ENCHAUTEGUI 823

The increased impoverishment of low-skilled immigrants can be addressed through a combination of immigration and immigrant policies. Le- gal immigration policy controls the number and the quality of legal immigrants entering the country. Immigration policy also encompasses the enforcement of immigration law along border zones. Reducing the number of low-skilled immigrants entering the United States may help prevent further deterioration of wages and employment since immigrants’ economic status is most affected by competition from other immigrants in the labor market. An active human capital development policy may help with the incorporation of low-skilled immigrants into the labor market. Certainly, programs aimed at improving English-language skills, including reading and writing, as well as arithmetic, should be at the center of this agenda. But other approaches are also needed, such as the diversification of job search methods for low-skilled immigrants to direct qualified persons to occupations that are not immigrant-dominated. Attention should also be given to improving immigrants’ access to union jobs and their ability to obtain trade licenses since immigrants are not well represented in skilled blue-collar occupations. Finally, promoting high school completion is crucial in a labor market where the prospects for persons without high school degrees are bleak. Employers must take part in efforts to upgrade the skills of immigrants since they are immediate beneficiaries of such efforts and have come to rely on immigrants to fill job vacancies. Some employers are helping to improve the skills of immigrants by offering English-language instruction or assistance with immigration papers for their employees (Yang 1996). As the supply of native low-skilled labor continues to diminish, immigrants will be entering employment sectors where employers have little or no experience with immigrant labor. Immigrant labor will also be seen outside of areas where immigrants have historically settled. Newly emerging immigrant employers need to be educated concerning the employment of immigrants, such as their work culture, the paperwork involved in hiring resident aliens, and the penalties for hiring undocumented immigrants.

Note

This research was conducted while the au- Andrew W. Mellon Foundation. The author thor was a Senior Research Associate at the gratefully acknowledges the research assis- Urban Institute. It was funded by the Immi- tance of Aaron Sparrow and the comments gration and Naturalization Service and by the of Michael Fix, both of the Urban Institute.

References

Acs, Gregory and Sheldon Danziger. 1993. “Educational attainment, industrial structure and male earnings through the 1980s,” Journal of Human Resources 28: 618–648.

Click to return to Table of Contents

824LOW-SKILLED IMMIGRANTS AND THE AMERICAN LABOR MARKET

Blackburn, McKinley L., David E. Bloom, and Richard B. Freeman. 1990. “The declining economic position of less skilled American men,” in Burtless 1990. Bound, John and George Johnson. 1991. “Wages in the United States during the 1980s and beyond,” in Kosters 1991. Burtless, Gary. 1990. A Future of Lousy Jobs? The Changing Structure of US Wages. Washing- ton, DC: The Brookings Institution. Enchautegui, María E. 1997a. “Immigration and wage changes of high school dropouts,” Monthly Labor Review 120, no. 9: 3–9. ———. 1997b. “Latino neighborhoods and Latino neighborhood poverty,” Journal of Urban Affairs 19, no. 4: 445–468. Glaessel-Brown, E. 1988. Immigration Policy and Colombian Textile Workers in New England. Report to the US Department of Labor. Boston: Journeylines. Grasmuck, Sherri and Patricia R. Pessar. 1991. Between Two Islands: Dominican International Migration. Berkeley: University of California Press. Grenier, Guillermo J., Alex Stepick, Debbie Draznin, Aileen LaBorwit, and Steve Morris. 1992. “On machines and bureaucracy: Controlling ethnic interaction in Miami’s apparel and construction industries,” in Louise Lamphere (ed.), Structuring Diversity: Eth- nographic Perspectives on the New Immigration. Chicago: University of Chicago Press. Holzer, Harry. 1988. “Search methods use by unemployed youth,” Journal of Labor Econom- ics 6, no. 1: 1–20. ———. 1996. What Employers Want: Job Prospects for Less-Educated Workers. New York: Russell Sage Foundation. Kasinitz, Philip and Jan Rosenberg. 1994. Missing the Connection: Social Isolation and Employ- ment on the Brooklyn Waterfront. Working Paper, Michael Harrington Center for Demo- cratic Values and Social Change, City University of New York. Kirshenman, Joleen and Kathryn M. Neckerman. 1991. “We’d love to hire them, but . . . : The meaning of race for employers,” in Christopher Jencks and Paul E. Peterson (eds.), The Urban Underclass. Washington, DC: The Brookings Institution. Kosters, Marvin H. 1991. Workers and Their Wages. Washington, DC: AEI Press. LaLonde, Robert J. and Robert H. Topel. 1991. “Labor market adjustments to increased immigration,” in John M. Abowd and Richard B. Freeman (eds.), Immigration, Trade, and the Labor Market. Chicago: University of Chicago Press. Maldonado, E. 1979. “Contract labor and the origins of Puerto Rican communities in the United States,” International Migration Review 13, no. 1: 103–121. Murphy, Kevin M. and Finis Welch. 1993. “Industrial change and the rising importance of skill,” in Sheldon Danziger and Peter Gottschalk, Uneven Tides: Rising Inequality in America. New York: Russell Sage Foundation. Myers, Dowell and Seong Woo Lee. 1996. “Immigration cohorts and residential overcrowding in southern California,” Demography 33, no. 1: 51–65. Newman, Katherine. 1994. “Working poor, and low wage employment in the lives of Harlem youths,” manuscript, Department of Anthropology, Columbia University. Parsons Carol A. 1990. “Immigration and regional labor market performance in the U.S. apparel industry,” Immigration Policy and Research Working Paper 7, US Department of Labor, Bureau of International Affairs. Portes, Alejandro and Rúben G. Rumbaut. 1990. Immigrant America: A Portrait. Berkeley: University of California Press. Sorensen, Elaine et al. 1992. Immigrant Categories and the US Job Market: Do They Make a Dif- ference? Urban Institute Report 92-1. Washington, DC: The Urban Institute Press. US Department of Labor. 1994. Occupational Projections and Training Data. Bulletin 2451. Waldinger, Roger. 1996.“Who makes the beds? Who washes the dishes? Black/immigrant competition reassessed,” in Harriet Orcutt Duleep and Phanindra V. Wunnava (eds.), Immigrants and Immigration Policy: Individual Skills, Family Ties, and Group Identities. Green- wich: JAI Press. Yang, Catherine. 1996. “Low-wage lessons: How Marriott keeps good help—even at $7.40 an hour,” Business Week 11: 108–116.

Click to print article Click to return to to T Tableable of of Contents Contents

ARCHIVES

Francis Hutcheson on the Rights of Society

In the Enlightenment tradition “rights” came chiefly to mean human rights, repos- ing in individuals. Initially few and general—in a classic enumeration, life, liberty, and the pursuit of happiness—human rights have multiplied in modern times through international declarations, convenants, and conventions. Several of these rights concern demographic matters, notably “reproductive rights.” The Universal Declaration of Human Rights, adopted in December 1948 (its 50th anniversary being celebrated in 1998), states in Article 16: “Men and women of full age . . . have the right to marry and to found a family.” The Proclamation of Teheran, issued by the International Conference on Human Rights held in Teheran in 1968, states that “Parents have a basic human right to determine freely and responsibly the number and the spacing of their children” (Paragraph 16). The World Population Plan of Action formulated at the 1974 Bucharest conference changed “parents” to “all couples and individuals” and added the words “and to have the information, education and means to do so” (Paragraph 14(f)). This is also the phrasing that appears as Principle 8 in Chapter 2 of the Programme of Action of the 1994 Cairo conference. The qualifier “responsibly” must presumably be read as a restriction on the right to reproduce, limiting the freedom to choose to decisions made with reference to a larger set of interests. Those interests are spelled out in the Bucharest statement in these terms: “the responsibility of couples and individuals in the exercise of this right takes into account the needs of their living and future children, and their responsibilities towards the community.” Similar language, with “society” added to “community,” is found in the recommendations of the 1984 Mexico City conference (Para- graph 26). Consideration of the obligations of individuals to society—in other words, the rights of society over its members—is a complementary line of thought to human rights, older but in recent times far less emphasized. (The comparative neglect of such obligations and of the need for a balance between the two sets of rights has been a criticism of the modern human rights movement.) This is the subject treated

POPULATION AND DEVELOPMENT REVIEW 24(4):825–829 (DECEMBER 1998)825

Click to print article Click toClick return to returnto Table to of Table Contents of Contents 826 A RCHIVES in the passage below by Francis Hutcheson, a prominent member of the Scottish Enlightenment. Hutcheson was born in Drumalig, Ireland, in 1694, of a Scots-Irish family. He graduated from the University of Glasgow, where from 1730 to his death in 1746 he was professor of moral philosophy. A renowned teacher (with Adam Smith as his best-known student), his Glasgow lectures, unusual in being delivered in En- glish rather than Latin, covered ethics, jurisprudence, government, and political economy. They were published posthumously as A System of Moral Philosophy (Glasgow, 1755). The excerpt below is the first half of Book II, Chapter 16: “Con- cerning the general rights of human society, or mankind as a system.” Hutcheson distinguishes between “perfect” and “imperfect” rights of society: the former, where compulsion may be justified; the latter, “to be left generally to the prudence and virtue of persons concerned.” Thus, society (“mankind as a system”) is accorded the right to insist that parents raise their children responsibly to be productive members, but individuals themselves have only a moral duty to cultivate their own talents. The first two perfect rights that Hutcheson lists are in essence demographic: preventing suicide and ensuring social reproduction. In both cases the arguments are purely pragmatic: it is in the society’s interests. In the reproduction case, Hutcheson concedes that attempting to compel people to have children would be unwise, but it is acceptable to make “celibacy burthensome and dishonourable.” The excerpt omits the last part of the chapter, discussing imperfect rights. The orthography has been modernized in minor respects. A facsimile edition of Hutcheson’s collected works was published by Georg Olms Verlagsbuchhandlung, Hildesheim, Germany (7 volumes, 1969–71).

Hitherto we have considered the rights and obligations peculiarly respect- ing certain individuals, and constituted for their happiness, in consistence with and subserviency to the general interest. But as we not only have the narrower kind affections, and a sense of duty in following their motions by good offices to individuals, and by abstaining from what may hurt them, but more extensive affections toward societies and mankind; and an higher sense of obligation to do nothing contrary to any publick interest, there are many obligations of a more extensive kind upon men to consult the general interest, even where no particular person is more concerned than others; and mankind, as a system, seems to have rights upon each individual, to demand of him such conduct as is necessary for the general good, and to abstain from what may have a contrary tendency, tho’ the wrong conduct no more affects one individual than another. Of these rights and obligations some are of the perfect kind, where compulsion may be just; others are of a more delicate obligation, not admitting compulsion, where our duty must be left to our own prudence and sense of virtue. Of the perfect kind are these following.

Click to return to Table of Contents F RANCIS HUTCHESON 827

I. As each individual is a part of this system, the happiness and duration of which depends on that of its part; as every one may be of some service to others in society, were it only by advice and example, if they have such dispositions as they ought to have: as we are formed by nature for the service of each other, and not each one merely for himself; each one is obliged to continue in life as long as he can be serviceable, were it only by an example of patience and resignation to the will of God; when no important interest requires his exposing his life to dangers. Human society has a right by force to prevent attempts of suicide from any unreasonable dejection, or melancholy, or chagrin; and these general rights of all, each one as he has opportunity, by what assistance he can obtain, may justly execute. No other bond is requisite to intitle a man to interpose in such cases, but the common tye of humanity. Did such immoderate pas- sions prevail; were suicide deemed a proper method to escape from the ordinary vexations of life, or from the severe chagrin upon disappointments, and were there no restraining sense of duty; many of the bravest spirits might rashly throw away those lives, which might have become joyful to themselves, and useful and ornamental to the world. Mankind have a right to interpose against such rash designs. II. Another general obligation on the individuals toward the system, respects the continuing the human race. Such as are not hindered from domestick cares by some more important services to mankind, seem obliged to contribute their part to this important purpose, by bearing their shares of the burden of educating human offspring, if they are in such circumstances as enable them to do it. This duty however, must in a great measure be left to the prudence of individuals: it would scarce be wise in any society to compel them to have offspring, whether they desired it or not; tho’ there may be strong reasons for making celibacy burthensome and dishonourable. In this matter, and that above-mentioned of preserving our own lives, nature by powerful instincts implanted has pretty generally se- cured the performance of our duty. Such instincts however are so far from preventing or superseding the notion of moral obligation, that they strongly establish it, and are the most direct indications of it. But as to such as desire offspring, or are the occasions of their coming into being, as they are under most sacred obligations, indicated abundantly by the parental affection, to give support and education to their offspring, and to form them into proper members of society; so mankind as a system, and every society, have a right to compell them to discharge these offices. They have a like right to prevent such ways of propagation as would make a proper education impracticable, by leaving the fathers uncertain, and thus casting the whole burden of it upon the mothers. ’Tis the interest of mankind to prevent such abuses, even tho’ the deluded mothers voluntarily yielded to the ensnaring solicitations. Of this more hereafter.

Click to return to Table of Contents 828 A RCHIVES

Mankind as a system have a like right to prevent any perversions of the natural instinct from its wise purposes, or any defeating of its end. Such are all monstrous lusts, and arts of abortion. III. There is a like right competent to mankind, and an obligation on each one, that nothing useful to men should be destroyed or spoiled out of levity or ill-nature; tho’ no individual, or society, has a more special right in it than another: that nothing useful should be destroyed, even by those who had acquired property in it, without any subserviency to some plea- sure or conveniency of theirs. To do so must evidence an envy or hatred of the rest of mankind. Thus fountains should not be stopped or poisoned; no useful products of the earth, which many stand in need of, should be ca- priciously destroyed by the owners, because they have abundance other ways for their use: no noxious creatures should be brought in, and turned loose in places formerly free from them. IV. ‘Tis also the right and duty of the system which each one should execute as he has opportunity, to assist the innocent against unjust vio- lence, to repel the invader, to obtain compensation of damage done, and security against like attempts for the future. Without this right all the en- joyments of men would be very precarious, since few could confide in their own strength to repel the combined forces of any cabals of the injurious. As the example of successful injuries tends to invite others to like practices, ’tis requisite for the general good that this bad influence be counter- acted as often as possible, by the inflicting of such evils upon the injurious, as by their terror may overballance in their minds, and those of others who may have like dispositions, all allurements to injustice from the hopes of secrecy and impunity. This is the foundation of the right of punishing, which, as we said above, men have in natural liberty, as well as in civil polity. The mistakes and inconveniences which may more frequently attend the execution of this right in natural liberty, do not take away the right, but shews the great use of civil polity. V. A like right we may justly assert to mankind as a system, and to every society of men, even before civil government, to compel any person who has fallen upon any fortunate invention, of great necessity or use for the preservation of life, or for a great increase of human happiness, to di- vulge it upon reasonable terms. The inventor, no doubt, may have a right to make large advantages of what his diligence, ingenuity, or felicity has discovered; and is justly intitled to a compensation for this advantage, in proportion to the good it brings to society, or the labour which the invention may have cost him, or the profit he could have made by it. But if a man is exorbitant in his demands, or so inhuman as not to employ his discoveries where they are wanted, or will always retain the secret to himself, so that it must perish with him; if the matter appears to be of great importance to mankind, a society has a right to compel him to arbitration

Click to return to Table of Contents

FRANCIS HUTCHESON829 about the proper compensation to be made for the discovery; and to force him, upon just terms, to make it. VI. ’Tis justly also reputed a right of human society to compel each person to such labours and industry as he is capable of, that he may not be an unnecessary burden upon the charity or compassion of the industrious: to compel parents so to educate and habituate their children, that they may be able to support themselves, while they continue in health. Enough is allowed to the natural liberty of mankind, and the parental power, that men may chuse occupations as they incline for themselves and their children. But as the universal diligence of all is plainly requisite for the good of mankind, each one is bound to his share of it, and is justly compelled to it, unless he can give sufficient security to the society that he shall be no burden to it. A like right any society has to exercise the parental power over orphans, in educating them to some useful industry, and obliging them to such reasonable services after they are adult, as may defray all prudent expences made upon them, and for their behoof, during their minority. VII. One may justly reckon among these rights of human society that one also of preserving a just veneration toward the dignity of our kind, and preventing any practices which would tend to make it despicable in the judgment of the vulgar; or which would have a tendency to introduce savage and inhuman dispositions. The dead carcases of men can have no rights, and yet ’tis plain men cannot so entirely separate in their thoughts the body from its former inhabitant, as to be unaffected with the treatment of the body. All nations have expressed their affectionate or grateful remembrance of the man, by some rites or other deemed respectful toward the body: and repute some sorts of treatment of dead bodies as inhuman and savage, expressing either hatred and contempt of the man, or a disrespect in general for our fellow-creatures. Every society therefore should prevent such practices as would incourage savage manners, or such as are the usual testimonies of contempt and hatred, toward any who have not by their vices incurred the just abhorrence of all good men. VIII. These and many of like nature we may count the perfect rights of mankind. There are others of the imperfect kind, to be left generally to the prudence and virtue of persons concerned, to which correspond the general duties or offices incumbent on each individual toward mankind in general, previous to any special tye. The same general maxim holds in these rights which we observed about the rights of individuals, viz. “The fulfilling or not-violating the perfect rights, argues rather mere absence of odi- ous vices, than any laudable degree of positive virtue: whereas the matter of praise, and the natural evidence of high virtue consists in a strict regard to the imperfect rights.”

Click to print article Click Clickto return to returnto Table to of Table Contents of Contents Click to return to Table of Contents

BOOK REVIEWS

The Mismeasure of Nations: A Review Essay on the Human Development Report 1998*

IAN CASTLES

The annual Human Development Report, published for the United Nations Development Programme (UNDP) by the Oxford University Press, is promoted by the UNDP as “the only comprehensive guide to global human development available on the market” (UNDP 1998c). Since the series was launched in 1990, these reports have been extraordinarily influential. Ac- cording to Amartya Sen, speaking at a memorial meeting for the originator of the series, Mahbub ul Haq, on 15 October 1998, the Human Development Report has “had a profound effect on the way policy makers, public servants and the news media as well as economists and other social scientists view social and economic advancement.” Professor Sen went on to describe the Report as “one of the major sources of information and understanding of the social and economic world” (Sen 1998). The release of each Human Development Report has become a significant global event. The 1998 Report was launched on 9 September at The Hague City Hall and “in more than 100 capitals worldwide.” In support of the Report’s theme of “Consumption for human development,” visitors to the launch and the associated events “were able to purchase ecologically and socially-sound products at a consumer’s market; sit in on small work- shops and discussion groups on consumption issues; and visit displays on development and the environment by dozens of non-government organizations” (UNDP 1998d).

*Review of United Nations Development Programme, Human Development Report 1998. New York: Oxford University Press, 1998. xii + 228 p. $34.95; $19.95 (pbk.).

POPULATION AND DEVELOPMENT REVIEW 24(4) (DECEMBER 1998)831

Click to print book review Click Clickto return to returnto Table to of Table Contents of Contents 832 PDR 24(4) B OOK REVIEWS

In all, Human Development Report 1998 was published in ten languages: English, French, German, Italian, Portuguese, Spanish, Catalan, Arabic, Japanese, and Russian. The associated Human Development 1998 Database on diskettes facilitates the compilation of tailored reports on the “demographic, social, economic and environmental issues of human development.” Prospective users are advised to “simply select one or more of 353 social indicators, select from 174 countries and/or 28 regional aggregates and run the report.” With unconscious irony, the 1998 Report refers to the difficulties for local and national producers that arise from “the opening of consumer markets with a constant flow of new products”:

Market research identifies “global elites” and “global middle classes” who follow the same consumption styles, showing preferences for “global brands.” There are the “global teens” . . . inhabiting a “global space,” a single pop-culture world, soaking up the same videos and music and providing a huge market for designer running shoes, t-shirts and jeans. (UNDP 1998a: 6)

This review will argue that the Human Development Report is itself the quintessential “global brand” of the 1990s, and that its dominant position in the global market for information on the social and economic world owes little to its intrinsic qualities and much to the packaging and promo- tional efforts of its multinational sponsor.

Statistical misunderstandings and errors

For every reader of the Human Development Report, there are thousands who read newspaper items purporting to summarize its contents. As few jour- nalists can be expected to read and comprehend a 240-page report, most of the articles are taken from news releases produced in the UNDP’s Public Affairs Department. The role of these news releases is well illustrated by the “media advisory” for an invitation-only advance briefing on the Human Development Report 1998, which was hosted by the UNDP Administrator at a New York restaurant just one week before the global launch at The Hague. The release was headed “De- spite record-high global consumption, ranks of poor, hungry and homeless are growing in rich nations as well as poor”; and it claimed that “homelessness and illiteracy in industrialized countries continue to rise, according to the 1998 Human Development Report . . . ” (UNDP 1998b). Perhaps no one noticed, but the Report did not say this. All of its claims about poverty and illiteracy in the industrial countries related to the situation at a point in time, not to trends over time. Reference was made to

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 833 rising levels of malnutrition in Eastern Europe and the Commonwealth of Independent States, but few readers would have thought of these countries as “rich nations.” The key observation about hunger was unexcep- tionable if platitudinous: it “is not as pervasive” in the industrial countries as in the developing countries (UNDP 1998a: 27). By way of contrast, the key observation about homelessness in the industrial countries was obviously wrong. In the 1997 volume it had been argued that “The most extreme housing deprivation is to have no home, and worldwide, an estimated 100 million are homeless” (UNDP 1997a: 29). In the 1998 Report, however, it was said that “around 100 million are thought to be homeless” in the developing countries alone and that “more than 100 million are homeless” in the industrial countries (UNDP 1998a: 2, 25, 27). Taken together with population data, these figures imply that the homeless represent only 2 percent of the population of the developing countries, but 8 percent of the population of the industrial countries. Clearly one or both of these percentages must be wrong. There are well-known methodological problems in measuring homelessness, and most studies agree that the number of “homeless people” at a particular time is not a satisfactory indicator. Alternative approaches have yielded more carefully stated conclusions: for example, “In New York nearly a quarter of a million people, more than 3% of the city’s population, have stayed in a shelter at some point in the past five years” (UNDP 1997a: 29). But the claim in Human Development Report 1998 is that more than 100 million people in the industrial countries are homeless (“a shockingly high number amid the affluence”), not that this number of people have experienced homelessness (UNDP 1998a: 27). In the same list of bullet points, under the heading “Measuring human poverty in industrial countries,” the 1998 Report asserts that “Unem- ployment among youth (age 15–24) has reached staggering heights, with 32% of young women and 22% of young men in France unemployed, 39% and 30% in Italy and 49% and 36% in Spain” (UNDP 1998a: 27). All of these figures are wrong, the correct percentages being far lower than those cited. For 1995—the reference year in the table from which the figures are drawn—9 percent of young women and 7 percent of young men in France were unemployed. In Italy the figures were 14 percent and 12 percent, and in Spain 18 percent and 16 percent. The errors relating to youth unemployment in this year’s Report can be attributed to inadequate documentation in the “Human Development Indicators” section of the Report itself. In 1998, for the sixth successive year, this section included statistics of the “youth unemployment rate” in industrial countries (UNDP 1998a: 192); but, as in earlier Reports, the compilers failed to mention that the rates relate unemployment to the labor force in

Click to return to Table of Contents 834 PDR 24(4) B OOK REVIEWS the relevant age/sex group, not to the population in that group. As a result, it is likely that many readers of the Reports, and many of the users of the Human Development 1998 Database, will have made the same mistake as this year’s authors. The ILO, the OECD, and many national statistical offices are striving to remove common misunderstandings about the scale and nature of youth unemployment in the industrial countries that the UNDP is helping to create. Additional sources of confusion have been introduced in the 1998 Report with the unveiling of “the new human poverty index (HPI-2).” Ac- cording to the authors, this index “shows conclusively that under-consumption and human deprivation are not just the lot of poor people in the developing world” (UNDP 1998a: 2). One element of HPI-2 is a measure identified as “People not expected to survive to age 60 (% of total population) 1995” (UNDP 1998a: 28, 186). This indicator does not represent the percentage of those living in 1995 who could not expect to live until the age of 60, but the percentage of a hypothetical population born in 1995 who would not survive to age 60 if that population were subject at each age to the mortality rates of 1995. Again the authors of the Report appear to have misinterpreted their selected indicator. They say that “Nearly 200 million people [in the industrial countries] are not expected to survive to age 60” (UNDP 1998a: 27), this being the product of the population of the industrial countries (about 1200 million) and the proportion of persons not surviving from age 0 to age 60 at 1995 mortality rates (16 percent). The proportion of persons not surviving from their present age will, of course, be much lower. And the number of people in the industrial countries who are not expected to survive from their present age to age 60 is about 100 million, not 200 million. In any case, the key point is that mortality rates are decreasing at all ages. During the twentieth century, expected survival rates from birth to age 60 in such countries as Britain and France have increased from about 45 percent to about 90 percent (Keyfitz and Flieger 1968: 324, 525; UNDP 1998a: 28, 186). In these countries, the chance of survival from age 0 to age 60 is now greater than was the chance of survival from age 0 to age 1 in 1900. Of course, there are still “millions” who will not survive to age 60, but it is somewhat perverse to present this as a manifestation of “human deprivation.”

The human development index: Crude measure or policy guide?

The first Human Development Report (1990) introduced a new measure of human development. Indicators of life expectancy, educational attainment, and income were combined into a composite “human development index” (HDI). In his speech at the memorial meeting for Mahbub ul Haq in Octo-

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 835 ber, Amartya Sen recalled his initial doubts about trying “to catch in one simple number a complex reality about human development and deprivation.” But Sen explained that he came to accept Haq’s view that the HDI was valuable “as an instrument of public communication.” This “deliberately-constructed crude measure” was a means of “getting the ear of the world through the high publicity associated with [its] transparent simplicity . . . ” (Sen 1998). But the proponents of the HDI saw other benefits in the “transparent simplicity” of the index. It was argued in the first Report that “Too many indicators could produce a perplexing picture—perhaps distracting policymakers from the main overall trends”; and that “Having too many indicators in the index would blur its focus and make it difficult to interpret and use” (UNDP 1990: 11, 13). The world was encouraged to believe that this simple device not only enabled “people and their governments to evaluate progress over time” but also permitted the determination of “priorities for policy intervention” (UNDP 1994: 91). Successive reports promoted the UNDP’s special new product. In the second Report (1991), it was claimed that the HDI was “a reliable measure of socio-economic progress” (UNDP 1991: 15). Readers were referred to a technical note in which the “robustness” of the HDI had been established by statistical tests:

Another way to check the index is to try a different weighting. Suppose we take the product of the three measures and then take the cube root—that is, the geometric rather than the arithmetic mean. This is the same as the equally weighted sum of the logarithms of all three variables. Taking the log of life expectancy, the log of literacy and the log of income . . . —and arranging them—gives a new index that also has an extremely high rank correlation with the original HDI. . . . (UNDP 1991: 88)

The conclusion was that “The HDI, simple as it is, stands up as a robust device for making intercountry comparisons” (UNDP 1991: 89). In the 1995 Report, readers were assured that “the HDI has been found to be a robust measure: even when different weights are tried, the country HDI values do not change significantly” (UNDP 1995: 122). Many development economists were less impressed. Allen Kelley argued in this journal that the HDI “offers only limited insights beyond those obtained by small modifications to simple measures of economic output” (Kelley 1991: 315). And in An Inquiry into Well-being and Destitution, Partha Dasgupta judged the index to be “not much good; it has too many unappealing properties [and] . . . is also excessively partial” (Dasgupta 1993: 107). There are grounds for believing that the vigorous promotion of the merits of the HDI has encouraged a simplistic approach to public policy

Click to return to Table of Contents 836 PDR 24(4) B OOK REVIEWS and administration in many developing countries. In the 1995 Report, it was said that “Its rankings [had] opened healthy competition among countries to improve their human development status”—that is, their HDI score (UNDP 1995: 119). And the 1998 Report records that President Fidel Ramos of the Philippines had “asked the National Statistical Coordination Board to include the human development index regularly in the system of statistics used to track variations across provinces,” and had “directed the De- partment of Interior and Local Government to closely monitor provincial and municipal human development indices—and to institute rewards for good performance” (UNDP 1998a: 18). As a “deliberately-constructed crude measure” that was devised as a means of “getting the ear of the world,” the HDI may not be a satisfactory indicator of the performance of provincial administrators. For reasons that will be explained below, it may also be an unsatisfactory tool for monitoring progress and determining priorities for policy intervention.

The HDI in practice: Country examples

Madagascar, Mauritius, and Tanzania

From the outset, the creators of the HDI reified the complex concept of “human development” as a single number. A country’s human development performance was related to its potential by means of a crude comparison of its relative ranking in “league tables” of HDI rank and GNP per capita. In the HDI table in each Human Development Report, countries were categorized according to whether their “HDI rank” was above or below their “GNP rank.” In this year’s Report, two island-countries in the Indian Ocean are fea- tured in the discussion of these relative rankings. Madagascar, which is almost as large as Texas, is one of four low-income countries that are said to “rank higher on the HDI than on GDP per capita (PPP$ [purchasing power parity dollars]), suggesting that they have converted economic prosperity into human capabilities very effectively”—an achievement described as “noteworthy.” Mauritius, 1000 kilometers to the east of Madagascar and smaller than Rhode Island, is one of four “more affluent” countries whose “ranking on the HDI is lower than that on GDP per capita (PPP$), suggesting that they have failed to translate economic prosperity into correspond- ingly better lives for their people.” This is described as “particularly dis- turbing” (UNDP 1998a: 20). The following table, which is drawn entirely from statistics in the Hu- man Development Report 1998, provides an alternative perspective on the development record of these two countries:

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 837

Madagascar Mauritius

Land area (‘000 ha) 58,154 203 Total real GDP ($PPPbn) 10 15 Real GDP per capita ($PPP) 673 13,294 Life expectancy at birth (years) 58 71 Under-5 mortality rate (per 1,000 live births) 164 23 Population without access to safe water (%) 66 2 Population without access to health services (%) 62 0 Population without access to sanitation (%) 59 0 Children not reaching grade 5 (%) 72 1 Adult illiteracy rate (%) 54 17

In 1955, Sir Arthur Lewis identified Mauritius as a country that faced problems in achieving development because its population was “much too large in relation to agriculture, and at the same time much too small to support a wide range of industrial development” (Lewis 1955: 324). Almost 40 years later, another eminent development economist, Partha Dasgupta, pointed out that Mauritius had been placed first in a “ranking of living standards data of 48 of the world’s poorest countries” and commented that “Her presence at the top of our ranking was a revelation to me” (Dasgupta 1993: 113). Lewis would not have believed it possible that Mauritius would achieve a level of real GDP per head 20 times that of Madagascar. While Mauritius has succeeded beyond all expectation, Madagascar has continued to disappoint. The country is regarded by the World Bank as having “enormous potential for growth,” with “abundant and varied natural resources.” But as a result of its failure to turn these resources to advantage, it is now

. . . among the poorest countries in the world. Poverty has increased and deepened substantially over the past two and a half decades, with real per capita income decreasing 40 percent between 1971 and 1991. The poverty assessment estimates that 70 percent of the population are poor and 59 percent are extremely poor (World Bank 1997: 88–90).

It is little short of tragic that the “team of eminent economists and distinguished development professionals” who prepare the Human Development Report have told the world that poverty-stricken Madagascar has turned its “economic prosperity into human capabilities very effectively”; and that affluent Mauritius has “failed to translate economic prosperity into corre- spondingly better lives for [its] people.” Regrettably, these examples from this year’s Report are part of a consistent pattern of misleading commentary on the relative performance of

Click to return to Table of Contents 838 PDR 24(4) B OOK REVIEWS different countries. In the 1995 Report it was claimed that “A look at the highest positive and negative differences between HDI and real GDP per capita ranks shows clearly that such countries as Costa Rica and Madagas- car used their economic growth to enhance the lives of their people” (UNDP 1995: 22). The fact was that Madagascar had had no economic growth to “use”: between 1965 and 1996, its GNP per capita had decreased at an average rate of 2.0 percent per annum (World Bank 1998: 24). But the authors of the Reports had become so enmeshed in the scrutiny of “positive and negative differences” that they had lost sight of realities that in the absence of the HDI would have been painfully obvious. The 1996 Report quoted with approval a statement by Julius Nyerere, who had been President of Tanzania for over 30 years: “Every proposal must be judged by the criterion of whether it serves the purpose of development—and the purpose of development is the people” (UNDP 1996: 46). Notwithstanding their leader’s rhetoric, the people of Tanzania are among the very few in the world who are even poorer than those of Madagascar, and their life expectancy at birth is seven years shorter. Yet Tanzania is also in the list of four low-income countries who are complimented in this year’s Report for their noteworthy achievement in converting their “economic prosperity into human capabilities very effectively.”

Oman

In the early Reports, oil-rich Oman was seen as the worst performer in the “negative difference” stakes. With an HDI rank 56 places below its GNP rank, the first Report (1990) listed it first among a group of countries “that . . . have yet to translate their income into corresponding levels of human development” (UNDP 1990: 16). In the second Report (1991), Oman was included in a group of “countries whose HDI rank is 20 or more places lower than their per capita income rank, showing that they have considerable potential to improve their human development levels—by spending their incomes better and planning their investment priorities more wisely” (UNDP 1991: 15). And in the third Report (1992), Oman was again singled out for its alleged failure to translate its high income into a commensurate HDI:

. . . Oman has a per capita income two and a half times that of Costa Rica, but its literacy rate is one-third of Costa Rica’s, its average life expectancy is nine years less and its child mortality rate is two and a quarter times higher. All this can be reduced to the information that Oman has an HDI of 0.589, while Costa Rica’s HDI is 0.842—43% higher. (UNDP 1992: 20)

The issues of principle that are raised by figuring of this kind will not be pursued here, except to note that they are not disposed of by tests of “ro-

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 839 bustness,” such as comparisons of results with the cube root of the product of three component measures. But there is also a practical question that has received little attention in the debate over the HDI: the quality of the underlying data. When the index first appeared in 1990, the estimated expectation of life at birth in Oman in the reference year (1987) was given as 57 years— less than in China, India, Pakistan, Brazil, Thailand, Philippines, Turkey, or Iran. It is now estimated that life expectancy in Oman in 1985–90 was in fact 68 years—greater than in all of these countries (United Nations 1998). The charge that Oman had failed to translate its income level into corresponding levels of human development was made on the assumption that the country’s average life expectancy was in the lowest quartile of the world’s population: we now know that it was in the highest quartile. The original calculation of Oman’s “deprivation” with respect to life expectancy in 1987, on a scale of 0 to 1, was 0.60; on the estimates now available, Oman’s “deprivation,” for the same year and calculated on the same basis, would have been 0.24. The first HDI calculations assumed that the adult literacy rate in Oman in 1985, on the basis of an estimate made by the UNDP itself, was 30 percent. UNESCO has now published an estimate that Oman’s adult literacy rate in 1993 was 60 percent (UNESCO 1998a). This implies that the rate in 1985 was far higher than the 30 percent which the UNDP estimated in the 1990 and 1991 Reports; and that the country’s level of adult literacy in the 1990s has been substantially higher than the figure of 35 percent that was published in every Human Development Report from 1992 to 1997 (and that went into the calculation that Costa Rica’s HDI was 43 percent greater than Oman’s). The estimates of “mean years of schooling” were even more astray, and here the errors are inexplicable. The HDI calculations in the 1992 Re- port put the “mean years of schooling” in Oman at only 0.9 years, compared with 5.7 years in Costa Rica. These figures were clearly inconsistent with UNESCO data, which showed that Oman had established a system of universal primary education. In fact, the UNESCO Statistical Yearbook 1992 (UNESCO 1993) gave the gross enrollment rates at primary level in 1990 as: Oman, 103 percent (up from 60 percent in 1980) and Costa Rica, 102 percent (down from 105 percent in 1980). At the secondary level the rates were: Oman, 54 percent (up from 14 percent in 1980) and Costa Rica, 42 percent (down from 48 percent in 1980). At the time of the 1992 Human Development Report, Oman’s per capita income was two and a half times that of Costa Rica. By 1996, Oman’s margin of advantage over Costa Rica had been reduced to one-third (World Bank 1998). It was becoming difficult to suggest that Oman had failed “to translate [its] income into corresponding levels of human development,”

Click to return to Table of Contents 840 PDR 24(4) B OOK REVIEWS as had been alleged in the 1990 Report. This was not because Oman had suddenly begun “spending [its] income better and planning [its] investment priorities more wisely,” as the authors of the 1991 Report had urged (1991: 15). It was because circumstances beyond control had diminished the country’s income (thereby reducing the “negative difference” between its HDI and GNP rankings). The crude technique of cataloguing these positive and negative differences implicitly assumes that a country’s progress in enhancing life expectancy and literacy levels can be sensitive to such temporary phenom- ena as rises and falls in the price of oil. It should always have been obvious that this assumption was invalid, and specifically in the case of Oman the 1997 Report effectively acknowledged that the judgments made in earlier reports had been unsound:

Beginning in 1970, Oman undertook a comprehensive programme of human development, achieving some of the most rapid advances ever recorded. Life expectancy has increased by 30 years. . . . Improvements in education have been even more impressive. . . . Most health problems associated with poverty and lack of schooling have been controlled or eradicated. . . . Ad- vances in health and education have been accompanied by rapid advances in other areas of human development. . . . Oil revenues . . . made possible such rapid progress and such a high standard of living. But without the commitment to human development, Oman might have been wealthy but unhealthy. Oman has been a global pace-setter in human development. (UNDP 1997a: 28)

Only five years earlier, the originator of the Human Development Report, Mahbub ul Haq, had singled out Oman as the prime example of a country that was “wealthy but unhealthy”:

. . . beyond the confusing maze of GNP numbers, beyond the curling smoke of industrial chimneys, beyond the endless fascination with budget deficits and balance of payments crises—it is people who matter. Costa Rica has a per capita income only one-third that of Oman but its literacy rate is three times higher, its life expectancy ten years longer, and its people enjoy a wide range of economic, social and political freedoms. (Haq 1992: 1)

Costa Rica

As this extract exemplifies, Costa Rica has always been seen by the HDI team as the prime example of a country that has become healthy without needing to be wealthy. The 1990 Report listed it first in a table headed “Top 15 countries in democratic human development” and asserted that “Costa

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 841

Rica shows that assigning a high priority to social sector expenditures, coupled with well-structured across-the-board policies, can dramatically improve the human condition despite only moderate growth and a poor distribution of income” (UNDP 1990: 16, 51). The 1991 Report proclaimed that “Costa Rica remains an outstanding example of human development in Latin America—and the world” (UNDP 1991: 59). And, as already noted, the 1992 Report put Costa Rica’s HDI at 43 percent higher than Oman’s. In the 1994 Report, Costa Rica was one of seven listed countries whose “HDI rank is far ahead of their income rank, showing that they have made more judicious use of their income to improve the capabilities of their people” (UNDP 1994: 95). The similar comment on Costa Rica and Mada- gascar in the 1995 Report has already been cited. Both the 1996 and 1997 Reports singled out Costa Rica and Viet Nam as countries that had “effectively translated the benefits of economic growth into improvements in the lives of their people”; and the 1996 Report also identified Costa Rica, together with Canada, China, and Sri Lanka, as “countries that convert income into capabilities more effectively than others, countries that constitute the ‘human development frontier’ of efficiency” (UNDP 1996: 31, 67; 1997a: 46). This year’s Report gives special praise to Costa Rica as “a global leader in environmental sustainability.” In a box in the chapter “Agenda for Ac- tion,” the country’s “coherent package” of environmental protection policies is outlined. These policies appear to rely heavily on public relations activities:

The hundred cleanest companies in Costa Rica are named annually, and a green seal of quality is given to gas stations with the best records in preventing air and water pollution and in treating waste water. A red stamp is for those with the worst records. The government and civil society also apply moral suasion by using ad campaigns to convince people that a healthy environment is good in itself, contributes to human well-being and is good for tourism. (UNDP 1998a: 99)

Summaries of Costa Rica’s own human development reports (UNDP 1997b), which are the outcome of a project sponsored by a consortium of national and international institutions (including UNDP, the European Union, and the Council of State Universities), take a quite different view of Costa Rica’s environmental record. The first report, published in 1995, was critical of the fact that “the country’s economic growth [had] relied heavily on dep- redation of the country’s natural endowment”; that “agroecological limitations of natural resources have not been taken into account”; and that “even worse, the nation’s developmental goals have been reached, in good part, at the expense of those resources.” With the country’s primary forest cov-

Click to return to Table of Contents 842 PDR 24(4) B OOK REVIEWS erage reduced from 56 percent of its area in 1960 to 22 percent in 1990, the report warned that “If this pace continues, Costa Rica will exhaust its primary forests within eight years. . . . ” Comments in other sections of this report, and in a second report published in 1996, did not support the glowing views of Costa Rica’s human development performance espoused in successive Human Development Re- ports. The country reports pointed out that “one in five children between 5 and 11 works and 78% of the youth [labor force] have only an incomplete primary education”; that 62 percent of all heads of households did not fin- ish primary school; that “nearly 25% of adolescents between 12 and 17 years old work with almost no job skills”; that “children and adolescents watch TV an average of 6.7 hours per day, while receiving less than 5 hours of school”; and that “In spite of governmental public relations campaigns, the majority of the people thought that the current administration’s performance was ‘bad’ or ‘very bad’” (UNDP 1997b). Remarkably, the Human Development Reports for the UNDP have not drawn upon these comments in Costa Rica’s own human development reports. In this year’s Report, for example, no mention is made of the finding that Costa Rican children spend far more time watching television than in school, although this was highly relevant to a section of the Report headed “Information imbalances,” which referred to findings that children in Ja- pan and the United States spent almost as much time watching television as in school (UNDP 1998a: 64). As “one of the major sources of information and understanding of the social and economic world” (to use Amartya Sen’s phrase), one might expect the Human Development Report to provide the same details about information imbalances in the “outstanding example of human development in Latin America” as it does for countries such as Japan and the United States.

Korea

This year’s Report includes a chart headed “Similar HDI, different income,” which shows graphically that the HDIs of Costa Rica and the Republic of Korea are virtually the same, while Korea’s real GDP per capita is almost twice as great. According to the explanatory text, the chart shows that “The link between economic prosperity and human development is . . . neither automatic nor obvious” (UNDP 1998a: 20). As in earlier Reports, the authors have reified the abstract concepts “economic prosperity” and “human development” as GDP and HDI numbers and have thereby assumed that the link between HDI and human development (unlike the link between GDP and human development) is automatic and obvious. In fact, the view that Korea and Costa Rica are at a similar level of human development would not be entertained, but for the fact that the near-identity of their HDIs gives it a superficial plausibility.

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 843

Between 1965 and 1996, Korea’s average rate of growth of GNP per capita was 7.3 percent per annum, the fastest rate of growth that has ever been achieved by a major economy over a 30-year period. As Costa Rica’s per capita GNP growth rate over the same period was a very modest 1.2 percent per annum (World Bank 1998), it appears that Korea’s level of real per capita income increased between the mid-1960s and the mid-1990s from less than one-third to almost double that of Costa Rica. Because the HDI calculations “reflect the diminishing returns to trans- forming income into human capabilities” (UNDP 1990: 12), Korea’s much higher per capita income raises its HDI by only 0.009 relative to that of Costa Rica. There are several reasons why a near-doubling in income would have a much larger impact on enlarging people’s choices than this small difference in the HDI implies. Perhaps the most important is that the con- comitant decrease in the amount of time that people need to spend in order to earn any given level of income provides them with the opportunity to choose more leisure, an aspect of human welfare that is not taken into account in the HDI. According to estimates prepared by Andrew Harvey for the United Nations (United Nations 1995), Koreans—both women and men—devote less time to work and household chores, and have more “free time,” than their counterparts in more than 20 other countries for which statistics are available (including all of the main industrial countries). The rate of development in the educational attainment of Korea’s adult population has been without precedent for a major country. Between 1960 and 1990, the average total years at schools and higher institutions of the population aged 15 and older increased from 4.25 years to 9.94 years. As a proportion of the average for the OECD countries, average years at school of the population aged 15 and older rose from 60 percent to 110 percent. The proportion of Korea’s population aged 15 and older who had been at secondary school or higher rose from 20 percent in 1960 (compared with 38 percent for the OECD total) to 76 percent in 1990 (compared with 63 percent for OECD) (Barro and Lee 1993). As noted above, the proportion in Costa Rica was far lower (UNDP 1997b). The much higher level of educational attainment in Korea than in Costa Rica is disregarded in the HDI, except to the extent that it is reflected in a fractionally higher level of adult literacy. According to UNESCO estimates (1998b), average length of schooling in Korea had reached 14.5 years in 1995, which was higher than in Swe- den (14.3 years) and Switzerland (14.0 years). Average length of schooling in Costa Rica was 10.3 years in 1995, which was lower than in all OECD countries except Turkey and also lower than in many other countries of Central and South America—including Peru (12.4 years), Chile (11.8 years), Cuba (11.3 years), Dominican Republic (11.2 years), Brazil (11.1 years), Jamaica (11.0 years), Trinidad and Tobago (10.8 years), Venezuela (10.5 years), and Colombia (10.5 years). The far higher level of average time in

Click to return to Table of Contents 844 PDR 24(4) B OOK REVIEWS school in Korea vis-à-vis Costa Rica is not fully reflected in the “gross enrollment ratios” component of their respective HDIs. The only element of the HDI in which Costa Rica ranks higher than Korea is “life expectancy at birth,” which is five years greater in Costa Rica. This has a huge impact on the relationship between the HDIs of the two countries (0.027). But the infant mortality rate, which is disregarded in the HDI because it “is almost perfectly correlated with life expectancy” (UNDP 1995: 121), is far lower in Korea (6 infant deaths per thousand births, compared with 13 per thousand in Costa Rica).

Conclusion

The review of the HDI by Allen Kelley, published in this journal in 1991 and cited earlier, was entitled “The Human Development Index: ‘Handle with care’”(Kelley 1991). Eight more Human Development Reports and eight more HDI “league tables” have demonstrated the need for even greater care in handling the index—and more circumspection by the UNDP in promoting the usefulness of its product.

References

Barro, Robert J. and Jong-Wha Lee. 1993. “International comparisons of educational attainment,” Journal of Monetary Economics 32, no. 3: 363–394. Dasgupta, Partha. 1993. An Inquiry into Well-being and Destitution. Oxford: Clarendon Press. Haq, Mahbub ul. 1992. Human Development in a Changing World. UNDP Occasional Paper 1. New York. Kelley, Allen C. 1991. “The Human Development Index: ‘Handle with care,’” Population and Development Review 17, no. 2: 315–324. Keyfitz, Nathan and Wilhelm Flieger. 1968. World Population. Chicago and London: Univer- sity of Chicago Press. Lewis, W. Arthur. 1955. The Theory of Economic Growth. London. George Allen and Unwin. OECD. 1997. Education at a Glance: OECD Indicators. Paris. ———. 1998. Quarterly Labour Force Statistics Number 2. Paris. Sen, Amartya. 1998. “Mahbub ul Haq: The courage and creativity of his ideas,” speech at the Memorial Meeting for Mahbub ul Haq, 15 October. UNDP website. UNDP. 1990. Human Development Report. New York: Oxford University Press. ———. 1991. Human Development Report. New York: Oxford University Press. ———. 1992. Human Development Report. New York: Oxford University Press. ———. 1993. Human Development Report. New York: Oxford University Press. ———. 1994. Human Development Report. New York: Oxford University Press. ———. 1995. Human Development Report. New York: Oxford University Press. ———. 1996. Human Development Report. New York: Oxford University Press. ———. 1997a. Human Development Report. New York: Oxford University Press. ———. 1997b. “Indicators and democratic governance: Measuring sustainable human development in Costa Rica,” UNDP website. ———. 1998a. Human Development Report. New York: Oxford University Press. ———. 1998b. Media Advisory. “Invitation-only press luncheon on 2 September to pre- view UNDP’s 1998 Human Development Report,” UNDP website.

Click to return to Table of Contents

BOOK REVIEWSPDR 24(4)845

———. 1998c. “About the Human Development Reports,” UNDP website. ———. 1998d. “Human Development Report: News and Events,” UNDP website. UNESCO. 1993. Statistical Yearbook 1992. Paris. ———. 1998a. Statistical Yearbook 1997. Paris. ———. 1998b. World Education Report 1998. Oxford: UNESCO Publishing. United Nations. 1995. Statistics and Indicators on the World’s Women. New York. ———. 1997. The Sex and Age Distribution of the World Populations: The 1996 Revision. New York. ———. 1998. World Population Prospects: The 1996 Revision. New York. World Bank. 1997. Poverty Reduction and The World Bank: Progress in Fiscal 1996 and 1997. Washington, DC. ———. 1998. World Development Indicators 1998. Washington, DC.

Click to print book review Click Clickto return to returnto Table to of Table Contents of Contents 846 PDR 24(4) B OOK REVIEWS

Styles of Population Economics: A Review Essay*

MARK PERLMAN

Professional economics is no longer a simple extension of what was taught years ago in Economics 101. During the past decades academic economics has been changing both with regard to its corpus—that is, its body of directed thought—and with regard to its self-image. Economists at the great American universities are now far more interested in developing the discipline’s scientific persona than in studying and solving social problems. The institutional or historical approach to economic interactions, long considered to be the most interesting part of the discipline because it focused on tangible issues, is no longer deemed to be at the cutting edge. Today the interactions between the institutions essential to fulfilling the functions of landlords, capitalists, entrepreneurs, and workers are denominated in ab- stractions, that is, algebraic symbols: k and r (capital and interest), l and w (labor and wages), and so forth. These functions are so expressed in order to be easily manipulated through mathematical (purely logical) transformations. Articles in the professional economics journals now read much like articles in the physics journals of the past century. They are full of equations, identifying and specifying competing models. Generally these models have two common characteristics: rationality (logic) and a fundamental assumption called “methodological individualism,” which posits that all choices, including production and consumption decisions, are explainable by individual preferences. These preferences can be more or less informed, but the inescapable factor is that what is chosen is chosen by a single mind, which may or may not operate as a tabula rasa. Results are generally seen in terms of static analysis, much like Newton’s mechanics. And as is the case with Newtonian mechanics, their goal is equilibrium or stability. Thus what constitutes modern economics is a divorce of economic study from what has been a more general social science concern. Indeed, so powerful is the attractiveness of this change that other social science

*Review of Mark R. Rosenzweig and Oded Stark (eds.), Handbook of Population and Family Economics. Amsterdam and New York: Elsevier, 1997. 1343 p. in two volumes. $155.00.

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 847 disciplines want to emulate this effort to become a “hard” science, for they too admire firm constructs. Not all economists march to this drumbeat. Some are still interested in the psychology of group choice and in historical institutions that both create and reflect that psychological framework. Usually their preferred approach is empirical—that is, it depends upon observation, both past and present. They seek observed regularities from which they can infer plausible (not necessarily logical) explanations. This approach is, of course, empiricism; its practitioners are called empiricists. Historically the first social science to use this empirical approach was demography. The earliest systematic demographic study in the English language was presented in 1662 by John Graunt in collaboration with Will- iam Petty, one of the first empiricist economists. Thus the marriage between demography and economics has lasted for more than three centuries, although the word “demography” itself is of much later coinage. I shall return to the history of this union below. The aims of both groups of economists (usually self-denominated theorists and empiricists) are purportedly the same. Each thinks its products offer the best path to understanding and even prediction. Both groups coun- sel patience because they realize that their techniques require improvement. The theorists start with simple models and over time seek to replace them with sophisticated versions. The empiricists seek continually to improve their data, making them more accurate and increasingly relevant.1 The conflict between practitioners of the two styles has existed for centuries. But in recent decades it has sharpened because many demographic economists have discovered a preferred basic abstract model explaining human choice. Developed by Gary Becker first when he was at Columbia University and then transplanted to the University of Chicago’s economics department when he moved there, it goes by the name of the new home economics. Its key assumption is that the individual is rational, and alone is able to choose what is best for him or her. And while what is best is seen as a product of economic efficiency—the largest ratio between benefits and costs—this school does not explain what the individual personally values most. If he or she has chosen what seems to others to be irrational, that choice can be explained by the fact that people value things differently. Marriage occurs when each of the parties figures that the match is as good as he or she can get. If later one of the parties becomes ill, then the question is whether the other values fidelity more than convenience. In this way not only are partner choices explained, but also the mainte- nance of such choices even after the original understandings have been upset. It is all personal calculation, conscious or subconscious.2 Rationality is now assumed to be at the heart of economics. Because he fully understood this, Vilfredo Pareto, in anyone’s book one of the greatest minds in economics, chose eventually to abandon economics, saying

Click to return to Table of Contents 848 PDR 24(4) B OOK REVIEWS that it answered few, if any, social questions. Pareto then turned to sociology, a discipline that tried to explain outcomes in terms other than logic. Pareto’s giving up the turf, however, failed to convince many who called themselves economists. Many economists of a more historical bent, like early this century Max Weber in Germany and Wesley Clair Mitchell and John R. Commons in America, asserted that the logical approach had no exclusive claim to ownership of the discipline of economics. Instead they asserted that good economics is what does the most good for economies rather than for economists. Time would show whether one or the other approach was more useful; indeed, time might even show that they were both useful. All of the essays in the two hefty volumes under review are concerned with studying human behavior in the various aspects of production and consumption and refer to the population angle, too. Some of the essays are methodologically monothematic, confined to one approach or the other, but more of them consider and even try to combine the two styles. Whatever one’s own predilections, in the 21 essays in the Handbook the editors have chosen to present what most graduate students expect, namely, a preponderance of the logical approach. But it is much to the credit of the editors that they have added major empirical studies, some questioning the quality of demographic and economic data. Rosenzweig and Stark’s 16-page introduction offers a neutral and sweeping survey of each essay’s highlights, and it is a fine place to start the demand- ing task of surveying the current state of the art of a 300-year-old literature.

The survey

As I have already observed, no modern economist has had a greater impact on today’s demographic economics than Gary Becker, since the 1960s the Christian Dior of intellectual styling.3 It is important to repeat that Becker’s methodological individualism is not predicated on homo economicus being a pecuniary maximizer; rather the principal tenet is self-determination. An individual may be altruistic, but only on his own terms. He may also be self-centered as in Becker’s Rotten Kid model. He or she marries and may stay married on the basis of a felicific calculus. And while the first basis of decisionmaking is rational individualism, the second basis is full knowledge, with recontracting if and when additional information surfaces.

The family

The essay by Theodore Bergstrom (University of Michigan) comprehen- sively surveys the spectrum of modern mainstream models, stemming from Becker’s theories of logically efficient family decisionmaking, ex ante investment in, and ex post accumulation of human capital. Families are seen

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 849 as production and consumption units, living in a world of private and public goods. Decisionmakers make choices with regard to their personal desires involving both their own and others’ attitudes toward selfishness and altruism, often but not always with regard to an optimal aggregate choice. These choices involve logical consideration of issues such as degrees of pa- ternalism and altruism, interpersonal competitiveness, and even the logical aspects of nonmonogamous arrangements. Bergstrom’s use of data to confirm his theoretical expectations leads to ambiguous results (see pp. 56– 57), but as his intention is different from confirming the theory empirically, that may not be a criticism which he would consider valid. In the second essay, Yoram Weiss (Tel-Aviv University) takes up marriage, separation, and divorce decisions—again from a modeled (economic) efficiency standpoint. He is comprehensive. He puts into professional economic terminology most, if not all, of the non–purely subjective elements relevant to the topic, focusing on decision-optimality. Of course, many people who are personally interested in the subject will be disappointed if they are looking for a specific treatment of their own experiences. Con- trary to what Tolstoy observed in the opening sentence of Anna Karenina, Weiss and most economists now seem to argue that all unhappy families are unhappy in their own way; happy families are in equilibrium. Jere Behrman (University of Pennsylvania) reports on intrahousehold distributions. While somewhat repetitive of the previous two essays, he adds an empirical dimension, mostly for industrializing countries. Behrman builds on a static model involving differences in personal capacity as well as information about the empirical outcomes of these differences. He finds some regularities confirming that most American parents bequeath equal amounts to each of their children, although in a significant number of families the lower-earning children are left more than are the higher-earning ones. But here the devil associated with empirical studies raises its familiar head—data on gifts given during the lifetimes of the donors are not collected (pp. 168–169). The flaw in the regularities found may stem from deficiencies in the relevant data. John Laitner (University of Michigan) writes on extended family and other interhousehold relationships, from the standpoints of both same-time and inter-temporal financial transactions. The foundation of the chapter concerns modeling, but he, too, offers some empirical findings. Particu- larly interesting to this reviewer are his observations about household shared consumption as differentiated from the shared pecuniary legacies or transfers, and his conclusions about varying terms of trade between households.

Fertility

Beginning just after World War II and continuing for decades, the best way to obtain research grants was to study the causes and consequences of high

Click to return to Table of Contents 850 PDR 24(4) B OOK REVIEWS fertility rates. This focus continued to be rewarded long after these rates peaked in the late 1950s or early 1960s, and what became a precipitous fall in industrialized countries began. Indeed, one remarkable thing about the essays in this section is the paucity of mention of two current international concerns: (1) the expected major future reductions in population totals in many industrialized countries unless policies encouraging immigration are adopted, and (2) the growing imbalance between the number of elderly economic dependents and the number in the active labor force. But given the foregoing reservation, to this reviewer’s taste Bernard Van Praag (University of Amsterdam) and Marcel Warnaar (National Insti- tute for Family Finance Information, Amsterdam) have written a short but interestingly balanced essay (drawn from others’ empirical studies) on the costs of raising children. Initially they present a “simple” model excogitated on their own, but they extend it with a more sophisticated statement integrating the model with some of the empirical facts. Their efforts to incorporate subjective elements are well detailed. One conclusion worth noting is that the full costs of children can be effectively measured only when they are fed at subsistence levels (pp. 263–264)—the kind of assumption that they, as competent scholars, think too small potatoes to take seriously. They conclude their contribution with appendixes that chart equivalence scales from various sources, some of which date back to the nineteenth century. The next essay, “The economics of fertility in developed countries,” by V. Joseph Hotz (University of Chicago), Jacob Alex Klerman (RAND), and Robert Willis (University of Michigan) starts with presentation of Ameri- can vital statistics and the usual interpretations of them. Included are period and cohort fertility data, percentage of childless women (1913–94), age-specific fertility rates as well as probabilities of first, second, and third births by age of women, and trends in nonmarital childbearing (1963–92). There are tables on contraceptive practices and on labor force participation of wives (with husbands present) by age of the youngest child. Having presented these data, the authors then fit them into the standard Beckerian models concerning the value of a woman’s time and the assumed cost of a “full-time child equivalent,” itself a function of income and other variables. The authors lay out their mathematical models, some of which have seem- ing empirical significance. But in the end they note that the theories are “much better developed” than the regularities derived from examination of the data (p. 342). This conclusion is the dilemma for many demographic economists: if the theory does not fit the data, then so much the worse for which—the theory or the data? T. Paul Schultz (Yale University) offers a readable, basically statistical study of the demand for children in low-income countries. Even more than the immediately preceding essay, it is an exemplar of empirical studies preceding theoretical treatments, for Schultz seeks regularities rather than logi-

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 851 cal rigor. His conclusions highlight institutional as well as individual factors. Schultz apparently lives comfortably with his refined observations. In this sense he is in the minority, but he follows faithfully the Kuznets empiricist tradition.

Mortality and health

Even during the first two post–World War II decades it was generally perceived that the improvements in life expectancy initially seen in the data for eighteenth- and particularly nineteenth-century industrialized countries were becoming generalized. The causes of improvements were often attributed to three factors: (1) better medical knowledge (though upon subsequent examination it became clear that, except for smallpox, few medical treatments went beyond the traditional rest and time); (2) purer drink- ing water and improved sanitary conditions; and (3) better nutritional standards for the mass of the population. But few realized until well after World War II what the plethora of revolutionary drugs would do to lower morbidity rates and to raise life expectancy worldwide. Even fewer realized the attendant social costs of these improvements in terms of pensions and health benefits. In his essay, something of a chef d’oeuvre, but one aimed at the Nobel laureate’s public speech audience rather than at the profession, Robert Fogel (University of Chicago) offers an empirical survey of the impacts of nutritional and other factors on selected European and American countries during the past two and a half centuries. In addition to paying attention to wars, epidemics, and other disease data, Fogel relies upon changes in adult height and weight, culled from military enlistment records, as proxies for measuring nutritional wellbeing. The essay, including the author’s use of older materials, is gratifyingly readable and informative. Yet, had it been expanded and refocused for a more professional audience, it could have contributed much more than it does. Kenneth Wolpin’s (University of Pennsylvania) “Determinants and consequences of the mortality and health of infants and children” is methodologically interesting. Addressing questions about the effect of infant and child mortality on fertility, and then the determinants of infant and child mortality themselves, Wolpin goes on to speculate whether data-oriented econometricians, as they become better able to handle their estimating techniques, will not reveal that data-driven theorizing will be a better path to understanding than logical modeling (p. 553). The essay on “Mortality and morbidity among adults and the elderly,” by Robin Sickles (Rice University) and the late Paul Taubman (University of Pennsylvania), offers some vital statistical data, but is mostly a survey of recent secondary materials. It ends on a hortatory note about the route that “demographic, epidemiologic, biological, and socio-economic paradigms

Click to return to Table of Contents 852 PDR 24(4) B OOK REVIEWS must take in order for the study of mortality and morbidity of adults and the elderly to bear fruitful results” (pp. 626–627).

Migration

The topic of population migration can suggest several things to economists. Some see in it the flow of labor from less-efficient to more-efficient occupations. Others speculate that migration is as much concerned with improved consumption as it is with more-efficient production. And still others see in migrant populations inherent social costs associated with increased demands for housing in short supply, as well as a need for increased social investment involving schools and health and recreational needs. And all of these programmatic issues tend to be simple when they are compared with possible cultural frictions. Michael Greenwood (University of Colorado) provides a sweeping examination of the evolution of the historical migration literature, involving prin- cipally Britain and the United States. The characteristics and result of as well as the reasons for internal migration have attracted scholars for more than a hundred years, with one of the first published econometric studies appearing in the initial 1938 issue of the Oxford Economic Papers (Makower, Marschak, and Robinson 1938, 1939a, and 1939b). Some of Greenwood’s findings are well known, for example that the educated move more than the uneducated, and the young (until recently) more than the old. But some important questions remain unanswered, such as “are movers likely to stay in a region that offers skills and training that are more specific to that region?”; and “do the involuntarily mobile adapt poorly compared with those who have had time to plan their moves?” Greenwood finds a paucity of studies relating to life cycle changes in migration, as in the case of graduated college students and divorced people. The essay ends with an expressed satisfaction that American data on migration have much improved. Robert Lucas’s (Boston University) “Internal migration in developing countries” covers a wide range of studies. Indeed, he opens by emphasiz- ing that the balance between investigations and creative theorizing has been adversely tilted in favor of the former. He focuses on the repeated efforts to fault the Harris–Todaro model4 by linking migration mostly to earnings differentials. He recommends more thinking about such factors as (1) the roles of job search costs, other labor market opportunities, personal information networking, the impact of capital, return migration patterns, and the like; (2) the influence of migration on production and wealth inequalities; and (3) the spectrum of policy options. “Economic impact of international migration and the economic performance of migrants,” by Robert Lalonde (Michigan State University) and Robert Topel (University of Chicago), concentrates on economic compari-

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 853 sons between the native-born and immigrants. To what degree do immigrants lower the earnings of the native-born? Not much, but usually to the detriment of the immigrants (p. 830). Why? For a variety of reasons including education, language difficulties, and problems of acculturation. But there are exceptions, the United States and Israel being two. Why? They are more open societies. The last essay in this section, “International migration and international trade,” by Assaf Razin and Efraim Sadka (both of Tel-Aviv Univer- sity) is a copy-book exercise in theoretical international trade geometry. No effort is made to use empirical data.

Aging, demographic composition, and the economy

For some inexplicable reason this topic has only recently become popular.5 “The economics of individual aging,” by Michael Hurd (State University of New York at Stony Brook), contains numerous data relating to private and public retirement plans, making it something of a model summary. Fortu- nately vital statistics and social security data were easily available, and Hurd has done considerable descriptive and projective work. This is followed by some desultory simple models used to explain retirement decisions—including an interesting variant that uses uncertainty as a factor. Uncertainty is resolved on the basis of stochastic dynamic programming.6 It would have been particularly interesting had there been discussion of studies showing the effects of the entry of technologically improved pharmaceuticals on retirement financing, but the absence of these is no fault of the author. Hurd concludes by stating that because we cannot reliably infer expectations from individual behavior, we need independent assessments of expectations: subjective probabilities will have to become “important variables in the estimation of stochastic dynamic models of decision-making” (p. 962). To which we can all say Amen. David Weil (Brown University) employs a macroeconomic approach to aging. Again, there are many data, and the analysis for the most part is empirical, although passing efforts are made to reduce the topic to the calculus. Weil contends that the social savings associated with reduced fertility and lower school enrollments approximate the increased expenditures by the swelling ranks of the retired (p. 1009); this conclusion seems dubi- ous to me. Significantly, Weil notes that the problems of financing the elderly are largely handled by increasing national productivity rates, a point having its own intriguing set of premises. For the period 1948–91 he finds an annual productivity increase of about 2 percent. The American growth rate of productivity increase per capita fell from 2.2 percent in the period 1948–72 to 1.7 percent during 1973–91, and this lower rate approximates

Click to return to Table of Contents 854 PDR 24(4) B OOK REVIEWS the projected effect of aging on consumption in the period 2010–30. Some- how, it sounds too neat. Weil’s essay is more of an analysis than it is a survey. As such, its tone is upbeat, providing a coherent, confident argument to what many have seen as a dismal topic. David Lam (University of Michigan) has provided intriguing insights in his essay, “Demographic variables and income inequality.” Lam displays a thorough knowledge of the data in discussing validations of the theoretical literature, of which he seems also to be a master. After noting that Becker’s 1991 analysis of assortative mating in marriage markets and al- lied subjects has led the latter to make one of his few explicit predictions, namely that there would be negative assortative mating with respect to wages, Lam finds no corroboration in the evidence at hand. Lam admits, however, that the data “are inevitably contaminated by the endogeneity of human capital investments and labor supply” (p. 1030). Lam holds that the model of assortative mating with household public goods as he presented it (Lam 1988) demonstrates that the magnitude and sign of assortative mating may differ over populations and time. For the most part, Lam concludes, the recent increasing inequality of income among males is attributable to structural changes, involving skills, in the labor market. One more point: Indicative of Lam’s performance as a “compleat de- tached observer” is his remark that what has been called family data should not be classified as household data.

Aggregate population change and economic growth

In “Population dynamics: Equilibrium, disequilibrium, and consequences of fluctuations” Ronald Lee (University of California, Berkeley) provides the closest approximation to a history of demographic economics to be found in the Handbook. Aside from its being the only real such effort, the essay is outstanding. Lee opens with Malthus’s Law of Population, not as Malthus later developed it, but as the British Classical Economists adopted it.7 He explains that there is some empirical validation that during early periods (1200–1800, pre–Industrial Revolution), a logged real wage series bears some visual relationship to a detrended logged population series (Figure 2, p. 1069). But after discussing the conceptual problems, he concludes cautiously that population aggregates responded to economic conditions most clearly only before the Industrial Revolution (p. 1073). After that revolution Malthus’s perceptions of the apocalyptic forces involving famine, disease, and so forth— Malthus’s feedback system—were less frightening (pp. 1073–1076). Space does not permit full discussion of Lee’s many finely phrased findings. One example, however, illustrates his analytical skill in evaluating empirical data. After showing that most studies over time and space

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 855 indicate a negative elasticity of fertility with respect to population density, he turns to a carefully pondered conclusion about the Classical system— namely that it should be viewed as an explanation of why population growth patterns varied, not why population levels fluctuated. Lee next moves to modern theories of population equilibrium, noting that natural resources have recently regained a place in the discussion in the sense that parents are aware of the necessity to limit the number of their children through preventive checks in order to assure higher living standards (p. 1078). Lee’s conclusion is that if classical Malthusianism made sense only for pre–Industrial Revolution society, the more modern problem has been a notion, based on the concept of environmental and resource limits, that there can be such a thing as an optimal equilibrium population level (pp. 1107–1108). The penultimate essay in the Handbook, by Marc Nerlove (University of Maryland) and Lakshmi Raut (University of California [San Diego]), “Growth models with endogenous populations: A general framework,” is another exemplary piece. Using the empiricism-to-theory route, it covers much of the same material as Lee. Nerlove and Raut question why in some instances as income increases fertility and mortality fall and investment in human and physical capital grows. Their answers involve a complex of factors, including nutrition, health care, education, and training. It is a com- prehensively neat treatment reviewing the results of intrafamily decisions on national aggregates. What Frank Notestein intuited and discussed in terms of vital statistics, Nerlove and Raut pick up, examine econometrically, and transform into an empirically derived model. The final essay, “Long-term consequences of population growth: Tech- nological change, natural resources, and the environment,” by James Robinson (University of Southern California) and T. N. Srinivasan (Yale University), is a brave and eminently successful effort at integrating the whole demographic–economics picture from the standpoint first of data and then of the development of formal models. As to methodology, Robinson and Srinivasan are clearly in the empirical Kuznetsian tradition of urging caution in the collection of data before even attempting analysis (e.g., pp. 1279–1280). As the volume editors note, Robinson and Srinivasan successfully track the evolution of thought on numerous aspects of their intricate subject: —To the question about a relationship between the finiteness of natural resources and the bounds to be placed on per capita income, their answer runs contrary to current opinion. They show that the interaction between environment and population growth currently experienced by developing countries differs in kind rather than in degree from these interactions in industrially developed countries. —They illustrate how a dysfunctional government and ill-designed incentive institutions in developing economies cannot reform inefficient

Click to return to Table of Contents 856 PDR 24(4) B OOK REVIEWS traditions pertaining to owners’ property rights. Nonexistent or inefficient financial markets preclude long-term economic advancement and also underlie dire demographic and other economic consequences. —They raise the question of whether the family as the basic statistical unit retains its traditional significance. Does women’s equality of career opportunity (including good if not equal compensation), with its effects on childbearing and stability of marriages, change the status of household as the appropriate demographic measurement unit? If not the household, then what? —In examining the Beckerian treatment of altruism, they question whether individual generosity can effectively replace social welfare programs designed for charitable or political stability purposes. Essential to Robinson and Srinivasan’s treatment is their belief that all of the foregoing are institutional questions that should not be assumed away. But the special merit of their chapter is that they have successfully linked all themes of the Handbook in a single comprehensive treatment.

Assessing the whole effort

From this reviewer’s standpoint, with his own clear preference for empiricism rather than model-building, the Handbook treats both styles equitably. While conceding that creating any handbook covering a large field is a daunting assignment, I conclude that Rosenzweig and Stark’s effort is truly successful—particularly from the standpoint of meeting current academic needs, which was their primary task. Their product excels in its ar- rangement, coverage, and rhetoric. But there is a serious flaw. I am troubled by the absence of virtually any explicit interest in the history of the discipline, for, as I have already remarked, thinking about the relationship between economic growth and demographic behavior is a very old human endeavor. Many modern academics opine with Henry Ford that “history is bunk.” And perhaps that is the reason why work associated with even such recent names as David Glass, Harvey Leibenstein, Frank Notestein, and Joseph Spengler is neglected. I hold the contrary view because I think that by the examination of the written record one observes how inventive minds have worked, and how they can inspire analysis today. There are many lessons to be learned. I will cite a few examples. In 1588 Giovanni Botero explained fluctuations in city size by relating growth to cheap food. When food prices rose because of poor crops and/or high transportation costs, city size declined: presumably in addition to experiencing increased deaths, many of the poor re-migrated to the country- side where transportation costs were small. Not only did the thesis in his Delle cause della grandezza delle città antedate Malthus’s 1798 Law, but Botero provided a special case that was both logical and could be (and was repeatedly) successfully tested empirically.

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 857

Malthus’s subsequent efforts to deal with the firm theoretical conclusions he had put forth in his 1798 Essay on Population carry to this day the lesson of a logical theory failing because its assumptions were both incomplete and likely partly erroneous. Starting with the 1803 2nd edition of that essay, Malthus tried unsuccessfully to substantiate the facile theory with observable facts. He could not do it, and when in 1820 he published Principles of Political Economy, his theme was that the pressing problem was under- not over-consumption of goods and services. Adam Smith’s earlier musings on the nature of empathy in individual psychology (as traced in his 1759 Theory of Moral Sentiments) show his in- sightful understanding of the costs as well as the benefits of technological specialization. These rich insights, careful readers know, are at the core of his arguments in the 1776 Wealth of Nations. The history of empirical measurement is every bit as rich. In analyzing the 1830 Belgian census, Adolphe Quételet (1796–1874) developed the concept of the “average man.” For him it possessed the characteristics of the ontological (factual) arithmetic means of the statistical observations of homogeneous populations. It is not so much the work of Quételet and of the Englishmen Francis Galton (1822–1911) and Karl Pearson (1857–1936), who contributed so fruitfully to comprehending measurements of scatter (deviations around measures of central tendency like the average), but the development of their (and others’) imaginative perceptions that demographic economists should find if not inspiring, at least useful. Why did each of these men “discover” his contribution, and why did he stop there? But this review essay ends on a positive note. Whatever my reservations about a profession that has no interest in, much less pride about, its past, I am fully of the opinion that the Handbook of Population and Family Economics should be on the shelves of every demographic economist—be it a student, a teacher, or a researcher. As Oliver Cromwell put it in his instructions to the artist Peter Lely on the painting of his portrait (in Walpole 1762–71),

Mr. Lely, I desire you would use all your skill to paint my picture truly like me, and not flatter me at all; but remark all these roughnesses, pimples, warts, and everything as you see me, otherwise I will never pay a farthing for it.

Notes

1 For example, if one wants to measure files, when there is increasing evidence that national economic wellbeing, should one use tax collection is deficient? Or should one use income data, derived from tax collectors’ wealth data, although they are generally

Click to return to Table of Contents 858 PDR 24(4) B OOK REVIEWS available only at time of the probate of an be constrained by ignorance about the con- estate? Or must one use consumption data sumption advantages of distant conurba- involving stratified random sampling, which tions. Accordingly some of these advantages are very expensive to obtain? are ignored when relocations take place. 2 In some sense this is a tautological pro- 5 In 1980 Cambridge University Press cess because any choice can be explained in published an excellent work, The Economics terms of preference. If I choose to sleep on a of Individual and Population Aging by Robert bench in the park it can be argued that that is Clark and Joseph J. Spengler. I had “com- my preferred bed; I prefer it to having to work missioned” the work, thinking that the topic and earn money to rent a room and purchase was interesting. The book caught on less a cot with sheets, pillow, and blankets. than any other in the series of Cambridge 3 Becker’s impact on the economics pro- surveys of economic literature. Now its time fession is as great as or greater than that of has surely come. any other living economist. Several of his 6 Most people consider uncertainty to be essays (Becker, 1965, 1967, 1974, 1976) and threatening, but as I have pointed out else- a book (Becker 1981, revised 1991), like where (Perlman 1980: 115–180) that is not Luther’s Articles, have defined the modern really the historical fact. Technological change, faith. In 1981 when I reviewed a National a very frequent source of uncertainty, is gen- Bureau of Economic Research volume on de- erally beneficent—even in the case of addi- mographic economics, I noted that Beckerian tional “good” life years as contrasted with un- modeling was likely replacing the older necessary terminal suffering attributable to NBER empirical tradition of Simon Kuznets reduced morbidity and mortality rates. (Perlman 1981). Little did I realize how 7 Nassau Senior, the first (and third) quickly that transformation would occur. Ac- Drummond Professor of Economics at Ox- cording to the index covering these two vol- ford, taught his students in the mid-nine- umes, Kuznets is mentioned rarely in the teenth century that the principles of Classi- text: only Schultz, Greenwood, Lam, Lee, cal Economics consisted of: 1) Logical and Robinson/Srinivasan actually refer to his maximization, or getting the most for the work. Even Fogel does not mention him in least; 2) Empirically, the expansion of agri- his bibliography. Sic transit gloria mundi. culture involved diminishing returns; 3) 4 The key criticism is that while urban Empirically, the expansion of industry in- employment is preferable to rural employ- volved increasing returns; 4) Malthus’s law ment, moves from rural to urban areas can of population limits.

References

Becker, Gary S. 1965. “A theory of the allocation of time,” Journal of Political Economy 75: 493–517. ———. 1967. “Human capital and the personal distribution of income: An analytical approach,” Woytinsky Lecture, University of Michigan. ———. 1974. “A theory of social interactions,” Journal of Political Economy 82: 1063–1094. ———. 1976. “Altruism, egoism, and fitness: Economics and sociobiology,” Journal of Eco- nomic Literature 14: 817–826. ———. 1981. A Treatise on the Family. Cambridge: Harvard University Press. ———. 1991. A Treatise on the Family, Enlarged edition. Cambridge: Harvard University Press. Botero, Giovanni. 1588. Delle cause della grandezza delle città. Clark, Robert L. and Joseph J. Spengler. 1980. The Economics of Individual and Population Aging. Cambridge Surveys of Economic Literature. New York: Cambridge University Press. Graunt, John. 1662. Natural and Political Observations . . . made upon the Bills of Mortality.

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 859

Lam, David. 1988. “Marriage markets and assortative mating with household public goods: Theoretical results and empirical implications,” Journal of Human Resources 23: 462– 487. Makower, H., J. Marschak, and H. P. Robinson. 1938. “Studies in mobility of labor: A tentative statistical measure,” Oxford Economic Papers 1: 83–123. ———. 1939a. “Studies in mobility of labor: Analysis for Great Britain, Part I,” Oxford Eco- nomic Papers 2: 70–97. ———. 1939b. “Studies in mobility of labor: Analysis for Great Britain, Part II,” Oxford Eco- nomic Papers 3: 39–62. Malthus, Thomas Robert. 1798. Essay on Population. ———. 1820. Principles of Political Economy. Perlman, Mark. 1980. Review of Imagination and the Nature of Choice by G. L. S. Shackle, Journal of Economic Literature 18: 115–118. ———. 1981. “Review of Population Change in Developed Countries by Richard A. Easterlin (ed.),” Journal of Economic Literature 19: 74–82. Smith, Adam. 1759. The Theory of Moral Sentiments. ———. 1776. The Wealth of Nations. Walpole, Horace. 1762–71. Anecdotes of Painting in England.

Click to return to Table of Contents 860 PDR 24(4) B OOK REVIEWS

HENRI LERIDON AND LAURENT TOULEMON Démographie: Approche statistique et dynamique des populations Paris: Economica, 1997. 440 p. 198 FF (C= 30.18).

JEAN BOURGEOIS-PICHAT La dynamique des populations: Populations stables, semi-stables et quasi-stables Paris: Presses Universitaires de France and INED, 1994. x + 296 p. 170 FF (C= 25.92).

Two recent books on demographic methods by French authors illustrate the rich- ness of the field of demography in France. If one wished to revive the traditional requirement that all English-speaking demography students should attain at least a reading knowledge of French, these two books would provide a powerful argument. Both contain interesting historical material, clear explanations of demographic methods, and a methodological perspective that deserves consideration by American demographers. The work by Henri Leridon and Laurent Toulemon is essentially a textbook on demographic methods that assumes knowledge of calculus. After an introduction containing a detailed and fascinating history of the field, the book is organized into five sections: population growth and inertia, formal analysis of vital events, statistical methods, causal analysis of demographic events, and historical patterns of population change. One strength of the Leridon–Toulemon book is its unified treatment of demographic methods and substantive issues. Although there is some specialization of sections and chapters, with some being focused more on methods and others more on substance, a balance between these two perspectives is maintained masterfully throughout. This is mostly a book on demographic methods, but the reader is never left to wonder “So what?” I can imagine that beginning demography students would find the work inspiring, and old pros would find it engaging and challenging as well. If translated into English (and I hope it will be), this book will be the rightful successor to Roland Pressat’s Demographic Analy- sis, first published in English in 1972. What I found most intriguing is that the authors’ perspective on quantitative methods is truly demographic. By that I mean it describes techniques for analyzing and presenting quantitative information that are familiar within demography but not elsewhere in the social sciences. In short, these are the tools that define the discipline of demography methodologically. Within American demography, however, these methods have become less popular in recent years. Instead of computing rates and building models of population processes, many demographers have embraced general techniques of data analysis, such as ordinary regression, logistic regression, and hazards models. Bowing to the influence of statistics and econometrics, and to a burgeoning micro-survey industry, traditional demographic analysis and modeling have become less common and, regrettably, less valued within the field. But what defines demographic methods? Judging by the balance between the sections of the book, Leridon and Toulemon’s perspective is that they are defined primarily by the study of life course transitions. The formal analysis of vital events is the longest section, followed by the chapters on the causal analysis of demographic events. These sections contain detailed discussions of demographic rates, including both theoretical concepts and practical concerns, such as how to calcu-

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 861 late rates in various substantive contexts using data from different sources. Even the section on “statistical methods” consists mostly of standardization techniques and model tables (for marriage, fertility, and mortality) and contains only two short chapters on logistic regression and event history analysis (or hazards models). Like more traditional demographic methods, logistic regression and event history analysis are methods for analyzing life course transitions. The main difference is that these statistical techniques are designed primarily for use with individual-level data (from surveys or clinical trials), whereas traditional demographic methods have been applied mostly to aggregate data (from censuses and vital statistics). However, the methodological perspectives of demography on the one hand versus statistics and econometrics on the other differ by more than a simple macro/ micro distinction, a point that the authors illustrate well. In Chapter 16, for example, survey data are analyzed as a means of describing differences in mortality by “socio-professional” categories. In the recent American literature on this topic (also based largely on survey data), it is common to present analyses whose end- products are odds ratios or relative rates, thus describing only relative levels of mortality in one social category compared to those of an arbitrarily chosen reference group. Although not discussed by Leridon and Toulemon in great methodological detail, this example reminds us that traditional demographic methods and measures can be used for the analysis of individual as well as aggregate data. Thus, instead of learning merely that the odds ratio or relative risk of death for some category is “significantly” different from that of an arbitrary reference group, we are able to compare actual estimates of life expectancy at age 35 for workers in different occupations. Of course, it is always important to know whether observed differences may be affected by chance variation in the sample of data being analyzed, but the implicit message conveyed by this example and throughout the book is that a careful description of an empirical pattern should take precedence over an evaluation of its statistical significance or other methodological details (such as correcting for “endogeneity” or unobserved heterogeneity). In addition to this no-nonsense approach to data analysis, the book also illustrates the rich tradition of demographic modeling. For example, the section on causal analysis contains a lucid description and evaluation of biological models of fertility processes (just as one would expect from Leridon, one of the world’s leading authorities on this topic). Overall, this book is a treasure trove of the principal methods that define and distinguish the field of demography. It has, to my knowledge, no equivalent in English. While the book by Leridon and Toulemon may serve as a general introduction to demographic methods, the book by Jean Bourgeois-Pichat is more specialized, covering in an exhaustive manner the theory of stable populations and related topics. Written toward the end of his life and finished posthumously by his former students and associates (led by Henri Leridon), the book is a major work of mathematical and technical demography. As explained in a preface written by Georges Tapinos, after a long and distinguished career as an internationally renowned demographer, Bourgeois-Pichat devoted himself to teaching a small circle of advanced students. This book, essentially a compilation of Bourgeois-Pichat’s lecture notes during 11 years in this capacity, illustrates the great seriousness with which he undertook that task.

Click to return to Table of Contents 862 PDR 24(4) B OOK REVIEWS

The treatment of stable population theory in Bourgeois-Pichat’s book is pref- aced by a detailed discussion of model life tables. One attractive feature of this discussion is its careful attention to the historical development of these methods, surely the most detailed description available. The reader can enjoy glimpses of earlier periods in the field of demography as the author recounts the motivation for developing each set of tables (Ledermann, Coale–Demeny, United Nations, OECD), the methods employed in each case, various criticisms of these tables or the methods used in deriving them, the substantive interest of each set of tables, and so forth. Bourgeois-Pichat’s discussion of stable, semi-stable, and quasi-stable populations is clear yet challenging. Stable populations are characterized by fixed birth and death rates, a constant growth rate, and an unchanging age structure. In general, a non-stable population whose birth and death rates become fixed at some moment will converge to stability over a period of several generations. Semi-stable populations resemble stable ones in that their population age structure is constant, even though their mortality, fertility, and growth rates may change over time. The essential requirement is merely that the population growth rate remains invariant across the age range. In this situation, actual growth rates at all ages equal the intrinsic rate of natural increase (Lotka’s r) for that moment of time, the population age structure is constant, and convergence to stability is in- stantaneous if age-specific birth and death rates stop changing. Whereas stable and semi-stable populations are mathematical constructs, quasi- stable populations are an “experimental concept.” Thus, Bourgeois-Pichat explains, when a hypothetical population is constructed based on a constant schedule of age-specific fertility rates and a mortality pattern that varies within a family of model life tables (e.g., Coale–Demeny West tables), the population age structure changes only slightly. According to Bourgeois-Pichat, semi- and quasi-stable populations come much closer to describing demographic realities than do stable populations. Thus, for purposes of indirect estimation or other modeling exercises, these more flexible constructs are preferable. In the book, each kind of stable population is applied to estimation problems as a means of illustrating their utility. Some readers may question the value of this elaborate mathematical machinery, given that its only practical application seems to be “indirect estimation,” a set of techniques that are slowly losing value as the quality of demographic data improves around the world. How- ever, improvements in vital registration and census data are proceeding slowly in many developing countries, hence these techniques are likely to remain useful for many years to come. Furthermore, their potential application to historical population dynamics is still largely unexplored. Overall, Bourgeois-Pichat’s last gift to the field of demography is an impressive piece of work. Although its practical applications are not always obvious, the book is highly recommended both for its historical descriptions of the field and for its cogent summary of the mathematical theory of certain fundamental aspects of population dynamics.

Department of Demography JOHN R. WILMOTH University of California, Berkeley

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 863

GAVIN W. JONES, ROBERT M. DOUGLAS, JOHN C. CALDWELL, AND RENNIE M. D’SOUZA (EDS.) The Continuing Demographic Transition Oxford: Clarendon Press, 1997. xi + 453 p. $95.00. John and Pat Caldwell have had a remarkable impact on the field of population studies over the last three decades. Their geographic scope alone is phenomenal— embracing, for example, much of Southeast Asia and sub-Saharan Africa, with the Indian subcontinent thrown in for good measure. Wherever one goes it seems that the Caldwells are either there, or they were there yesterday, or they will be tomorrow. Then there are the many publications on individual topics, ranging from famine in the Sahel to the Ford Foundation’s contribution to population studies. Still more, there is the highly influential concentration of writings around several major themes—such as the reasons for fertility decline and the context of high fertility, sexual networking and the AIDS epidemic, the demographic consequences of mass education and ideational change, and, of course, the potential of anthropological demography: the so-called micro-approach. Just how this dynamic duo have managed to write so much, while traveling so far, and all the time being so generous and so good-natured, remains one of the marvels of our field. Small wonder, then, that to mark John Caldwell’s formal retirement from the Australian National University, jumbo jets filled with demographers and other social scientists descended on Canberra in August of 1995. From the resulting symposium have already come two special editions of the Health Transition Review (of which Jack has been both founder and editor), one focused on the AIDS epidemic in the third world, the other dealing with the contemporary demographic transition.1 This, the third publication, appears under the title that denoted the symposium’s central theme: “The Continuing Demographic Transition.” The title is fitting because so much falls under its umbrella. Few people will read this book from cover to cover. Given its origins, that would probably be asking too much. Following a concise Introduction by Gavin Jones and Robert Doug- las, the book has 18 disparate chapters, by a host of authors, grouped in four parts and ranging over wide terrain. The quality of the chapters is generally reasonable and often strong. All contributors refer to the Caldwellian literature in an entirely easy, natural way. In what follows I will try to give a flavor of the contents. But, I conclude with a comment about where this book’s treatment of the demographic transition—and perhaps our discipline’s treatment of it more generally—is, I believe, often making an error of omission. Fittingly, Part I of the book contains solely a chapter by John and Pat Caldwell. It deals with the fertility transition, which they rightly see as now a global phenomenon, and arguably the most important turn of events of our time. Short though it is, the piece contains much detail and insight that cannot be addressed here (they paint on a big canvas). In essense, they argue that pretransitional societies did not practice fertility control and that the first major fertility decline, in France, has to be seen in the context of the ideational ferment associated with the French Revolution. The Caldwells tell us that subsequent European fertility declines were lengthy social learning experiences, for example about the practice of withdrawal. On the key question of causation, they assert (p. 19) that “the fertility transition has been economically determined, but . . . its timing and pace have

Click to return to Table of Contents 864 PDR 24(4) B OOK REVIEWS major social and ideological components.” The Caldwells have no doubt that family planning programs have hastened fertility declines. And they single out increased child education and reductions in child mortality as other powerful contributing forces. For connoisseurs (and there is affection here), the chapter contains several uses of the phrase “almost certainly,” a brief synopsis of “wealth flows theory,” and an ending with typical openness and sweep (p. 24): “Every assertion in this paper needs testing in many places.” What can one possibly say! Part II contains five chapters under the heading “Recent insights into demographic transition.” The first, by F. B. Smith, is the only chapter that is squarely focused on mortality per se. In an interesting mixture of social and demographic history, the author uses the rich Australian records for the period 1880–1910 to explore the basis of mortality decline. The most important reduction occurred for pulmonary tuberculo- sis, but there were also significant falls in infant mortality. There is some nice detail here—for example, that in the 1890s the average Queenslander gnashed through some 370 pounds of (often tainted) meat each year (compared to a mere 109 pounds in Britain). I am unconvinced, however, by Smith’s dismissal (on pp. 31 and 41) as “in- apt” and “irrelevant” for our understanding of Australia’s mortality decline issues related to the role of medicine and preventive and curative services that have been explored for Britain by writers like McKeown and Szreter. In the second chapter, Gavin Jones takes us to contemporary East and Southeast Asia, where signs indicate that an increasing proportion of young women (especially among the educated and urban) are not marrying. Marriage is not a particularly attractive option for many of these women, for whom life offers many other opportunities. As well as informing us about an interesting aspect of recent world marriage trends, Jones’s chapter is useful in highlighting what we do not know—for example, the extent to which this rise in nonmarriage may be partially offset by increased infor- mal cohabitation, and whether there are signs of similar trends in China. Jones also discusses some of the implications of this growth in nonmarriage, for example for the care of the elderly. This topic is taken up by Graeme Hugo in greater detail in a chapter on Indonesia. Hugo notes that while the family remains the chief source of support for older people in Indonesia, the level of support they now experience has declined in relative terms, and, quite possibly, in absolute terms too. Appropriately, Hugo develops his analysis within a “wealth flows” framework. For those who may have thought that wealth flows theory was on the wane, this book demonstrates an unparalleled resurgence. The two remaining chapters in Part II take us respectively to tropical Africa and to the increasing number of countries that today have low—often extremely low—levels of fertility. In a short methodological piece, William Brass, Fatima Juarez, and Anne Scott apply a parity progression ratio approach to Demographic and Health Surveys (DHS) data and conclude (p. 88) that “a transition to regimes of lower fertility through family limitation became widespread over Africa south of the Sahara in the 1980s.” This conclusion is not entirely new, but it is important. Moreover, given the usual lengthy lag between the start of fertility decline and its detection, one is tempted to say that falling national fertility must now be the rule, rather than the exception, in sub-Saharan Africa. After all, the UN’s 1996 revision suggests that even between 1985–90 and 1990–95 total fertility was falling in 24 of the region’s 42 countries (albeit very slightly in some ).2 At the other

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 865 end of the fertility spectrum, Paul Demeny’s chapter shows how widespread below-replacement fertility has become in Europe, in the countries of onetime Eu- rope Overseas, and also, increasingly, in Asia. Drawing on the later work of Malthus and the American physician John Billings, Demeny illustrates the longstanding theoretical support for sustained total fertility well below replacement. He cites (p. 106) Billings writing in 1893 to the effect that “young women are gradually being imbued with the idea that marriage and motherhood are not to be their chief ob- jects in life . . . that they should aim at being independent of . . . husbands, and should fit themselves to earn their own living in . . . one of the many ways in which females are beginning to find . . . employment.” Presumably this would apply aptly to the young women of countries like Thailand and Malaysia who are the concern of Gavin Jones. In both this rationale and the African fertility transition, we are surely engaging with aspects of what the Caldwells (p. 20) refer to as the growing “global society.” Incidentally, I would comment to the book’s editors (who raise the issue) that rather than being part of the continuing demographic transition, below-replacement fertility is a feature of the post-transitional world. Indeed, this is so even by Demeny’s definition—which sees the transition as over when both fertility and mortality are low.3 The post-transitional demographic world emerges from the transition, but it is a different beast (and, arguably, more closely related to creatures of the pretransitional world). Part III contains nine chapters grouped under the heading “The place of culture in the explanation of demographic transition.” Several of these are of limited significance, mainly because they examine specific topics and/or locations. Two chapters deal with Bangladesh. In what is the book’s only example of what might be termed the “survey genre” of contemporary demographic research, Sajeda Amin, Ian Diamond, and Fiona Steele employ multi-level logistic regression to data from the 1989 Bangladesh Fertility Survey to demonstrate that variation in “religiosity” accounts for much of the district-level variation in contraceptive acceptance. And in another chapter Amin and Mead Cain show that a “marriage squeeze”—mainly consequent upon mortality decline—has been the chief cause of the major shift from brideprice to dowry payments in Bangladesh. Anthropologists who have de- nied this causal link have simply got things wrong. That said, the country’s strong patriarchal regime has meant that it has not experienced all possible consequences of a squeeze, such as a reduction in the age gap between men and women at marriage (which the Caldwells did find in their fieldwork in South India during the 1980s). This chapter demonstrates well that reference to a blend of demographic, cultural, institutional, and other factors is required in order to provide a satisfactory account of the shift in marriage payments. Nevertheless, we should remem- ber that the main cause of the “shift” is the “squeeze”—and behind the squeeze lies the demographic phenomenon of declining mortality. One other chapter in Part III is devoted to South Asia. Indrani Pieris explores the complex basis of Sri Lanka’s relatively favorable performance in the “health transition stakes.” She argues convincingly that traditional Sinhalese culture combined a deep- rooted sensitivity to illness with a very practical approach toward treating it. Several other contributions to Part III adopt a broadly anthropological approach in trying to “get under the skin” of demographic change in particular societies. In

Click to return to Table of Contents 866 PDR 24(4) B OOK REVIEWS what is perhaps the best, and certainly the most finely textured chapter, Tom Fricke explores the moral dimensions of marriage in his Timling community of upland Nepal. As in most traditional societies, until recently reciprocity was the hallmark of almost everything Timling. However, as the institution of marriage has begun to change—ultimately because of the influence of such external factors as entry into the wage labor economy—so the moral dimension of marriage, and Timling life in general, are changing too. There is some analogy here with the chapter by Susan Cotts Watkins, Naomi Rutenberg, and David Wilkinson, which deals with the consequences of what they contend (p. 239) is the virtual flooding of Kenya with modern contraceptives. This development—again, coming from outside—is having profound implications for rural women’s control of their own fertility, their autonomy, and hence the general gender balance of power. Again, one feels the wind of the global society. In another chapter Caroline Bledsoe and Fatoumatta Banja explore the context of contraception in a Gambian community where women have their own theory of “body expenditure.” Prudent behavior in conserving body expenditure, or “reproductive capital,” perhaps with the help of modern contraception, may have strong repercussions in the moral realm. Thus a woman’s conjugal and familial support are more likely to be maintained if she can preserve her body’s innate vital substances, partly through the judicious spacing of her births. Part III also contains three theoretical chapters. Alaka Malwade Basu provides a framework for anthropological demographers wishing to explore the social basis of early-age mortality. Her chapter contains an illuminating discussion of the complexities behind the “deliberate neglect” of children. And she warns against the simplistic ethnocentrism that often flavors writing on this subject. In her chapter Karen Oppenheim Mason addresses the relationship between gender systems and demographic change. She reviews the often persuasive evidence that less-stratified gender systems tend to be more conducive to declines in fertility and mortality. However, her main point—on which she is surely right—is that gender systems usually have a conditioning rather than a causal influence on demographic change. The point is elementary, but very important. If a gender system is to be causally related to a demographic change, then it must be shown that a change in the system has produced a change in the demographic conditions. And, unfortunately, good evidence that this has happened is hard to find (though, certainly, the possibility cannot be excluded). Finally, the chapter by David Kertzer discusses the role of culture in explaining demographic change. Kertzer contends that, in general, demographers ignore culture, or they treat it as an error term, or they try to operationalize it as just another variable on the right-hand side of a regression equation. Illustrating his argument with a discussion of the abandonment of new- born babies in historical Europe, Kertzer calls for a historical approach to demographic explanation, allowing for contingency and incorporating culture. Culture is defined as “a continuously changing process, one intimately interwoven with the changing institutional structure and field of political power” (p. 152). I have no problem with the ingredients in Kertzer’s recipe for demographic explanation. Indeed, inter alia the factors he mentions must surely all be there. He may depict culture as a problem for demographers, but most demographers I know do not regard it as such. Indeed, I do not recognize the limited and barren landscape of demography that Kertzer describes. On the contrary, most of the chapters in this very

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 867 book address institutions, history, contingency, and culture in a sophisticated way. Only in one—that by Amin, Diamond, and Steele—might it be said that culture is reduced to a right-hand expression; but why should there not be a place for that too? This brings us to Part IV of the book, in which all three chapters deal squarely with fertility decline. Thus, and with humor, Geoffrey McNicoll reflects on the “governance” (nice word!) of fertility transition in Asia. Following from his earlier writings, his main concern here is with the oft-neglected role of institutional context. He identifies two dimensions of this context, argues convincingly that they have both been important in Asia, and exemplifies their independent operation apropos of the experience of such states as China and India. McNicoll’s first dimension is “regularity”—essentially the creation of an environment of order and stability that facilitates the processes of policy formulation, planning, and implementation. The second dimension is “duress”—the use of administrative and/or political pressure (including physical force) to attain fertility objectives. The next chapter, by Terence and Valerie Hull, exemplifies the operation of institutions of governance and socialization in contemporary Indonesia. With obvious knowledge and understanding, the Hulls too are basically concerned with the governance of that country’s fertility transition. The final chapter in the book, by John Bongaarts, returns to one of his own long-time concerns, namely the net effects of family planning programs on fertility, especially in the light of Lant Pritchett’s recent salvo emanating from the pages of this journal.4 Although Bongaarts gives a clear treatment of the issues—and concedes a point or two to Pritchett—this debate is in danger of becoming rather esoteric. Like the Caldwells, I find it hard to deny Bongaarts’s general contention that family planning programs have hastened fertility declines in many countries and that, accordingly, they have had a significant effect in slowing world population growth. Where does this leave us? Overall this is a pretty good book—although, taken individually, most of the chapters are perhaps not quite as “provocative” or “stimu- lating” as the editors promise at the start. Nevertheless, the majority are well worth reading. And university teachers will probably find several that should be included on their reading lists. Apart from the chapter by the Caldwells, however, only in this book’s Introduction do we find any discussion at a level of generalization that could be used to draw things together. This brief discussion occurs around pp. 2– 3, where Jones and Douglas mention that mortality has declined everywhere and that in many populations “an apparently linked fall in fertility has followed.” They then refer to the familiar “stylized version” of the demographic transition and say that there is enough reality in this version to explain the tremendous growth in world population that has occurred during the twentieth century. Still, that is as far as Jones and Douglas seem prepared to go, for they then remark on the wide range of circumstances prevailing when countries go through the transition, including claimed instances where mortality did not decline before fertility. This brings me, in the end, to my own brief lament. I feel a little shamefaced in making it, because the key point is so old. Indeed, I would like to believe that it is so obvious to the editors and authors in this book that they simply felt it requires no mention at all (though, on this, I have some doubt). The key point is that the cause of the fertility transition is the mortality transition.5 There is no real exception.6 There is no other cause. Everything else is con-

Click to return to Table of Contents 868 PDR 24(4) B OOK REVIEWS ditioning context—whether it be changes in technology, the existence of family planning programs, or the nature of prevailing economic, political, gender, administrative, educational, or other institutions. It is impossible to imagine a fertility transition in the absence of the mortality transition. The former just could not happen without the latter, irrespective of any economic or other changes in the conditioning context. However, given the occurrence of mortality transition, some rough approximation to demographic equilibrium has to be restored. So eventually, inevitably—and usually with little appreciation of the underlying cause—people reduce their fertility. If the 450 pages of this book do have a unifying question, then it is how to account for the fertility transition. And if this transition indeed has a claim to be the most important turn of events of our time, then its basis deserves to be clearly and correctly stated. But in this book—as so often elsewhere—the mortality transition and the role of its manifold consequences in influencing fertility are barely touched on. “Apparently linked” just will not do! In failing to highlight the ultimate demographic cause of what is a demographic effect, we are doing the subject a disservice. Certainly, historians and anthropologists (especially), and other social scientists too (even, sometimes, economists) can help us to appreciate the particular idioms and conditions of fertility transition as they apply in specific locations. Sometimes, however, our subject becomes so en- grossed in particular cultures and contexts that we take our eyes right off the ball.

London School of Economics TIM DYSON

Notes

1 See Robert M. Douglas, Gavin Jones, 4 See Lant H. Pritchett, “Desired fertil- and Rennie M. D’Souza (eds.), ”The Shap- ity and the impact of population policies,” ing of Fertility and Mortality Declines: The Population and Development Review 20, no. 1 Contemporary Demographic Transition,” (1994): 1–55; and Lant H. Pritchett, “The im- Supplement to Health Transition Review 6 pact of population policies: Reply,” Popula- (1996) (Canberra: Australian National Uni- tion and Development Review 20, no. 3 (1994): versity Press), especially the Preface. Also see 621–630. I. O. Orubuloye, John C. Caldwell, Pat Cald- 5 For an excellent recent statement see well, and Shail Jain (eds.), “The Third World Dudley Kirk, “Demographic transition theory,” AIDS Epidemic,” Supplement to Health Tran- Population Studies 50, no. 3 (1996): 361–387. sition Review 5 (1995) (Canberra: Australian For convenience here I leave aside the causes National University Press). of mortality decline itself, but on some of its 2 See United Nations, World Population ultimate historical stirrings see, for example, Prospects: The 1996 Revision, Annex 1: Demo- Richard A. Easterlin, Growth Triumphant: The graphic Indicators (New York: 1996), p. 120. Twenty-first Century in Historical Perspective (Ann 3 See Paul Demeny, “Early fertility de- Arbor: University of Michigan Press, 1996), cline in Austria-Hungary: A lesson in demo- especially chapter 2. graphic transition,” in D. V. Glass and Roger 6 Not even France, where the popula- Revelle (eds.), Population and Social Change tion roughly doubled in size while the (London: Edward Arnold, 1972), pp. 153–172. country’s mortality declined.

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 869

JOHN M. RIDDLE Eve’s Herbs: A History of Contraception and Abortion in the West Cambridge, MA and London: Harvard University Press, 1997. 341 p. $39.95. In Eve’s Herbs John Riddle documents the fact that for thousands of years various plants have been credited with contraceptive and abortifacient properties. His thesis is simply described. “[A]ncient women” (women of ancient times) “took certain substances, mostly herbs, in order to control their reproduction. . . . Women thought that what they took worked successfully, and I found that modern scientific reports tend to confirm their practices as probably being effective” (pp. 6–7). Over time this knowledge, which was fragile because the “chain of learning about antifertility drugs was forged by vocal cords” (p. 51), was dissipated. What may have happened is that along with women burned as witches perished their “special knowledge that to others was considered suspicious, magical, and, above all, not scientific as defined by the new age” (p. 167). In addition, “knowledge of antifertility agents was eroded by centuries of laws, religious doctrine, [and] social conventions” (p. 234). Even in the face of such difficulties, however, remnants of the ancient lore have lingered to the present in tiny pockets of the population who find that the old recipes remain effective. Let us examine Riddle’s points one by one. It is true that the contraceptive and abortifacient measures taken by people in the past included herbal compounds, but they included, in addition, nonherbal substances, magical spells and amulets, vaginal sponges and pessaries, other substances introduced intra-vaginally, physical procedures during or after intercourse, and various forms of incomplete intercourse (Himes 1970). Nevertheless, the herbal pharmacopoeia was large. Lists of rue, savin (juniper), pennyroyal, tansy, saffron, colchicum (autumn crocus), artemisia (wormwood and mugwort), black hellebore (Christmas or Lenten rose), apiol (parsley), sage, and mint send one scrambling to one’s botanical encyclopedia and then out to the garden, wondering just how many emmenagogues or abortifacients one is, however innocently, harboring. Quite a few, it would seem. Over the ages a large variety of herbs has been recommended to women who wanted to bring on a menstrual flow, to induce an abortion, or to avoid conceiving in the first place, and undoubtedly some or even many women used them. Shortly before World War I it was estimated that between six and eight out of ten “working women” in York took “drugs” to bring on a late menstrual period or to induce an abortion (Elderton 1914: 136–137). But as to the second point, is it true that “Women thought that what they took worked successfully”? Riddle’s source material tends to be of the prescriptive sort: such and such an herb is effective; prepare this, in this way; mix these herbs; drink that. It therefore reflects someone’s belief that a substance thus recommended will be effective, but the recommendation is unsupported by the independent tes- timonials of grateful users that it actually was. Even advocates of the use of emmenagogues did not guarantee their efficacy. Henry Garrigues (1900), for example, a physician who looks back into the nineteenth century rather than forward into the twentieth, lists both chemical compounds and herbs as emmenagogues. Yet he advises that “as their effect is very uncertain, it is wise to combine several in one prescription” (p. 257)—the example he gives con-

Click to return to Table of Contents 870 PDR 24(4) B OOK REVIEWS tains strychnine, aloes, quinine, ferrous sulphate, savin, and gentian—and “to combine the use of drugs with the other remedial agents recommended”—that is,

proper alimentation, tonics, moderate exercise in the open air, horseback riding, mild gymnastics, or massage . . . hot vaginal and rectal injections, warm hip-baths, warm foot-baths with or without mustard, long warm general baths . . . [and] electricity . . . bipolar intra-uterine faradization, with secondary current, or, best of all, galvanism, with the negative pole in the uterus.

Riddle, who does not comment on such energetic regimens of back-up treatment, takes a different view of polypharmacy:

Another version of the Antidotarium has around eight recipes for expelling the men- strua, each of which contains a number of abortifacient drugs. . . . The unknown factor here is not whether these compound recipes would work, because they probably would. We know that the ingredients put in the compounds acted in various ways to produce abortions. (p. 106) One compound . . . was made of forty-four substances, including artemisia, pennyroyal, a number of other mints, dittany, juniper, rue. . . . On the basis of what we know about the ingredients, it is difficult to imagine what the recipe would do except to cause an abortion. (p. 145)

Further on, however, Riddle comments, “The wares sold in the apothecary shops appear to have been complicated and (if I may judge without the benefit of scientific proof) less effective than the simple drugs of antiquity and the Middle Ages” (p. 166). He presents evidence not that women thought their herbal concoctions were effective but that many herbs were deemed contraceptive, emmenagogic, or abortifacient, and that these herbs are mentioned frequently in different sorts of writings. His evidence, therefore, is accumulative and indirect. Because a certain herb has been credited in one source with emmenagogic or abortifacient powers, mention of it in another source suggests to Riddle that it has been used with such powers in mind. He infers, for example, from Ophelia’s mention of rue that:

Ophelia may have been more involved with Hamlet than the modern reader knows. Ophelia’s use of the word rue may be different, not just a symbol of remorse, as it was for Laertes. She may have been referring to her use of the plant for an abortion, hence we better understand her madness. (p. 49)

It is only when we come to comparatively recent times—which is where Riddle’s account suddenly begins to move very fast indeed—that we know what women themselves thought about the efficacy of such preparations. The last few decades of the nineteenth century and the first few of the twentieth were the era of the pharmacist, and among the drugs he sold were ones to remove so-called irregularities or obstructions, coded references to the procurement of abortion that were fully recognized as such at the time. Riddle does not report what women thought about either the patented products or their constituent herbs, but it is plain that they maintained a degree of skepticism about both:

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 871

One woman said to me, ‘I’d rather swallow the druggist shop and the man in’t than have another kid.’ She used to boil ten herbs together, I forget the names on ‘em now, mixed up with gin and salts, and take a glass every morning before breakfast . . . nau- seous stuff it was. . . . And the kid came in the end. (Elderton 1914: 136)

Perhaps this woman could not afford the commercial product; perhaps she had more faith in the one she could prepare on her own. But wherever the truth lies, at the end of the requisite number of months she achieved a live baby. Likewise:

I confess without shame that when well-meaning friends said: “You cannot afford another baby; take this drug,” I took their strong concoctions to purge me of the little life that might be mine. They failed, as such things generally do, and the third baby came. (Women’s Co-operative Guild 1916: 45)

Similarly:

Many of our patients have at one time or another applied to their chemist, who has supplied them with “medicine” which usually “didn’t do no good.” (Florence 1930: 123)

And what of Riddle’s claim that “modern scientific reports tend to confirm their practices as probably being effective”? Even with the qualifications of “tend” and “probably,” such a statement cannot be allowed to pass without comment. Let us return to the case of Ophelia. Riddle explains that:

Rue does cause an abortion. Chinese investigators administered chloroform extracts of the whole rue plant in daily oral doses of 0.8 to 1.2 grams per kilogram of body weight to female rats over the first 8 to 10 days of pregnancy. The extracts reduced the number of pregnancies in the experimental group by from 20 percent to 75 percent (depending on the potency of the extract administered). (p. 49)

What would this mean for a woman rather than a rat? Pro-rating in terms of body weight and the average length of live-birth gestation (nine months in humans and 23 days in rats), a woman weighing 65 kilograms or 143 pounds would need about 65 grams of rue extract daily, or one pound weekly for three and a half months. Thus the required quantity of rue extract would amount to more than 10 percent of the woman’s own body weight, and the required quantity of the actual plant would undoubtedly be some multiple of this. The woman would need her own herb garden and a sizable garden at that. Given the seasonality of production and the volume of leafy matter required, the woman would need to turn her entire garden over to rue, yet even then it is not sure that she would abort. The fact that our calculations lead to absurdity suggests we should not equate women with the rats who were the subject of the experiment. But there is a more fundamental reason for eschewing the comparison, namely that, apart from their similarity in both being mammals, rats and women have very different reproductive physiologies. Rats have estrus, but women, unlike all other mammals except some of the other higher primates, have menstrual cycles. Are we to believe that a substance will have the same effect on either subject? Before we return to this question, let us move on to the next point in Riddle’s argument, that to do with witches. Scholars have contended that the targets of

Click to return to Table of Contents 872 PDR 24(4) B OOK REVIEWS witch hunts were, variously, the sexually deviant, women, the poor, the outcast, the old, and the schizophrenic or plain mad. Riddle thinks women were the target and, with token academic diffidence, prefers the reason given by two contemporary German scholars, a sociologist and an economist: women were the primary target because their secret knowledge fit poorly with the knowledge of the new scientific age (pp. 167ff). This explanation accords well with Riddle’s search for an explanation for the break with the past in the secret knowledge that allowed women to control their reproduction. It accords less well, as monocausal explanations tend to do, with the historical evidence. How does one explain the fact, for example, that men also were accused as witches? In France, indeed, more men were so accused than women. Riddle’s final point, that even today women use herbal preparations to control their fertility, is undoubtedly correct. Twice (pp. 50, 257) he mentions the case of a friend of a friend who conceived because she went on holiday without her Mason jar of Queen Anne’s lace and thus could not take her usual tablespoonful of seeds with a glass of water immediately after intercourse. For the demographer, probably the most difficult section in Eve’s Herbs is the one about demography. Riddle concludes that European population growth was slow until the middle of the eighteenth century. He jumps bewilderingly back- ward and forward between different periods, from country to country, from topic to topic, and from birth rates to life expectancy to marriage rates. But:

When the data are considered as a whole, . . . the patterns I see support the idea of continuous use of contraceptive and abortifacient medicines among some, but limited, populations over a long period of time. (p. 173)

I think Riddle would then like to say that rates of population growth increased after the middle of the eighteenth century because—the witch connection—women no longer had control over their reproduction. But he does not actually say this. Hav- ing battled one’s way through these puzzling pages it is anticlimactic to read that:

The data do not conclusively support the thesis that birth control was a modern (that is, late eighteenth or early nineteenth century) discovery. On the other hand, those data do not prove that a characteristic of modernity was the loss of birth control means once known to wise women and witches. (p. 178)

What a shame. One hopes Riddle will be braver in his next book. It would be incorrect to say that Riddle ignores mortality completely. Indeed, he remarks (in parentheses) that “some demographers believe the nineteenth century’s [population] increase is better understood by postulating a decrease in the mortality rate in part because of improvements in hygiene and public health” (p. 174), giving as a reference McKeown and Brown’s chapter in Glass and Eversley’s 1965 volume Population in History, a source of demographic information to which he returns repeatedly. Not recalling any other mention of mortality but anxious not to be unfair, I checked Riddle’s index: nothing on mortality, next to nothing on life span, nothing even on death (although there are several references to “death carrot”). Is it possible that Riddle sees fertility as the sole engine of population growth? Certainly he conflates the two. In a passage with which I had

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 873 particular difficulty, he states that “The European population rarely increased at an estimated rate of over 38 per 1,000 people annually in the pre-modern period prior to the late eighteenth century” (p. 170). Indeed, I thought it never did. In fact, Riddle’s source (Hajnal, in Glass and Eversley, again) was talking not about growth rates, but about birth rates. Enough of demography. What happened to the rats in the rue experiment? I think they were poisoned. Such a verdict ties in well with the views of human users of emmenagogues and abortifacients, as well as with the views of the medical and scientific communities of the last century or so. John Harvey Kellogg— physician and author of medical texts, vegetarian, Seventh-day Adventist, co-pio- neer (with his brother) of dry breakfast cereals—paused in his Ladies’ Guide (1895) to present a warning:

We wish to say a word . . . about a class of drugs known to the physician as emmenagogues, because of their supposed power to restore the menstrual function. There is quite a long list of these remedies. . . . Those which are the most efficient as stimulants of the uterus are so poisonous and potent for evil that much more harm than good is likely to come from their use, and hence none of them are to be recommended. (p. 515)

Kellogg’s was not a lone voice, even then. Herbal substances do poison, and deaths still occasionally occur from the use of pennyroyal oil to procure abortion, as Riddle himself mentions. Even a century ago some medical professionals made the important distinction between herbal infusions and highly concentrated extracts: the former they deemed ineffective; the latter toxic. Lella Florence (1930), an unsung heroine of the birth-control movement, mentions a woman with nine pregnancies (and seven living children) who had “often taken drugs to try to bring about a miscarriage. Once she was blind and deaf for three days as the result of taking too much quinine” (p. 87). Such accounts are consistent with later evidence on the workings of quinine. Pascal Whelpton, seeking the advice of physicians about how to classify abortions that respondents reported in the Indianapolis study, was told that if women reported inducing an abortion with a substance such as quinine without mentioning “considerable bleeding, great discomfort, or some type of sick- ness, in short anything that laid the women up, or caused considerable pain or inconvenience” (Taylor 1944: 35), then they had probably not been pregnant in the first place. If the herbal substances used to bring on the next menstrual period or to induce an abortion were either dangerous or ineffective, then how did they achieve their reputation? One explanation is that the woman who is anxious because her period is delayed, resorts to an herbal preparation, and then bleeds is likely to ascribe the bleeding to her intervention. Nevertheless, she may not have been pregnant at all: menstruation is irregular, and pregnancy is not the only reason a period does not appear when expected. Or her pregnancy may have aborted spontaneously, as do about 20 percent of unrecognized pregnancies, and around 15 percent of recognized pregnancies. But it may be wrong to exaggerate the reputation of these herbal contraceptives, emmenagogues, and abortifacients. It is all too likely that women took such substances not because they believed they would be effective, but because they hoped they would be. Moreover, they were willing to try them before they re-

Click to return to Table of Contents 874 PDR 24(4) B OOK REVIEWS sorted to more invasive measures. The sequence of progressive recourse to the hot bath, the controlled fall, strenuous exercise, pennyroyal or tansy tea, gin, pills from the druggist, and, as a last resort, the knitting needle or the curette, is well docu- mented. At some point in this sequence—although perhaps not until its end— women bleed, whether because they were not pregnant in the first place, or because they aborted spontaneously, or because this brutal regimen finally achieved the required result. It is easy to sympathize with the woman who yearns instead for some magic pill that would allow her to avoid both the protracted anxiety and discomfort of this time-honored regimen and the mechanical solution that awaits her at its end if the nonsurgical methods prove ineffective. But Riddle’s history of contraception nowhere mentions the fact that such a magic pill now exists, and that we have had it for close to 40 years. True, the oral contraceptive is not an abortifacient, because it prevents pregnancy in the first place; but the book claims to deal with contraception as well as abortion. Riddle does not distinguish clearly between the two—indeed, he ignores the distinction, however important it may be for the reproductive physiology by which women are biologically governed, and however profound the distinction may be for women themselves. Riddle is interested in herbs. He is not, at heart, interested in contraception. Naturally, then, he is uninterested in the intrauterine device, which, like the pill, is another recent invention. But he is also uninterested in—if not hostile to—old contraceptive practices if they make no use of herbs. He is incredulous, for example, that withdrawal could ever have been an important means of fertility control—let alone that in some societies it continues to be important—and derides those who are so foolish as to suggest as much:

Demographer Massimo Livi-Bacci found evidence that aristocracies and Jews engaged in fertility control prior to the modern era, but his findings are confined by other demographers to a restricted few. Emmanuel Le Roy Ladurie and Philippe Ariès assumed that the means employed by the elites was coitus interruptus. Were this so, the aristo- crats and elites were remarkable in their self-restraint, enough so to deserve their position. To withdraw before a climax is a torturous act for a male, and a male’s coopera- tion is necessary for the effectiveness of the technique. Not suffering the dangers of pregnancy and childbirth or the difficulties of raising small children, males as a rule are less concerned about family size. (p. 177)

This is a remarkable statement. Did all the ordinary men who exercised self-restraint thereby demonstrate their fitness for a higher station than the one to which they had been born? Are issues concerning family size decided mainly on the basis of the dangers of childbirth and the burden of caring for small children? And even if one imagines that one would personally find withdrawal difficult, is it not imprudent to extrapolate from such a personal belief—the while ignoring the weight of historical and contemporary evidence—to the behavior of others? But such is Riddle’s approach: the personal is everything. Repeatedly he offers us his opinion: on the actions of particular herbs, on the forces of history, even on what lies behind a muddled collection of grossly aggregate demographic statistics. He believes also, as many people would not, that “It is perfectly possible that people in past ages knew some aspects of science better than we do in the modern age” (p. 255). This belief, in combination with the seriousness with which he regards

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 875 personal revelation, perhaps makes it inevitable that he seems to want not so much to persuade as to convert.

Demography Unit GIGI SANTOW Stockholm University

References

Elderton, Ethel M. 1914. Report on the English Birthrate. Part I. England, North of the Humber. London: Dulau and Co. Florence, Lella Secor. 1930. Birth Control on Trial. London: George Allen and Unwin. Garrigues, Henry J. 1900. A Text-Book of the Diseases of Women. Philadelphia: W. B. Saunders & Company. Himes, Norman E. 1970 [1936]. Medical History of Contraception. New York: Schocken Books. Kellogg, J. H. 1895. Ladies’ Guide in Health and Disease: Girlhood, Maidenhood, Wifehood, Moth- erhood. Battle Creek, MI: Modern Medicine Publishing Company. Taylor, Howard C. 1944. The Abortion Problem: Proceedings of the Conference Held under the Aus- pices of the National Committee on Maternal Health, Inc., at the New York Academy of Medi- cine, June 19th and 20th, 1942. Baltimore: The Williams and Wilkins Company. Women’s Co-operative Guild. 1916. Maternity: Letters from Working-Women. London: G. Bell and Sons.

OCHIAI EMIKO The Japanese Family System in Transition: A Sociological Analysis of Family Change in Postwar Japan Tokyo: LTCB International Library Foundation, 1997. xii + 197 p. ¥3000. This book examines changes in the institution of the family in Japan following World War II. A major theme is that the Japanese family is converging toward a structure that Ochiai calls “the modern family,” which is typical of industrial societies everywhere. She says that in Japan, however, the concept of the family has been shaped by constant contrast with Western countries, and that the emphasis on Japanese uniqueness has obscured many commonalities with the West. Com- bining the concept of the modern family with the particular demographic conditions of postwar Japan, Ochiai proposes a new concept called the Japanese “postwar family system.” According to Ochiai, the postwar family system achieved a stable structure during a certain period after the war, was preceded by a distinct period during which this structure emerged, and was followed by a distinct period during which it underwent considerable change. Ochiai never precisely defines the time boundaries of these three periods, but in part of the book the boundaries seem to coin- cide roughly with the timing of major shifts in the birth rate (e.g., p. 74). Japanese fertility declined rapidly during the first half of the 1950s, remained fairly constant at about two children per woman between 1955 and 1975, and then re- sumed declining as it continues to do today. But in other parts of the book (e.g., p. 65), 1950 instead of 1955 seems to be the first dividing point, which is somewhat

Click to return to Table of Contents 876 PDR 24(4) B OOK REVIEWS problematic because of the major fertility decline that occurred between 1950 and 1955. The author never discusses this inconsistency. The postwar family system, as defined by the author, has three hallmarks: the “housewifization” of women, “reproductive egalitarianism” (by which is meant that most families had a similar number of children, either two or three, during the stable period) and “effects of demographic transition.” Housewifization is associated with rural–urban migration and the emergence of the urban Japanese “salaryman” who works long hours away from home while his wife stays home to take care of the children and the household. At many points in the book, Ochiai emphasizes that the postwar housewife role was not typical for women historically in Japan (or in West- ern countries either, for that matter), and that housewifization represents a major shift in family structure. The “effects of demographic transition” encompass several aspects. One is age at marriage, which was fairly constant between 1955 and 1975, but increased steadily after 1975. Another aspect concerns the traditional Japanese extended family, or ie, usually a stem family consisting of a nuclear family and the husband’s parents. The number of such households has remained remarkably constant, even as the number and proportion of nuclear family households or one-person households have increased substantially. Another demographic aspect is the extent to which housewives interact with non-kin. Since about 1975, interaction between housewives and non-kin (e.g., neighbors) has increased, Ochiai says, as the number of the typical housewife’s (or her husband’s) adult siblings has declined as a consequence of the fall in fertility during the first half of the 1950s. She says that non-kin interactions have replaced kin interactions, although the extent of this replacement is never made clear. Despite the logic of her argument, Ochiai presents no evidence that neighborhood community interactions have in fact increased. They may in fact have decreased as a consequence of the decline of small shops and the general weakening of neighborhood communities in Japan. The discussion of demographic trends is enlivened with descriptions of attendant changes in popular Japanese culture (such as the popular Sazae-san and Chibi Maruko-chan comic strips, which over the years have portrayed amusing yet revealing aspects of family life in Japan) as well as an interesting chapter on the evolution of the Japanese women’s movement, which got underway after major student unrest in 1968. Although this movement has never been as vigorous as its counterparts in many Western countries, it has nevertheless influenced popular thinking about the roles and status of women in Japan. The book also includes a number of asides (sometimes lengthy) about sociological theories of the family and comparative historical evidence about the nature of family life in Western countries and in Japan. The book ends with chapters on the decline of the extended family and the trend toward individualism in Japanese society, including a discussion of the special problems that this trend poses for women who choose the dependent role of housewife. As Ochiai acknowledges at the outset, this is not an academic book. She says, “My aim is not to deliver explanations in cut-and-dried textbook fashion but to challenge the generally accepted way of looking at the family and propose a different one. I invite readers to approach this book as if listening to a lecture series presented live, with each chapter being a separate lecture devoted to one topic.”

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 877

Although this approach makes for lively reading (especially since Ochiai writes from a well-articulated feminist perspective), a difficulty with the conversational mode of presentation is that careful marshaling of evidence in support of her arguments is often lacking. For example, one piece of evidence marshaled to show “housewifization” is a series of graphs of the female labor force participation rate plotted against age in five-year age groups, with one piecewise-linear graph for each of several successive birth cohorts of women (p. 13). The graphs are M-shaped with a dip at ages 25–29, reflecting departure from the labor force after marriage or first birth followed by re-entry some years later. The dip at 25–29 is deepest for the cohort of women born during 1946–50, who built their families mainly during the 1970s. The dip becomes progressively shallower for subsequent cohorts. Ochiai concludes from this graph that housewifization reached its peak among women born during 1946–50. A problem with this analysis is that there is no control for marital status. After 1975, women increasingly married at ages above 25, so that the proportion in the labor force at 25–29 was influenced not only by changes in labor force participation rates for single and married women in the age group, but also by increases in the proportion of single women in the age group. Because single women have a much higher rate than married women, the labor force rate for all women aged 25–29 can increase as the proportion single in the age group increases, even if the labor force participation rates for single women and married women separately do not change at all. Another problem is that the 1946–50 cohort was a baby boom cohort (Japan had a brief baby boom between 1947 and 1949), many of whom reached ages 25–29 during the 1973–75 recession when many women in this unusually numerous cohort had difficulty finding jobs or were laid off as marginal workers. Thus the unusually large dip in the age curve of labor force participation for this cohort of women may have stemmed largely from causes other than housewifization. No mention is made of these weaknesses in the evidence. A second piece of evidence of housewifization presented by Ochiai is the trend in the overall female labor force participation rate. She presents a graph which shows that this trend was downward between 1950 and 1980 (p. 18), concluding that a major shift toward housewifization occurred during this period. However, the graph she presents does not bear out this claim. The labor force participation rate varies within a narrow band, dropping from slightly above 50 percent in the late 1940s to about 47 percent in 1980. Based on the evidence presented, the trend toward housewifization after the war appears to have been rather slight. A disag- gregation of the trend in women’s labor force participation by type of work, which could have been done, would undoubtedly show a different picture, but Ochiai does not discuss this line of evidence. In the absence of such evidence, one is left wondering whether housewifization pertained more to changes in the popular concept of the ideal role for a married woman than to changes in actual behavior. Ochiai presents no convincing evidence for any period since the war in which Japanese family structure could be said to have been truly stable. Of course, certain aspects were stable for a while, such as the birth rate and the mean age at marriage. But even here there are problems. For example, constancy of the birth rate between 1955 and 1975 following the rapid fertility decline between 1950 and 1955 does not imply that family composition also remained constant between

Click to return to Table of Contents 878 PDR 24(4) B OOK REVIEWS

1955 and 1975. Indeed, a rapid fertility decline typically generates changes in family composition for many years afterward. Also, the late 1950s and the 1960s were times of very rapid urbanization and economic growth, during which the nature of female labor force participation changed markedly, even while the overall female labor force participation rate changed little. Most notably, the proportion employed as family workers declined steeply while the proportion working as paid employees outside the home increased sharply, with important consequences for family life. Levels of women’s education also increased rapidly between 1955 and 1975, with major implications for how women viewed themselves and their role in the family and in society. Also during this period the proportion of arranged marriages fell substantially, as did the proportion of newly married couples who lived with parents. And the proportion of couples who expected their children to support them in their old age declined steadily during the entire postwar period. None of these trends relating to family life appears to have stabilized at any time during the postwar period, and none of them receives adequate attention in this book. Although there is much of interest in this book, many of the author’s arguments, including the central argument that there was a period during which a stable postwar family system existed, must be viewed as speculative and in need of further empirical investigation and substantiation. The results of such investigation might require refinement or even revision of some of the arguments presented in this volume.

Program on Population ROBERT D. RETHERFORD East-West Center Honolulu

SHORT REVIEWS

by Martin Brockerhoff, Geoffrey McNicoll

JOSÉ ALVARADO AND JOHN CREEDY Population Ageing, Migration and Social Expenditure Cheltenham, U.K. and Northampton, MA: Edward Elgar, 1998. xii + 191 p. $70.00. This loosely constructed book is partly a general review of the literatures on the economic effects of aging and of immigration in industrialized countries and partly a simulation study of the effects of different levels of immigration on public expenditures in the Australian context. The authors defend high immigration as an economic benefit to industrialized countries on the grounds that it lessens the burden of social expenditures. They dispute the demographic argument that fertility

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 879 has a greater effect than immigration on population aging—though by the expedi- ent of assuming that total population size is of no moment. Constant annual immigration at a level corresponding initially to about one percent of the recipient population, an historically high rate even for Australia, is found to yield a social expenditure-to-GDP ratio of 29 percent by 2050 (up from 20 percent currently), whereas immigration at a quarter of that level yields a 32 percent ratio. The gain is only a deferral, of course, since the population is aging in both scenarios. More- over, some of the gain is lost if the immigrants are from non-English-speaking backgrounds, a category that has higher reliance on unemployment and other benefits. A concluding remark is well taken: “In the wider debate on social expenditure and population aging, it should of course be remembered that there are many other factors to be taken into consideration.”—G.McN.

MILICA ZARKOVIC BOOKMAN The Demographic Struggle for Power: The Political Economy of Demographic Engineering in the Modern World London and Portland, OR: Frank Cass, 1997. 273 p. $47.50; $24.50 (pbk.). An ethnic group, according to a definition quoted by the author, is “a people united by a common dislike of their neighbors and a common myth of their origin.” In this study of political demography the players are ethnic groups—more precisely, the leaders of such groups. The book presents a catalogue of situations around the world where political unrest is deemed to have demographic causes or consequences. The descriptions tend to be cursory, however, and it is not always apparent that a demographic element is salient—as in the case of nascent independence movements or other conflicts between center and region. The documentation is un- critical, often relying on newspaper accounts (and sometimes leaving blatant absurdi- ties in them uncorrected: “In 1990 the natality of Serbs dropped to 11.4%”). The only situation treated in some detail is the Balkans—certainly deserving the attention, but making clear by contrast the superficial treatment of other regions. The “demographic engineering” of the title refers to all kinds of measures to alter population numbers: by pronatalist policies, boundary changes, resettlement or forced population transfers (such as “ethnic cleansing”), and assimilation. The manipula- tion of census definitions and figures is also included as a kind of virtual engineering—appropriately so, given the role that statistics play in the political construction of ethnicity. Less persuasively, these measures are melded together to yield a “demographic engineering index,” which is calculated for each country or region and used in some tentative quantitative analysis. In sum, the book serves to call attention to the range of ethnic conflict around the world, and to the ways, usually unpleasant, that political ends may involve the pursuit of demographic goals. It cites the prominent theorists in the field but does not itself have much theoretical ambition. It is empirically lightweight. The author is professor of economics, St. Joseph’s University, Philadelphia. Appendix on inter-ethnic conflict in selected regions. Index.—G.McN.

Click to return to Table of Contents 880 PDR 24(4) B OOK REVIEWS

MURRAY FRIEDMAN AND NANCY ISSERMAN (EDS.) The Tribal Basis of American Life: Racial, Religious, and Ethnic Groups in Conflict Westport, CT and London: Praeger, 1998. viii + 168 p. $59.95. The premise of this collection of essays is that, as in Central Africa, the former Yugoslavia, and other parts of the world, “in the United States we are witnessing a startling increase in racial, ethnic, and religious tensions.” Murray Friedman, the lead editor, attributes growing fissures in American society to changes in the composition of immigrants since 1965 and the failure of the dominant “Protestant Es- tablishment” to assimilate newcomers; a consequent trend among minorities to seek group remedies for obtaining economic and political power; the decline of the ideology of individual rights and responsibilities; and unyielding beliefs on divisive issues such as abortion, homosexuality, and affirmative action. While pointing to growing conflict, Friedman emphasizes that social fragmentation resulting from such circumstances “can also be a source of strength, adding vigor and variety to our society.” Contributing authors address a broad range of topics. On the issue of plural- ism, Gary Rubin argues that the “new immigration” of Asians and Hispanics, unlike that of Europeans earlier this century, is based on a calculated social mobility strategy of separation and isolation rather than integration. Despite the economic and social benefits of this strategy for new immigrants, Rubin perceives such sepa- ratism negatively, inasmuch as ethnic groups will fail to develop the understanding and respect for one another that result from constant interaction. James Davison Hunter and Kimon Howland Sargeant find the roots of America’s contemporary “culture wars” in cleavages between communities whose worldview is either “or- thodox” or “progressive.” Their provocative discussion notes that these two categories derive from religious convictions, but nonetheless cut across religious de- nominations and political affiliations depending on the social issue involved. Elsewhere in the volume case studies of the riots in Crown Heights, Brooklyn and Los Angeles in the early 1990s are less satisfactory, being descriptive, journalistic accounts that offer scant interpretation. Judith Goode describes results of the ethnographic Philadelphia Changing Relations Project sponsored by the Ford Foun- dation between 1988 and 1990. Her sobering conclusion is that state and institutional policy celebrating ethnicity (for instance, multicultural schooling and neighborhood development) “frequently not only reinforces stereotypes but ironically strengthens the boundaries it hopes to reduce.” James Kurth proposes an ambitious theory of ethnic relations that distinguishes models of (1) a stable West- ern Europe, where the state existed before the nation and ethnic assimilation was induced, (2) a volatile Eastern Europe, where preexisting nations have undermined the state’s capacity to suppress conflict, and (3) a peaceful America, where the assimilation and social mobility of ethnic groups (other than African Americans and Mexican and Puerto Rican immigrants) have been facilitated until recently by the perception that the individual preceded, and was therefore largely unbound by, both the nation and the state. A final chapter by Nathan Glazer laments America’s collective failure to discuss divisive issues of racism and hate without recourse to the judicial system. The breadth of topics addressed in this book as- sures that readers interested in almost any aspect of American ethnicity and social change will find chapters of value. —M.B.

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 881

FRANK FUREDI Population and Development: A Critical Introduction New York: St. Martin’s Press, 1997. vi + 201 p. $59.95; $19.95 (pbk.). Frank Furedi, who lectures in development studies at the University of Kent, as- pires to write an intellectual and political history of the population and development question. His book starts in the era of Social Darwinism in the 1930s but most attention is paid to the last four decades. It is well written, sharply observed, and with criticisms often well-taken. Understandably, on this complicated terrain he is not always sure-footed. Nicholas Eberstadt would be astonished to find himself described as “another vocal supporter of population policies.” Again, several pages are spent recounting the fatal errors of the Coale–Hoover study of 1958, although—demonstrating a common hazard in this field—the most recent research on the subject has restored the Coale–Hoover arguments to favor. While the author claims to take “neither a positive nor a negative view of population growth,” he in fact accepts the (1980s) revisionist view in which the question is seen as settled and population planners as having lost. However, Furedi’s story is not primarily about empirical facts. It is about “the manner in which the relationship between population growth and development has been conceptualized, fought out and altered.” He is concerned, in the postmodern manner, with “a critical interrogation” of “the narrative of demography.” The interrogation, in this account, clearly points to the guilt of what the author terms the international population lobby, a loosely defined group of mainly US and other Western government agencies, international bodies (notably UNFPA), and NGOs. The population lobby represents the interests of the elite of the North, and is covertly motivated by apprehension about immigration and political instability engendered by population growth in the South. Its concern is with curtail- ing that population growth, in the pursuit of which it practices “tactical expedi- ency,” opportunistically switching from one rationale to another, from one instrument to another, in “a permanent search for effective arguments.” It had to distance itself from its “openly eugenic past,” but has subsequently found its justifications succes- sively in poverty and economic development, in environmental degradation, and in female empowerment. Thus, when the revisionist ascendancy put paid to the developmental rationale, “the shift from famine to pollution [as the main harmful effect of population growth] represents a pragmatic accommodation to contemporary reality. . . . The population-environment synthesis is an artificial way of blaming environmental degradation on the numbers of people.” With that link too being seen as shaky, “women’s health has become the latest in a long line of fashionable policies through which the population lobby seeks to influence fertility.” Unfortunately, “from past experience, it is difficult to see why the new women-oriented rationale should be any more durable than the development one.” Of course, rationales may be mere window-dressing where there is scope to apply blatant political pressure. To Furedi, the spread of population programs in the 1990s is linked to the diminished influence of poor countries in the aftermath of the Cold War. “Numerous African countries which opposed such policies in the past have now fallen into line.” Furedi’s own stance is not paraded but is implicit throughout the study. He holds that “the disposition to interpret social problems through the prism of population . . . detract[s] attention from the fundamental socio-economic problems facing

Click to return to Table of Contents 882 PDR 24(4) B OOK REVIEWS societies.” The preferences of ordinary people are to be respected, not second- guessed and certainly not manipulated. He is for a resolute noninterference in how societies are ordered, even to disallowing foreign objections to practices like female circumcision or child marriage that “have existed for hundreds of years” and are “integral to the moral and social code of the societies concerned.” “The real issue worth considering is from where do a group of population professionals get the authority to decide what is in the best interest of people in societies around the world.” The assumption by Western experts that they know what is best for poor people is “forcefully expressed in one of the key terms of the Cairo consensus—empowerment,” a term the author finds patronizing and paternalistic. Bibli- ography, index.—G.McN.

ANRUDH JAIN (ED.) Do Population Policies Matter? Fertility and Politics in Egypt, India, Kenya, and Mexico New York: Population Council, 1998. xix + 203 p. $14.95 (pbk.). The four country studies examine population policymaking and the politics surrounding it from historical and contemporary perspectives. Studies are based on sources that reflect perceptions of population issues within the country: official documents, public statements, mass media, and religious commentary. Issues related to the evolution of population policies include the role of stakeholders, the influence of internal politics and international agencies, and the degree of flexibility and autonomy at the local level. An introductory essay offers observations on the politics of fertility transition and a concluding essay speculates on the future of population policies.

NATIONAL RESEARCH COUNCIL, COMMITTEE ON THE HUMAN DIMENSIONS OF GLOBAL CHANGE Environmentally Significant Consumption: Research Directions Washington, DC: National Academy Press, 1997. viii + 143 p. $34.00 (pbk.). Responsibility for environmental harm due to human action can be formally allocated between the effects of population growth and the effects of per capita economic activity (“consumption”). In international discourse, this division permits blame for trends such as deforestation or global warming to attach to both rich and poor countries. Increasingly, international declarations call symmetrically for curtailment of population growth among the poor and of consumption among the rich. Yet consumption in this usage clearly differs from what economists mean by the term, and from what sociologists mean. As the editors of this brief but illuminating report remark, “Consumption . . . has neither well-defined and accepted units of measurement nor a scientific community devoted to studying its dynamics. . . . It lacks even a shared definition . . . ” (p. 3). They contrast its elusiveness with the conceptual simplicity of population. The NRC Committee on the Human Dimensions of Global Change is here concerned with the environmental impacts of whatever consumption is. For Paul Stern, the committee’s study director, it is “environmental consumption,” defined as “hu-

Click to return to Table of Contents B OOK REVIEWS PDR 24(4) 883 man and human-induced transformations of materials and energy.” Matters that fall within this subject include the amount of stuff in the economy (“in 1990, the average American consumed over 50 kg of material per day, excluding water”), the relationship between population size and CO2 emissions (“there are disproportionately large effects for the most populous nations”), the environmental impact of travel (which “is emerging as the primary leader of growth in carbon emissions in the wealthy, industrialized countries”), cross-cultural comparisons of wellbeing (“by objective indicators other than life-span, the quality of life in some ultra-low-consumption societies seems rather high— . . . by many indicators higher than ours today”), and the force of emulation of Western life-styles (“simple emulation remains an empirically weak model for prediction”). The volume consists of summaries of papers and presentations from a 1995 workshop held by the committee. The editors—Paul Stern, Thomas Dietz, Vernon Ruttan, Robert Socolow, and James Sweeney—are also the authors of a final chapter, “Strategies for setting research priorities.”—G.McN.

UNITED NATIONS, DEPARTMENT OF ECONOMIC AND SOCIAL AFFAIRS World Urbanization Prospects: The 1996 Revision New York: United Nations, 1998. viii + 190 p. $37.50. The biennial series of World Urbanization Prospects prepared by the United Nations Population Division is the most widely used source of urban population statistics. In addition to updating estimates and projections of city sizes, rates of urban growth, national urban populations, and other measures for all countries and large agglomerations, the 1996 Revision is notable in two respects. First, whereas previous issues of the series provided insufficient text to describe a vast body of data, the latest revision is exceptionally skimpy in its discussion of urban population trends and is devoid of interpretation. Second and more important, the 1996 Revision offers for the first time in 20 years a description of the methodology used by the UN to estimate and project urban populations and agglomerations. The one significant change from the methods developed in the 1970s (outlined in detail in the UN’s 1980 publication Patterns of Urban and Rural Population Growth) is the imposition of an unspecified “dampening factor” to cities in countries where the aggregate of city populations is projected to grow more rapidly than the national urban population (p. 36). One hopes that this procedure helps to avoid some of the egregious errors that have plagued the UN’s short-term city population projections in the past—for instance, the forecast in 1980 that Mexico City’s population would by now exceed 30 million, as compared to the current estimate of about 18 million. Yet, implausible projections of future city sizes remain. In the next 15 years, for instance, Dhaka and Lagos are expected to double in size—to about 20 and 25 million, respectively—and to rank among the world’s five largest cities, despite weak prospects for economic growth in coming years. Since these projections, like several others, are derived from highly suspect census data, it is appropriate to treat them with skepticism. World Urbanization Prospects is the only publication to provide the definition of “urban” area for each country of the world. Given that these definitions vary considerably from country to country, and given the common misuse of the terms

Click to return to Table of Contents 884 PDR 24(4) B OOK REVIEWS

“urban” and “city” in describing populations, this information is perhaps the most consistently valuable contribution of the series. To lend greater credibility to the data, future issues would benefit from a review of data presented in previous publications. Current projections of the developing world’s urban population in the year 2000, for instance, are 11 percent lower than those projected in 1980. The Population Division is uniquely positioned to assess the extent to which this change is largely attributable to its estimation and projection methods, rather than to the availability and use of more recent data from the 1990 census round or to unfore- seen demographic changes in recent years.—M.B.

UNITED NATIONS CHILDREN’S FUND Education for All? The MONEE Project Regional Monitoring Report No. 5 Florence: UNICEF International Child Development Centre, 1998. xi + 135 p. $20.00. Detailed comparative data and analysis of the education sector are presented for the 27 countries of Central and Eastern Europe and the former Soviet Union. An initial chapter is described as an update on welfare changes in the region. It is followed by three chapters specifically on education: a general comparative assessment of conditions and trends in the sector; an examination of problems of access to education; and a discussion of financing, governance, and policy issues and of the steps required to provide “education for all.” The text balances broad description of the mostly bleak educational situation with specifics on particular countries and topics. A lengthy statistical annex presents an assortment of data by country and year (1989–96) on population, the family, health, the economy, crime, and school enrollment. Data provenance is clearly a critical question in such a publication. The report’s major sources are the national statistical offices in the region (which, however, “should not be considered responsible for the data included here”). Where published sources have been drawn on, the tables are meticulously footnoted: particular country series are attributed to agencies ranging from the Interstate Statis- tical Committee of the Commonwealth of Independent States and the National Statistical Committee of the Kyrgyz Republic to the US Census Bureau. The World Bank is not a source, nor are its World Development Indicators compared to the report’s data where they overlap (indeed, the ICDC data “are not necessarily consistent with those found in other UNICEF publications”). The entire data base, called TransMONEE 3.0, can be accessed at http://eurochild.gla.ac.uk/Documents/ monee/monee_project.htm. The MONEE Project—in full, Public Policies and Social Conditions: Monitor- ing the Transitions in Central and Eastern Europe—was begun in 1992 as an activity of UNICEF’s Innocenti Centre in Florence. Its coverage has steadily expanded, most recently to include the Yugoslav successor states and Central Asia. Previ- ously published monitoring reports are: Public Policy and Social Conditions (1993), Crisis in Mortality, Health and Nutrition (1994), Poverty, Children and Policy (1995), and Children at Risk in Central and Eastern Europe (1997). The 1998 report’s editor was John Micklewright. Glossary, bibliography.

Click to return to Table of Contents DOCUMENTS

Population Aging and the US Federal Budget

The effect of demographic changes on the cost of the two major US federal programs affecting the old-age population—Social Security and Medicare—features prominently in the discussions aimed at reforming these programs. A report of the Congressional Budget Office, Long- Term Budgetary Pressures and Policy Options (May 1998), provides a detailed analysis of the prospective influence the upcoming retirement of the large baby-boom generation will have on federal budget outlays and considers potential measures for reducing growth in spending for Social Security and Medicare. Excerpts from the Summary of the report are reproduced below.

Some simple demographic facts lie behind a period of favorable demographics that be- concerns about the long-run budgetary situ- gan with the retirement of the generation ation facing the United States. This country’s born between World War I and World War population is aging. The Social Security Ad- II, whose relatively small numbers are now ministration estimates that between now and providing a respite to the spending growth 2030, the number of people age 65 or older of Social Security and other entitlement pro- will double, while the number of people ages grams for the elderly. 20 to 64 will increase by only about 15 per- Besides straining entitlement programs, cent (see Table 1). the retirement of the baby boomers will also Some of that demographic change re- significantly slow the growth of the labor flects the welcome fact that people are liv- force. The effect of having such a large group ing longer today. Thanks to improved health of workers leave the labor force will be ac- care and healthier lifestyles, a growing pro- centuated by the fact that the high birth rate portion of the adult population now reaches during the baby boom was followed by a age 65, and life expectancy at that age has much lower rate (a baby “bust”). As a re- increased by about 15 percent since 1970. sult, the growth of the labor force will slow When Medicare was created in 1965, the av- to a crawl between 2010 and 2020 and al- erage person in the United States was ex- most to a standstill between 2020 and 2030. pected at birth to live about 71 years. By That projection stands in stark contrast to the 1990, that expected life span had risen to 75 2 percent annual growth that the labor force years; by 2010, it is projected to increase to 78. recorded from 1960 to 1989, and even to the A second factor behind the demographic 1 percent average annual growth rate exchange is the baby boom, the large genera- pected over the next 10 years. tion of Americans born between 1946 and With more retirees and little growth in 1964. In 2008, the oldest members of the the number of workers, the share of elderly baby boom will turn 62 and become eligible people in the adult population will increase sig- to claim early retirement benefits under So- nificantly in coming decades. One measure cial Security. That date will mark the end of of that demographic pressure is the 65-plus

POPULATION AND DEVELOPMENT REVIEW 24(4) (DECEMBER 1998) 885

Click to return to Table of Contents 886 D OCUMENTS

TABLE 1 Population of the United States by age, calendar years 1950–2070 Age group 1950 1970 1990 2000 2010 2030 2050 2070

In millions of people Less than 20 years old 53 81 75 81 81 83 84 85 20 to 64 years old 92 113 153 168 186 192 203 206 65 or older 13 21 32 35 40 68 75 84 Total 158 215 260 285 307 344 362 376 As a percentage of the total population Less than 20 years old 34 38 29 29 27 24 23 23 20 to 64 years old 58 53 59 59 61 56 56 55 65 or older 8 10 12 12 13 20 21 22 Total 100 100 100 100 100 100 100 100 As a percentage of the population ages 20 to 64 Less than 20 years old 58 71 49 48 44 43 42 42 20 to 64 years old 100 100 100 100 100 100 100 100 65 or older 14 18 21 21 21 36 37 41

SOURCE: Congressional Budget Office based on data from the Social Security Administration. NOTE: Population as of July 1. dependency ratio—the ratio of the number of By contrast, Medicare and Medicaid are people age 65 or older to the number of people open-ended entitlement programs that place ages 20 to 64. In 1950, the 65-plus dependency no dollar limits on the benefits provided to ratio was a little less than 15 percent. That ra- each enrollee. Over most of the programs’ his- tio rose to about 20 percent in 1990, and the tories, benefits per enrollee have risen rapidly. Social Security Administration expects it to Indeed, the growth in per-enrollee costs swell to about 35 percent in 2030 and to more is the main reason that federal spending for than 40 percent by 2070. Although other fore- Medicare and Medicaid, now more than casters have different estimates, they generally three-quarters of that for Social Security, is agree on the basic story: the 65-plus depen- projected to overtake spending for Social Se- dency ratio will increase substantially over the curity within 10 years. The persistent growth coming decades. in spending per enrollee reflects an increase As a result, both the outlay and revenue in the volume and intensity of services pro- sides of the budget will be increasingly vided through Medicare and Medicaid, and strained after 2008. Revenues will grow without a change in policy, those factors will more slowly as the number of people work- continue to increase the burden of federal ing—and the economy—grows more slowly. health costs in the years ahead. Thus, even At the same time, outlays for government if the 65-plus dependency ratio did not climb programs that aid the elderly (Social Secu- with the retirement of the baby boom, fed- rity, Medicare, and Medicaid) will rise as the eral health spending would still be projected number of people eligible to receive benefits to rise faster than GDP and would put in- from those programs expands. creasing pressure on the budget. The projected increase in Social Security spending as a share of gross domestic product (GDP) results largely from the surging Slowing the growth of number of people eligible for benefits, but spending on Social Security the growth in Medicare and Medicaid also and Medicare reflects an increase in spending per enrollee. The Social Security benefit paid to each re- In 1997, federal spending for Social Security cipient is set legislatively by a formula that and Medicare exceeded $500 billion, or depends on the recipient’s history of wages. about 7 percent of GDP. By 2030, when most

Click to return to Table of Contents D OCUMENTS 887

FIGURE 1 Projected growth in spending be able to make the necessary adjustments and for Social Security and Medicare, calendar could therefore have lower incomes when they years 1995–2070 stopped working. Spending as a percentage of GDP Second, the age at which a worker would 16 become eligible for full retirement benefits— 14 the “normal retirement age”—could be 12 Total raised to reflect increases in life expectancy. 10 Under legislation enacted in 1983, the nor- 8 mal retirement age is already scheduled to Social Security rise from 65 to 67 by 2022. Some proposals 6 would speed up the transition to age 67 and 4 Medicarea then further increase the age to keep up with 2 future gains in life expectancy. Raising the 0 age at which a worker would become eligible 19952010 2030 2050 2070 for full benefits (without changing the ear- SOURCE: Congressional Budget Office based on liest eligibility age) is, for most purposes, intermediate assumptions from the 1997 reports of the equivalent to cutting initial benefits, with boards of trustees of the Social Security and Medicare trust funds. similar advantages and disadvantages. NOTES: Data are plotted at five-year intervals. Third, future annual cost-of-living ad- a Medicare spending is shown net of premium receipts. justments (COLAs) could be reduced. Cur- rent law indexes the basic Social Security benefit by the increase in the consumer price baby boomers will have retired, those two index (CPI), beginning when a worker be- programs will consume 12 percent of GDP comes eligible for benefits. Many analysts (see Figure 1). Nearly all of the increase in feel that the CPI overstates increases in the Social Security’s share of GDP between now cost of living, although the magnitude of the and 2030, and two-thirds of the increase in overstatement and what should be done Medicare’s share, will occur after 2010, as about it are subject to much debate. More- retired baby boomers become eligible for over, the Bureau of Labor Statistics recently those programs. Those projections are based made changes in how the CPI is calculated on the intermediate assumptions from the that address several of the analysts’ concerns. programs’ trustees in their 1997 annual re- Unlike across-the-board reductions in ben- ports; CBO [Congressional Budget Office] efits and increases in the normal retirement modified the projections for Medicare to re- age, substantial changes in COLAs would flect the changes enacted in the Balanced eventually reduce benefits the most for the Budget Act of 1997. oldest beneficiaries and those who initially became eligible for Social Security on the Social Security basis of disability. Each of these approaches could be used Three approaches illustrate the trade-offs that to achieve considerable savings, with the the Congress would face in trying to reduce amount depending on the specific changes the growth in spending for the Social Security made. Estimates provided by the Social Se- program. First, the initial benefits of future curity Administration’s Office of the Actu- Social Security beneficiaries could be reduced ary illustrate the magnitude of the changes below the levels current law would provide. that would result from several specific op- Announcing across-the-board cuts in initial tions (see Table 2). Cutting the benefits of benefits long before they took effect could pro- each successive cohort of workers who be- duce substantial savings while still preserving came eligible for Social Security disability or the basic benefit structure of the Social Secu- retired-worker benefits by 0.5 percent per rity system and giving people time to adjust year, starting in 1998 and ending in 2032, to the reduced benefits. In principle, workers would ultimately reduce spending by about could offset the cut in their future Social Se- 16 percent. But the full savings would take curity benefits by either working longer or sav- a long time to achieve. By 2030, spending ing more. However, some people would not would be about 10 percent below the pro-

Click to return to Table of Contents 888 D OCUMENTS

TABLE 2 Effects of illustrative options for reducing growth in spending for Social Security Option 2010 2030 2050 2070

Spending as a percentage of GDP Continue current law 4.9 6.6 6.5 6.7 Phase in a 16 percent reduction in initial benefitsa 4.8 5.9 5.5 5.6 Raise the normal retirement ageb 4.8 6.1 5.6 5.5 CPI minus onec 4.5 5.9 5.7 5.9 Savings as a percentage of projected spending under current law Phase in a 16 percent reduction in initial benefitsa 2101516 Raise the normal retirement ageb 1 8 14 18 CPI minus onec 8111212

SOURCE: Congressional Budget Office based on estimates provided by the Social Security Administration, Office of the Actuary, March 3, 1998, using the intermediate assumptions in the 1997 report of the Board of Trustees of the Federal Old-Age and Survivors Insurance and Disability Insurance Trust Funds. NOTE: CPI = consumer price index. aStarting in 1998 and ending in 2032, the benefits of each successive cohort of workers becoming eligible for Social Security disability or retired-worker benefits would be reduced by 0.5 percent a year. Thus, workers becoming eligible in 2032 or later would receive about 84 percent of the benefits that they would have received under current law. bThe normal retirement age of workers who turn 62 in 2011 would be 67; that age would increase by two months a year until it reached 70 in 2029 and then would increase by one month every other year for the remainder of the projection period. cBeginning in December 1998, the cost-of-living adjustment would be set to equal the increase in the consumer price index minus 1 percentage point. As under current law, no adjustment would be made in years in which there was no increase. Any reductions would be accumulated until a net increase was achieved in a future year. jected amount for that year under current to improve the financial status of Social Se- law. curity and, instead, presented three alterna- Speeding up the increase in the normal tive plans. Much of the public attention di- retirement age to 67 and then linking it to rected toward those plans has focused on increases in longevity would achieve simi- aspects that involve either requiring work- lar savings. Under the specific option ana- ers to invest a certain percentage of their lyzed here, the age at which retirees would earnings in retirement accounts or investing receive full benefits would rise to 70 in 2029 a portion of the balance in the Social Secu- (for workers born in 1967) and then go up rity trust funds in equities rather than Trea- by one month every other year. This option sury securities. A number of other such would reduce spending by less than 10 per- “privatization” proposals have been made in cent in 2030. recent years. Savings could accrue more rapidly by Ultimately, the success of privatization cutting COLAs, because doing so would af- proposals in preparing the economy for the fect all beneficiaries, not just new ones. Re- retirement of the baby boomers rests on the ducing the COLA by 1 percentage point each extent to which the proposals would increase year, starting with the next COLA, would national saving. Some of the specific provi- also reduce spending in 2030 by about 10 sions in one or more of the plans would do percent. that by slowing the growth in spending for The Advisory Council on Social Security Social Security—for example, through re- considered these and other approaches in its ductions in initial benefits or increases in the recent report. The members of the council normal retirement age. Other provisions were unable to reach a consensus about how could boost national saving if they required

Click to return to Table of Contents D OCUMENTS 889

TABLE 3 Effects of illustrative options for reducing growth in net spending for Medicare Option 2010 2030 2050 2070 Net federal spending as a percentage of GDP Continue current law 3.2 5.5 6.2 6.8 Raise the age of eligibility to 70a 3.1 4.9 5.5 6.1 Collect 50 percent of SMI costs from enrolleesb 2.7 4.6 5.3 5.8 Restructure the Medicare market and limit growth in Medicare’s defined contribution to 4 percent a yearc 2.6 3.9 4.0 4.3 Savings as a percentage of projected spending under current law Raise the age of eligibility to 70a 2111111 Collect 50 percent of SMI costs from enrolleesb 15 15 14 14 Restructure the Medicare market and limit growth in Medicare’s defined contribution to 4 percent a yearc 18 29 35 37

SOURCE: Congressional Budget Office. NOTE: SMI = Supplementary Medical Insurance. aThe age of eligibility for Medicare would be increased to 70 by 2032. bPremiums for Medicare enrollees would be increased to cover 50 percent of SMI costs by 2000. c The growth of Medicare’s per-enrollee contribution would be indexed to the average growth rate of GDP per capita (about 4 percent) after 2000. workers to save more than would otherwise Medicare. The Congress could reduce the be the case or if they raised taxes without number of people eligible for benefits, collect increasing expenditures. more of the costs from beneficiaries, or restructure Medicare to reduce total costs per beneficiary. The estimated effects on net federal Medicare spending for Medicare under illustrative op- Medicare has been highly successful in tions for each approach are shown in Table 3. achieving its primary objective of ensuring access to mainstream medical care for the Reduce eligibility aged and the disabled, but the program’s costs have placed an increasing burden on One way to reduce the number of people eli- the economy. In 1997, Medicare’s spending gible for benefits would be to increase the net of premiums paid by enrollees was 2.3 age of eligibility from 65 to 70, using the percent of GDP. If no changes are made in schedule presented above for increasing the current law, net spending is expected to normal retirement age for Social Security reach 3.2 percent of GDP by 2010 and 6.8 benefits. That approach would ultimately re- percent by 2070. Underlying those projec- duce federal spending for Medicare by about tions is an assumption (consistent with the 11 percent compared with current law. Net trustees’ report) that growth in Medicare’s spending would continue to grow relative to spending per enrollee will gradually slow GDP but at a slower rate, reaching 6.1 per- between 2008 and 2020 to be more in line cent of GDP by 2070. Increasing the age of with growth in the average wage. That as- eligibility, however, would do little to reduce sumption is optimistic, though, since policies total health care costs, and it would lengthen designed to achieve that result are not yet the period during which people who opted in place. for early retirement under Social Security Three fundamental approaches exist for might have difficulty obtaining private insur- slowing the growth in federal spending for ance coverage.

Click to return to Table of Contents 890 D OCUMENTS

Raise premiums risks from higher growth in health care costs would be shifted to health plans and enroll- The second approach would raise the pre- ees. Although the federal subsidy per en- miums collected from enrollees to cover 50 rollee would be smaller than it would be un- percent of Medicare’s costs for Supplemen- der current law, competition among plans tary Medical Insurance (Part B). This option and providers could spur efficiency and in- would reduce net Medicare spending by crease health benefits per dollar spent. about 15 percent a year, but it would sim- For example, Medicare’s defined contri- ply shift costs to beneficiaries rather than bution could be set to equal net spending per constrain the growth in total health care enrollee in 2000 and indexed to the average costs. Without any changes to improve the growth rate of GDP per capita thereafter. efficiency of the Medicare program, premi- Under this option, federal savings would be ums would consume an ever larger share of 29 percent of currently projected spending enrollees’ income, rising from about 3 percent by 2030 and 37 percent by 2070. However, currently to more than 12 percent by 2070. the effects of this option on total costs for a basic benefit package—and therefore on the costs that beneficiaries would bear—are un- Restructure Medicare certain. If the incentives generated only A third approach would be to restructure the enough cost-conscious behavior to match the program, giving patients and providers slowdown in the growth of health care costs greater incentives to make cost-effective per enrollee assumed in the base scenario, choices. One way to do that would be to set enrollees’ premiums would steadily increase, up a system of competing health care plans reaching 23 percent of their average income and limit growth in the amount Medicare by 2070. If, instead, growth in costs per en- contributed toward the premiums charged rollee slowed to match the annual growth by the various plans. In such a restructured in the federal defined contribution, premi- system, Medicare’s fee-for-service sector ums would increase little relative to the av- could be one of the plans, competing for en- erage income of enrollees. rollees on the same basis as all other plans. In practice, the effects of this restructur- Because enrollees would be responsible for ing would probably differ among various any premium amounts in excess of Medi- enrollee groups. Some basic plans would care’s contribution, they would have finan- keep their costs low enough to avoid having cial incentives to be prudent purchasers of to charge supplemental premiums, but ac- health plans. Also, because plans would be cess to providers and quality of services avail- at risk for any costs above their predeter- able in those plans might limit their appeal mined premium collections, they would primarily to low-income enrollees. Higher- have financial incentives to operate effi- income enrollees might gravitate instead to ciently. Control of federal Medicare spend- plans that charged supplemental premiums ing would be assured because the financial and provided better access and quality.

Click to return to Table of Contents D OCUMENTS 891

The 1998 Revision of the United Nations Population Projections

The main results of the 1998 round of population estimates (from 1950 to the present) and population projections (up to 2050) were released by the Population Division of the Depart- ment of Economic and Social Affairs of the United Nations on 28 October 1998. These estimates and projections cover 228 countries and areas of the world (“from Pitcairn with 46 persons to China with 1.3 billion”). Full details of the exercise will be forthcoming by the end of the year in the publication Population Prospects: The 1998 Revision. An excerpt from a “Briefing Packet” issued by the Population Division is reproduced below. The projections, and to a much less pronounced degree the retrospective estimates, are revised biennially. (From time to time, the Population Division also prepares supplementary world population projections, extending the regular projections in regional detail into the long term. For a summary of the most recent such exercise, World Population Projections to 2150, issued in February 1998, see the Documents section in the March 1998 PDR.) Compared with the two earlier sets of projections, the 1998 set shows some marked changes. The global population for 2050 according to the 1998 “medium” projection, perhaps the single most watched and cited summary figure, is 8.9 billion. The corresponding forecast was 9.4 billion in the 1996 revision and 9.8 billion in the 1994 revision. Only to a minor degree are these differences attributable to revised current population data, as the 1998 global estimates in the three sets show only modest shifts. (The global population estimate for 1998 was 5.98 billion in the 1994 revision, as compared with 5.93 billion and 5.90 billion in the 1996 and 1998 revisions, respectively.) Rather they reflect reconsiderations of the likely outlook for the future evolution of fertility and mortality, driven by changed assumptions concerning the most plausible extrapolation of recent population trends. Salient features of this shift include the acceptance of the notion that return to replacement-level fertility is unlikely in countries in which fertility is now below, and often well below, replacement level; a more rapid convergence of fertility to replacement level (albeit not below it) in countries in which fertility is still relatively high; and a darkened outlook on mortality attributable to the AIDS epidemic in significant portions of the developing world.

At mid-1998, world population stood at 5.9 world population reaches 8.9 billion in 2050 billion. It is expected that the world popula- (figure I). tion will reach the 6 billion mark in 1999. The mid-1998 world population stood at Between 1995 and 2000 the world popu- 5,901 million, with 4,719 million (80 per lation is growing at 1.33 per cent per year, cent) in the less developed regions and 1,182 adding an average of 78 million persons each million (20 per cent) in the more developed year. In the mid 21st century world popula- regions. Asia accounted for 3,585 million, i.e. tion will be in the range of 7.3 to 10.7 bil- 61 per cent of the world total. During the lion, depending on the assumed future fer- last two years, Africa’s population (749 mil- tility trends. In the medium variant, the lion in 1998) became larger than Europe’s

Click to return to Table of Contents 892 D OCUMENTS

FIGURE I World population size: Past estimates and medium-, high- and low fertility variants, 1950–2050 (billions) 12 11 High 10 9 Medium 8 7 Low 6 5 4 3 2 1950 1960 1970 1980 1990 2000 2010 2020 2030 2040 2050

SOURCE: United Nations Population Division, World Population Prospects: The 1998 Revision, forthcoming.

(729 million). The population of Latin America ant results in a rapid decline of annual rate and the Caribbean is estimated at 504 million, of population change, to a negative value of and that of Northern America at 305 million. –0.23 per cent per year in the middle of 21st The world population is growing at 1.33 century. The population in 2050 will be 10.7 per cent per year between 1995 and 2000, billion according to the high variant and 7.3 which is significantly less than the peak billion according to the low variant. growth rate of 2.04 per cent in 1965–1970, Ninety-seven per cent of the world popu- and less than the rate of 1.46 per cent in lation increase takes place in the less devel- 1990–1995. The annual population incre- oped regions. Every year the population of ment also declined from its peak of 86 mil- Asia is increasing by 50 million, the popula- lion in 1985–1990 to the current 78 million. tion of Africa by 17 million, and that of Latin It will further decline gradually to 64 mil- America and the Caribbean by nearly 8 million in 2015–2020, and then sharply to 30 lion. Africa has the highest growth rate million in 2045–2050. among all major areas (2.36 per cent). Middle In the medium-fertility variant, it is pro- Africa, Eastern Africa and Western Africa have jected that the annual population growth growth rates of 2.5 per cent and over. Europe, rate will continue declining from 1.33 per on the other hand, has the lowest growth rate cent in 1995–2000 to 0.34 per cent in 2045– (0.03 per cent), with a negative rate of –0.2 2050. From 1804, when the world passed the per cent in Eastern Europe. 1 billion mark, it took 123 years to reach 2 Sixty per cent of the world population billion people in 1927, 33 years to attain 3 increase is contributed by only 10 countries, billion in 1960, 14 years to reach 4 billion in with 21 per cent contributed by India and 1974, 13 years to attain 5 billion in 1987 and 15 per cent by China (table 1). 12 years to reach 6 billion in 1999. It will Currently 2 out of 5 people in the world take 14 years to reach 7 billion in 2013, 15 live in either China (1,256 million) or India years to reach 8 billion in 2028, and, with (982 million). There are eight other coun- the slowing down of population growth, it tries with a population over 100 million: the will take 26 years to reach 9 billion, in 2054. United States of America, Indonesia, Brazil, According to the high-fertility variant, Pakistan, Russian Federation, Japan, Bang- the annual population growth rate will de- ladesh and Nigeria. According to the me- crease more slowly, reaching 0.87 per cent dium-fertility variant projection, by the year per year in 2045–2050. The low-fertility vari- 2050 eight additional countries will have

Click to return to Table of Contents D OCUMENTS 893

TABLE 1 Top ten contributors to world population growth, 1995–2000 (net annual additions in thousands) Net Cumulative No. Country addition Per cent per cent

1 India 15,999 20.6 20.6 2 China 11,408 14.7 35.3 3 Pakistan 4,048 5.2 40.5 4 Indonesia 2,929 3.8 44.2 5 Nigeria 2,511 3.2 47.5 6 United States of America 2,267 2.9 50.4 7 Brazil 2,154 2.8 53.1 8 Bangladesh 2,108 2.7 55.9 9 Mexico 1,547 2.0 57.9 10 Philippines 1,522 2.0 59.8 Sub-total 46,494 59.8 59.8 World total 77,738 100 100

SOURCE: United Nations Population Division, World Population Prospects: The 1998 Revision, forthcoming.

exceeded the 100 million mark: Ethiopia, the tal decreased to 17.5 in 1998, and it will fur- Democratic Republic of the Congo, Mexico, ther decline to 11.5 per cent in 2050. Con- Philippines, Viet Nam, Iran, Egypt and Tur- versely, the world population share of Af- key. The ranking will be somewhat differ- rica increased from 8.8 per cent in 1950 to ent; India will then be the most populated 12.7 per cent in 1988 and is projected to country (1,529 million) followed by China reach 19.8 per cent in 2050. The shares of (1,478 million), the United States of America Asia and Latin America are relatively more (349 million) and Pakistan (346 million). stable at approximately 60 and 10 per cent, According to the medium variant, by respectively. All projection variants yield 2045–2050, 56 countries will experience a similar results with respect to the distribu- negative population growth, including all tion of the world population. European countries, Japan and China. The The United Nations Population Division population of the more developed regions as considered the demographic impact of AIDS a group is expected to reach a peak of 1,617 in 34 countries with a population of at least million in 2020, then it will start a gradual 1 million and an adult HIV prevalence of 2 decline and by 2050 will be 2 per cent per cent or more, or with very large infected smaller than in 1998. By contrast, the popu- adult populations. Among these countries, lation of the less developed regions will in- 29 are in Sub-Sahara Africa, three are in Asia crease by 64 per cent, from 4,719 million in (Cambodia, India and Thailand) and two in 1998 to 7,754 million in 2050. The fastest Latin America and the Caribbean (Brazil and population growth will take place in Africa: Haiti). Of the 30 million persons currently its population will more than double during infected by HIV in the world, 26 million (85 the first half of the 21st century; and Africa's per cent) reside in these 34 countries. share in the world population growth will The 1998 Revision shows a devastating toll increase from the current 22 per cent to 55 from AIDS with respect to mortality and per cent in 2045–2050. population loss. In the 29 African countries Different demographic growth rates lead in which the impact of AIDS was studied, to a redistribution of the world population life expectancy at birth is projected to de- among major geographic areas and groups crease to 47 years in 1995–2000 whereas it of countries. While in 1950, Europe and North- would have expected to have reached 54 ern America accounted for 28.5 per cent of the years, in the absence of the AIDS epidemic, world population, their share of the world to- a loss of 7 years. The demographic impact of

Click to return to Table of Contents 894 D OCUMENTS

TABLE 2 Countries with a TABLE 3 Population of the population of over 100 million, major regions of the world, 1950, 1998 and 2050 (population in 1998 and 2050 (population in millions, medium variant) millions, medium variant) 1998 1950 1998 2050

1 China 1,256 World 2,521 5,901 8,909 2 India 982 More developed regions 813 1,182 1,155 3 United States 274 Less developed regions 1,709 4,719 7,754 4 Indonesia 206 5 Brazil 166 Africa 221 749 1,766 6 Pakistan 148 Asia 1,402 3,585 5,268 7 Russian Federation 147 Europe 547 729 628 8 Japan 126 Latin America and the Caribbean 167 504 809 9 Bangladesh 125 Northern America 172 305 392 10 Nigeria 106 Oceania 13 30 46 2050 SOURCE: United Nations Population Division, World 1 India 1,529 Population Prospects: The 1998 Revision, forthcoming. 2 China 1,478 3 United States 349 4 Pakistan 346 these countries is projected to reach only 47 5 Indonesia 312 years, instead of 64 years in the absence of 6 Nigeria 244 AIDS: 17 years of life expectancy lost to AIDS. 7 Brazil 244 Even in the worst cases, the toll of AIDS 8 Bangladesh 213 is not expected to lead to declines of popu- 9 Ethiopia 170 lation, because fertility in these countries is 10 Democratic Republic of the Congo 160 high. In the hardest-hit country, Botswana, 11 Mexico 147 with an adult HIV/AIDS prevalence of 25 per 12 Philippines 131 cent, the population in 2025 is expected to be 23 per cent smaller than what it would 13 Viet Nam 127 have been in the absence of AIDS. Never- 14 Russian Federation 122 theless, the population is still expected to 15 Iran 115 nearly double between 1995 and 2050. 16 Egypt 115 According to the 1998 Revision, 61 coun- 17 Japan 105 tries of the world exhibit a total fertility rate 18 Turkey 101 (TFR) in 1995–2000 at or below the level of 2.1 children per woman, which is the level SOURCE: United Nations Population Division, World Population Prospects: The 1998 Revision, forthcoming. necessary for the replacement of generations. The combined population of those 61 countries (2.6 billion in 1998) amounts to 44 per AIDS is even more dramatic when one fo- cent of the global population. cuses on the hardest hit countries, for ex- In practically all countries of the more ample the 9 countries with an adult HIV developed regions, fertility is currently sig- prevalence of 10 per cent or more: Botswana, nificantly below 2.1. In 20 of these countries Kenya, Malawi, Mozambique, Namibia, the TFR has stayed at below-replacement Rwanda, South Africa, Zambia and Zimbab- level for more than two decades. In the we. In these countries the average life ex- 1980s–1990s fertility has decreased to levels pectancy at birth is estimated to reach 48 below replacement in several countries from years in 1995–2000 whereas it would have the less developed regions, including all been expected to reach 58 years in the ab- countries in the populous region of Eastern sence of AIDS, a loss of 10 years. By 2010– Asia (except Mongolia). Consequently, in its 2015, the average life expectancy at birth in medium variant, the 1998 Revision assumes

Click to return to Table of Contents D OCUMENTS 895 that fertility in these countries will not re- Overall in the world there are still three turn to replacement level within the time times as many children (30 per cent) as older horizon of the projections, i.e. until 2050. persons (10 per cent). The 1998 Revision for the first time pre- However, in the more developed regions, sents estimated and projected numbers of oc- in 1998 the number of older persons ex- togenarians, nonagenarians and centenar- ceeded that of children for the first time. Italy ians, for all countries of the world. In 1998, has the oldest population, with 60 per cent 66 million persons were aged 80 or over, that more older persons than children. Greece, is, about 1 of every 100 persons. That pro- Japan, Spain and Germany have between 50 portion was 5 times higher in the more de- per cent and 40 per cent more older persons veloped regions than in the less developed than children. By the year 2050, in the me- regions (3.0 versus 0.6 per cent). Among dium variant, in the more developed regions them, 6.4 million were aged 90 years or over, there will be more than twice as many older and about 135 thousand are estimated to be persons as children (figure II). aged 100 or over. The population aged 80 In the less developed regions, the pro- or over is projected to increase almost 6-fold portion of older persons will increase from and reach 370 million in 2050. The number 8 to 21 per cent between 1998 and 2050, of centenarians is projected to increase 16- while that of children will decrease from 33 fold to reach 2.2 million. to 20 per cent. For the world as a whole, the As a result of the combined effects of the proportion aged 60 or over will increase from decrease in fertility and the increase in life the current 10 per cent to 22 per cent in expectancy, the population of the world is 2050, while the proportion aged less than 15 becoming older, with a diminishing propor- will decrease from 30 per cent to 20 per cent. tion of children, aged less than 15 years, and By the year 2050, there will be more older an increase of older persons, aged 60 or over. persons than children in the world.

FIGURE II Proportion of total population aged 0–14 and 60 or over, more and less developed regions, 1950–2050 (medium variant projections)

50%

40%

30%

20%

10%

0% 1950 1960 1970 1980 1990 2000 2010 2020 2030 2040 2050

More developed regions, proportion below 15 years of age More developed regions, proportion aged 60 or over Less developed regions, proportion below 15 years of age Less developed regions, proportion aged 60 or over

SOURCE: United Nations Population Division, World Population Prospects: The 1998 Revision, forthcoming.

Click to return to Table of Contents Click to return to Table of Contents ABSTRACTS

Malthus and the Less Developed the emergence of a more efficient mix of pri- World: The Pivotal Role of India vate and public providers competing on an equal basis. JOHN C. CALDWELL Malthus’s Essay had a powerful influence on how English-speaking people interpreted Immigration Policy Prior to the population issues. This was particularly true 1930s: Labor Markets, Policy with regard to India, Britain’s huge colony, Interactions, and Globalization which, unlike other large agrarian countries Backlash of the less developed world, was progressively described by census statistics and other ASHLEY S. TIMMER demographic observations. Its administrators JEFFREY G. WILLIAMSON and civil servants, both British and Indian, increasingly saw it in Malthusian terms. This What determines immigration policy? The tradition persisted into the twentieth century literature here is not nearly as mature as that and played a powerful role in the establish- for trade policy, so this article must be ment of national family planning programs viewed as an initial effort to establish the in India, Pakistan, Bangladesh, and beyond. main empirical outlines. The authors con- In turn the British experience in India helped struct an index of immigration policy for five to shape the attitudes of the English-speak- countries of immigration—Australia, Argen- ing peoples to poor, densely populated coun- tina, Brazil, Canada, and the United States— tries and rapid population growth. for 1860–1930, that is, during and shortly after the age of mass migration. The exercise reveals that the doors to the New World did not suddenly slam shut on immigrants Population and Reproductive after World War I, as is typically illustrated Health: An Economic Framework by citing the passage of the Emergency for Policy Evaluation Quota Act by the US Congress in 1921. In- stead, there was a gradual closing of the JERE R. BEHRMAN doors, although the rate and timing of the JAMES C. KNOWLES closing varied across countries. The authors find that poor wage performance and the This article provides a standard economic perceived threat from more, low-quality for- framework to evaluate policies in the popu- eign workers were the main influences on lation and reproductive health fields and il- shifts in immigration policy. They also offer lustrates its use in order to facilitate cross- some support for the idea that immigration disciplinary exchanges between economists policy was as much an interactive process as and others working in these areas. This were the tariff policies of the time. framework justifies policy interventions to increase efficiency and productivity and to redistribute resources. The article illustrates this framework with a cost-benefit analysis The Onset of Fertility Transition of a safe motherhood project in Indonesia in Pakistan and a distributional analysis of family planning and reproductive health services in ZEBA A. SATHAR Vietnam. Application of the policy frame- JOHN B. CASTERLINE work to a number of resource and finance issues in population and reproductive health Recent trends in fertility and contraceptive suggests that a significant program bias fa- prevalence indicate that the fertility transi- vors publicly provided services and hinders tion in Pakistan has begun in the 1990s. Be-

POPULATION AND DEVELOPMENT REVIEW 24(4) (DECEMBER 1998) 897

Click to return to Table of Contents 898 A BSTRACTS / FRENCH fore that decade, the total fertility rate had Low-Skilled Immigrants and the exceeded six births per woman for at least Changing American Labor Market three decades, and fewer than 10 percent of married women practiced contraception. The MARÍA E. ENCHAUTEGUI most recent survey data, collected in 1996– 97, show a total fertility rate of 5.3 births per Low-skilled immigrants are an integral part woman and a contraceptive prevalence rate of the US labor market: almost 30 percent of 24 percent. Underlying this development of US workers without a high school diploma are macroeconomic trends that have led to are immigrants. If current trends persist, im- widespread economic distress at the house- migrants will become the majority of US low- hold level, and social changes that have di- skilled workers in the near future. While luted the influence of extended kin and re- low-skilled immigrants maintain strong em- sulted in greater husband–wife convergence ployment levels, they are concentrated in the in reproductive decisionmaking. The direct most menial low-skilled jobs, and their wages causes of declining fertility are a crystalliza- are declining relative to those of natives. The tion of existing desires for smaller families, substantial deterioration of the economic sta- along with a decline in family size desires and tus of low-skilled immigrants in the last de- a reduction in the social, cultural, and psy- cade raises important policy questions con- chic costs of contraception. Improvements in cerning ways to address the plight of this family planning services have contributed growing segment of US workers. little to the onset of fertility decline but could be decisive in sustaining the decline over the next decade.

Malthus et les pays en La population et la santé en matière développement : le rôle pivot de de reproduction : plan d’ensemble l’Inde économique pour l’évaluation des politiques JOHN C. CALDWELL JERE R. BEHRMAN La thèse de Malthus a eu une profonde JAMES C. KNOWLES influence sur la façon dont les personnes d’expression anglaise ont interprété les Le présent article offre un plan d’ensemble questions démographiques, surtout en ce qui économique standard visant à évaluer les concerne l’Inde, l’immense colonie politiques de population et de santé en britannique. En effet, à la différence des autres matière de reproduction, et fait ressortir son pays agraires en voie de développement, utilité lors d’échanges interdisciplinaires entre l’Inde a été définie progressivement par des économistes et autres professionnels du statistiques de l’état de la population et domaine. Ce plan d’ensemble justifie les d’autres observations démographiques. Ses interventions de principe qui servent à administrateurs et ses fonctionnaires, tant accroître l’efficience et la productivité et à britanniques qu’indiens, en sont venus à la redistribuer les ressources. Le présent article percevoir en termes malthusiens. Cette met en évidence ce plan d’ensemble au moyen tradition s’est prolongée jusqu’au XXe siècle d’une analyse coût-avantage d’un projet de et a joué un rôle important lors de la création maternité sans risques en Indonésie, ainsi que de programmes de planification familiale en d’une analyse relative à une distribution des Inde, au Pakistan, au Bangladesh et au-delà. services de planification familiale et de santé L’expérience britannique en Inde, quant à en matière de reproduction au Vietnam. elle, a contribué à façonner les attitudes des L’application du cadre d’action à un nombre pays d’expression anglaise envers les pays de ressources et de questions financières pauvres, à forte densité de population et à concernant des programmes de population et croissance démographique rapide. de santé en matière de reproduction suggère

Click to return to Table of Contents A BSTRACTS / FRENCH 899 qu’une partialité importante dans les qu’une transition procréatrice a débuté au programmes favorise les services publics et Pakistan dans les années 1990. Durant les trois empêche l’émergence de meilleurs décennies antérieures, l’indice synthétique de organismes dispensateurs de services privés fécondité dépassait les six naissances par et publics qui pourraient se livrer con- femme, et moins de 10 pour-cent des femmes currence sur une base égale. mariées pratiquaient la contraception. Les données du plus récent sondage effectué en 1996-1997 révèlent un indice synthétique de Politique en matière d’immigration fécondité de 5,3 naissances par femme et un avant les années 1930 : marché de taux d’utilisation de contraceptifs de 24 pour- l’emploi, interactions des politiques cent. À la base de ce développement, on et contrecoup de la mondialisation retrouve des tendances macro-économiques qui ont suscité des contraintes économiques généralisées dans les ménages, ainsi que des ASHLEY S. TIMMER changements sociaux qui ont dilué JEFFREY G. WILLIAMSON l’influence de la famille et rapproché maris Quels sont les facteurs déterminants d’une et femmes en ce qui concerne les décisions politique en matière d’immigration ? La en matière de reproduction. Les causes documentation sur cette question est loin directes du taux de fécondité à la baisse sont d’être aussi évoluée que celle qui traite de la la cristallisation du désir d’avoir des familles politique commerciale. Le présent article doit moins nombreuses, de pair avec un déclin donc être perçu comme un premier effort du nombre des enfants désirés et d’une visant à tracer les grandes lignes empiriques. réduction des coûts sociaux, culturels et Les auteurs ont créé un répertoire des psychiques en matière de contraception. politiques en matière d’immigration pour cinq L’amélioration des services de planification pays d’immigration, à savoir l’Australie, familiale a peu contribué à la baisse du taux l’Argentine, le Brésil, le Canada et les États- de fécondité, bien qu’elle puisse bien Unis, de 1860 à 1930, c’est à dire pendant et constituer un facteur décisif pour maintenir un peu après l’ère de migration massive. la tendance à la baisse au cours de la L’étude a révélé que les portes du Nouveau prochaine décennie. Monde ne se sont pas brusquement refermées pour les immigrants après la Première guerre mondiale, comme la promulgation du Immigrants peu spécialisés et le Emergency Quota Act par le Congrès des États- marché de l’emploi en mutation aux Unis en 1921 l’avait fait croire. Les portes se États-Unis sont plutôt refermées graduellement à un rythme propre à chaque pays. Les auteurs ont MARÍA E. ENCHAUTEGUI conclu que la mauvaise performance salariale et la menace apparente de l’arrivée de Les immigrants peu spécialisés font partie travailleurs étrangers de qualité inférieure ont intégrante du marché de l’emploi aux États- été les facteurs principaux qui ont le plus Unis. Presque 30 pour-cent des ouvriers sans influencé les changements de politiques en diplôme d’études secondaires aux États-Unis matière d’immigration. Ils ont également sont des immigrants. Si les tendances actuelles suggéré que les politiques en matière se maintiennent, les immigrants compteront d’immigration constituent un processus très bientôt pour la majorité des ouvriers peu interactif du même ordre que les politiques spécialisés aux États-Unis. Bien que les tarifaires de l’époque. immigrants peu spécialisés conservent des niveaux d’emploi élevés, on les retrouve principalement dans les emplois les moins L’apparition d’une transition spécialisés, et leur salaire sont à la baisse par procréatrice au Pakistan rapport à celui de ceux qui sont nés aux États- Unis. La détérioration substantielle du statut économique des immigrants peu spécialisés au ZEBA A. SATHAR cours de la dernière décennie soulève des JOHN B. CASTERLINE questions administratives sérieuses sur le sort Des tendances récentes en matière de fécondité de ce segment grandissant des ouvriers aux et d’utilisation de contraceptifs indiquent États-Unis.

Click to return to Table of Contents 900 A BSTRACTS / SPANISH

Malthus y el mundo menos familiar y de salud reproductiva en Viet- desarrollado: El papel central de la Nam. La aplicación del esquema de políticas India a varios aspectos de recursos y finanzas en población y salud reproductiva sugiere que un sesgo importante de programa favorece JOHN C. CALDWELL a los servicios proveídos públicamente y El Ensayo de Malthus ejerció una enorme dificulta el surgimiento de una combinación influencia sobre cómo el mundo de habla más eficiente de proveedores privados y inglesa interpretó los problemas relacionados públicos compitiendo en base de iguales. con la población. Esto se manifestó especial- mente con respecto a la India, la enorme colonia británica, que a diferencia de otros grandes países agrarios del mundo menos desarrollado, fue descrita de manera progresiva por medio de estadísticas de censo La política de inmigración anterior a y otras observaciones demográficas. Sus los años 1930: Mercados laborales, administradores y empleados públicos, tanto interacción de políticas y reacción británicos como indios, la vieron más y más ante la globalización en términos maltusianos. Esta tradición perduraría hasta el siglo veinte, jugando un ASHLEY S. TIMMER papel poderoso en el establecimiento de pro- JEFFREY G. WILLIAMSON gramas nacionales de planificación familiar en India, Pakistán, Bangladesh y allende. A ¿Qué determina la política de inmigración? su vez, la experiencia británica en India Las publicaciones en esta área no tienen la ayudó a moldear las actitudes del mundo de misma madurez que las de las políticas de habla inglesa hacia países pobres densamente comercio, por lo cual debe mirarse este poblados y hacia el crecimiento rápido de la artículo como un esfuerzo inicial para población. establecer los perfiles empíricos más importantes. Los autores construyen un índice de políticas de inmigración para cinco países de inmigración—Australia, Argentina, Brasil, Canadá y los Estados Unidos—para 1860–1930, durante y poco después de la era Población y salud reproductiva: Un de migración en masa. En el ejercicio se esquema económico para evaluar revela que luego de la Primera Guerra políticas Mundial las puertas al Nuevo Mundo no se cerraron súbitamente al inmigrante como se ha típicamente dado a entender citando la JERE R. BEHRMAN aprobación por el congreso de los Estados JAMES C. KNOWLES Unidos en 1921 del Emergency Quota Act Este artículo provee un esquema económico [Ley de cuotas de emergencia]. En lugar de estandard para evaluar políticas en los ello, las puertas se cerraron gradualmente, campos de población y salud reproductiva y y el ritmo y coordinación del cierre varió de se ilustra su uso para facilitar el intercambio acuerdo a los países. Los autores hallaron que a través de disciplinas entre economistas y el rendimiento bajo de salarios y la amenaza otros trabajando en estas áreas. Este esquema percebida de un número mayor de trabaja- justifica las intervenciones de políticas para dores extranjeros de baja calidad fueron las aumentar la eficacia y productividad y para principales influencias sobre los viros en la redistribuir recursos. En el artículo se ilustra política de inmigración. Los autores también este esquema con un análisis de costo- dan cierto apoyo a la idea de que la política beneficio de un proyecto de maternidad de inmigración fue tanto un proceso de segura en Indonesia y un análisis de la interacción como lo fueron las políticas distribución de servicios de planificación arancelarias de aquel tiempo.

Click to return to Table of Contents A BSTRACTS / SPANISH 901

El inicio de la transición de la servicios de planificación familiar al inicio del fecundidad en Pakistán descenso de la fecundida ha sido pequeña pero puede ser concluyente en cuanto al ZEBA A. SATHAR mantenimiento del descenso a través de la JOHN B. CASTERLINE década venidera. Las tendencias recientes en la fecundidad y prevalencia anticonceptiva indican que la transición de la fecundidad en Pakistán ha Inmigrantes poco calificados y el comenzado en los años 1990. Antes de esa mercado laboral cambiente de los década, la tasa global de fecundidad había Estados Unidos sido mayor de seis nacimientos por mujer durante por lo menos tres décadas, y menos MARÍA E. ENCHAUTEGUI de un 10 por ciento de las mujeres casadas Los inmigrantes poco calificados son parte practicaban la anticoncepción. Los datos de integral del mercado laboral en los Estados encuesta más recientes recolectados en Unidos: casi el 30 por ciento de los 1996–97, muestran una tasa global de trabajadores estadounidenses sin diploma fecundidad de 5.3 nacimientos por mujer y secundario son inmigrantes. Si las tendencias una tasa de prevalencia anticonceptiva de 24 actuales persisten, los inmigrantes en el por ciento. Subyacente a este desarrollo hay futuro próximo llegarán a ser la mayoría de tendencias macroeconómicas que han dado los trabajadores poco calificados de los lugar a angustias económicas extensas a nivel Estados Unidos. Aunque los inmigrantes del hogar, y a cambios sociales que han poco calificados mantienen niveles firmes de disminuido la influencia de la familia empleo, se concentran en aquellos trabajos extendida y han resultado en una con- que requieren sólo un mínimo de calificación, vergencia mayor entre esposo-esposa en la y sus salarios disminuyen en relación a toma de decisiones reproductivas. Las causas personas originarias del país. El deterioro directas del descenso de la fecundidad son considerable en la última década de la una cristalización de deseos existentes por situación económica del inmigrante poco tener familias más pequeñas, junto a una calificado da lugar a importantes inter- disminución en el tamaño final deseado de rogantes de políticas con respecto a cómo familia y una reducción del costo social, responder a la situación crítica de este creciente cultural y psíquico de la anticoncepción. La segmento de trabajadores estadounidenses. contribución de un mejoramiento en los

Click to return to Table of Contents AUTHORS FOR THIS ISSUE

JERE R. BEHRMAN is Professor of Economics, University of Pennsylvania.

JOHN C. CALDWELL is Emeritus Professor of Demography, Australian National University, Canberra.

JOHN B. CASTERLINE is Senior Associate, Policy Research Division, Population Council.

IAN CASTLES is Vice President, Academy of the Social Sciences in Australia. He was Australian Statistician from 1986 to 1994.

PAUL DE SA is Research Fellow, Belfer Center for Science and International Affairs, Harvard University.

MARÍA E. ENCHAUTEGUI is Assistant Professor of Economics, University of Puerto Rico, Río Piedras.

JAMES C. KNOWLES is an economic consultant, Chapel Hill, North Carolina.

FREDERICK A. B. MEYERSON is Director, Global Change Policy Project, Yale Center for Environmental Law and Policy, and Lecturer in Environmental Studies, Yale University.

MARK PERLMAN is University Professor of Economics Emeritus, University of Pittsburgh.

ZEBA A. SATHAR is Associate, Population Council, Islamabad, Pakistan.

ASHLEY S. TIMMER is Assistant Professor of Public Policy, Duke University.

JEFFREY G. WILLIAMSON is Laird Bell Professor of Economics, Harvard University.

902 POPULATION AND DEVELOPMENT REVIEW 24(4) (DECEMBER 1998)

Click to return to Table of Contents POPULATION AND DEVELOPMENT REVIEW

CONTENTS TO VOLUME 24, 1998

ARTICLES

Bahchieva, Raisa A., see Michael S. Rendall Banja, Fatoumatta, see Caroline Bledsoe Behrman, Jere R. and James C. Knowles, Population and reproductive health: An economic framework for policy evaluation 4: 697–737 Bledsoe, Caroline, Fatoumatta Banja, and Allan G. Hill, Reproductive mishaps and West- ern contraception: An African challenge to fertility theory 1: 15–57 Bongaarts, John and Griffith Feeney, On the quantum and tempo of fertility 2: 271– 291 Brennan, Ellen, see Martin Brockerhoff Brockerhoff, Martin and Ellen Brennan, The poverty of cities in developing regions 1: 75–114 Bryant, John, Communism, poverty, and demographic change in North Vietnam 2: 235–269 Caldwell, John C., Malthus and the less developed world: The pivotal role of India 4: 675–696 Casterline, John B., see Zeba A. Sathar Eberstein, Isaac W., see Robert A. Hummer Feeney, Griffith, see John Bongaarts Golini, Antonio, How low can fertility be? An empirical exploration 1: 59–73 Hill, Allan G., see Caroline Bledsoe Huang Banghan, see Kay Johnson Hummer, Robert A., Richard G. Rogers, and Isaac W. Eberstein, Sociodemographic differentials in adult mortality: A review of analytic approaches 3: 553–578 Johnson, Kay, Huang Banghan, and Wang Liyao, Infant abandonment and adoption in China 3: 469–510 Knowles, James C., see Jere R. Behrman Lesthaeghe, Ron, On theory development: Applications to the study of family formation 1: 1–14 Reher, David Sven, Family ties in Western Europe: Persistent contrasts 2: 203–234

POPULATION AND DEVELOPMENT REVIEW 24(4) (DECEMBER 1998) 903

Click to return to Table of Contents 904 C ONTENTS TO VOLUME 24

Rendall, Michael S. and Raisa A. Bahchieva, An old-age security motive for fertility in the United States? 2: 293–307 Rogers, Richard G., see Robert A. Hummer Sathar, Zeba A. and John B. Casterline, The onset of fertility transition in Pakistan 4: 773–796 Seltzer, William, Population statistics, the Holocaust, and the Nuremberg trials 3: 511– 552 Timmer, Ashley S. and Jeffrey G. Williamson, Immigration policy prior to the 1930s: Labor markets, policy interactions, and globalization backlash 4: 739–771 Wang Liyao, see Kay Johnson Williamson, Jeffrey G., see Ashley S. Timmer Zlotnik, Hania, International migration 1965–96: An overview 3: 429–468

NOTES AND COMMENTARY

Ahlburg, Dennis A., Julian Simon and the population growth debate 2: 317–327 de Sa, Paul, Population, carbon emissions, and global warming: Comment 4: 797– 803 Ipsen, Carl, Population policy in the age of Fascism: Observations on recent literature 3: 579–592 McNicoll, Geoffrey, Malthus for the twenty-first century 2: 309–316 Meyerson, Frederick A. B., Population, carbon emissions, and global warming: The forgotten relationship at Kyoto 1: 115–130 Meyerson, Frederick A. B., Toward a per capita–based climate treaty: Reply 4: 804– 810

DATA AND PERSPECTIVES

Enchautegui, María E., Low-skilled immigrants and the changing American labor market 4: 811–824 Marcoux, Alain, The feminization of poverty: Claims, facts, and data needs 1: 131– 139 Pollard, Michael S. and Zheng Wu, Divergence of marriage patterns in Quebec and elsewhere in Canada 2: 329–356 Wilmoth, John R., Is the pace of Japanese mortality decline converging toward international trends? 3: 593–600 Wils, Annababette and Anne Goujon, Diffusion of education in six world regions, 1960– 90 2: 357–368