63069 Determinants and Consequences of High Fertility: A Synopsis of the Evidence Portfolio Review The World Bank June 2010 Determinants and Consequences of High Fertility: A Synopsis of the Evidence June 2010 iii Contents Acknowledgments iv Acronyms and Abbreviations vi Introduction 1 Consequences of High Fertility 3 Consequences for Health (Child and Maternal) 4 Consequences for Economic Growth 6 Consequences for the Natural Environment 8 Determinants of High Fertility 11 Supply of Children 12 Motivation: Mortality Change 13 Demand for Children: Economic and Social Determinants 15 Demand for Children: Current Patterns in High Fertility Countries 16 Costs of Fertility Regulation 17 Policy Options 20 References 22 Figures Figure 1. Total ODA Commitments for Health, High-Fertility Countries, 1995–2007 2 Figure 2. Family Planning Program Effort Scores for Major Developing Regions (percent maximum) 18 Boxes Box 1. Main Points on the Implications of High Fertility 3 Box 2. Main Points on the Determinants of High Fertility 11 iv Acknowledgments T his report was prepared by John B. Caster- Bank Review of Population and High Fertility, line (Robert T. Lazarus Professor in Popu- with an external advisory group comprising: lation Studies, Department of Sociology, Stan Bernstein (United Nations Population Ohio State University). This paper is based Fund), John Bongaarts (Population Council), on two reports: (i) Determinants of Fertility in John Casterline (Ohio State University), Bar- Developing Countries: A Review of Recent Lit- bara Crane (IPAS), Adrienne Germain (In- erature prepared by Pia Axemo (HDNHE), Ro- ternational Women’s Health Coalition), Jean dolfo Bulatao (HDNHE), Sadia Chowdhury Pierre Guengant (L’Institut de recherché pour (HDNHE), Anu Garg (HDNHE), Inez Mik- le développement), Jose Guzman (United kelson-Lopez (HDNHE), Brian Chin (Uni- Nations Population Fund), Karen Hardee versity of Pennsylvania Population Studies (Population Action International), Daniel Center), Ajay Tandon (EASHH), and Vaibhav Kraushaar (Bill and Melinda Gates Foun- Gupta (HDNHE); and (ii) Implications of Sus- dation), Gilda Sedgh (Guttmacher Insti- tained High Fertility: A Review of the Evidence tute), Amy Tsui (Johns Hopkins University, prepared by Samuel Mills (HDNHE), See- Bloomberg School of Public Health), and meen Saadat (HDNHE), Neelakshi Medhi Wasim Zaman (International Council on (HDNHE), Sadia Chowdhury (HDNHE), Management of Population Programmes). and Ijeoma Okeigwe (University of California- The World Bank advisory group comprised: San Francisco School of Medicine). Martha Ainsworth (IEGWB), Peter Berman The authors are grateful to the World (HDNHE), Eduard Bos (HDNHE), Ro- Bank Library Research Services for assisting dolfo Bulatao (HDNHE), Hugo Diaz Etch- with the literature search. Mukesh Chawla, evere (HDNVP), Rama Lakshminarayanan Sector Manager, and Julian Schweitzer, Sector (HDNHE), John May (AFTHE), Eliza- Director, of the Health, Nutrition, and Popu- beth Lule (AFTQK), and Thomas Merrick lation unit of the Human Development Net- (WBIHS). work (HDNHE) at the World Bank provided Editing services provided by General Ser- overall guidance and support. Thanks to Vic- vices Department Translation and Interpre- toriano Arias (HDNHE) for providing ad- tation Services (GSDTI). The authors would ministrative support. like to thank the government of the Neth- This case study was part of a larger World erlands, which provided financial support Bank Economic and Sector Work (ESW) en- through the World Bank-Netherlands Partner- titled Addressing the Neglected MDG: World ship Program (BNPP). Determinants and Consequences of High Fertility | A Synopsis of the Evidence v Correspondence Details: Æ This report is available on the following website: http://www.worldbank.org/ Æ Sadia Chowdhury (HDNHE), World hnppublications. Bank, Mail Stop G7-701, 1818 H Street N.W., Washington, DC 20433, USA, Tel: 202-458-1984, email: schowdhury3@ worldbank.org vi Acronyms and Abbreviations AIDS Acquired Immunodeficiency HIV Human Immunodeficiency Virus Syndrome MDG Millennium Development Goal DHS Demographic and Health Survey TFR total fertility rate GDP gross domestic product Determinants and Consequences of High Fertility | A Synopsis of the Evidence 1 Introduction I n the six decades since 1950, fertility has substantial fertility decline. But uncertainty fallen substantially in developing coun- remains as to how rapidly that decline will tries. Even so, high fertility—defined as occur, what policies and programs can accel- five or more births per woman over the re- erate decline, and whether fertility will fall to productive career—characterizes 33 coun- low levels (i.e., less that 2.5 births per woman) tries.1 Twenty-nine of these countries are in in all countries. Sub-Saharan Africa. High fertility poses health The high-fertility countries lag in many risks for children and their mothers, detracts development indicators, as reflected for ex- from human capital investment, slows eco- ample in their rate of progress toward nomic growth, and exacerbates environmental achievement of the Millennium Develop- threats. These and other consequences of high ment Goals (MDGs). These countries have fertility are reviewed in the first half of this also received less development assistance for paper. Recognizing these detrimental con- population and reproductive health than sequences motivates two inter-related ques- countries more advanced in their transitions tions that are addressed in the second half of to lower fertility, and the assistance they did the paper: Why does high fertility persist? and receive increased only marginally from 1995 What can be done about it? to 2007, a period during which commit- In recent years demographic concerns ments to both health and HIV/AIDS rose have shifted increasingly to the consequences substantially, as shown in figure 1. of fertility decline, such as population aging, and to other demographic phenomena such as urbanization. Although high fertility persists in some countries, based on global experience 1 This is as of 2000–2005, the last period for which since 1950 there is good reason to expect that the United Nations (2009) makes data-based esti- these countries too will eventually experience mates rather than relying on projections. 2 Figure 1 | Total ODA Commitments for Health, High-Fertility Countries, 1995–2007 All recipient countries 35 high fertility countries: TFR > 5 15000 15000 ODA commitments, current US$ ODA commitments, current US$ 10000 10000 5000 5000 0 0 1995 1997 1999 2001 2003 2005 2007 1995 1997 1999 2001 2003 2005 2007 Year Year Total Health HIV/AIDS POP/RH Source: OECD DAC. Determinants and Consequences of High Fertility | A Synopsis of the Evidence 3 Consequences of High Fertility H igh fertility is defined as a total fertility ture is relatively rapid population growth rate rate (TFR) of 5.0 or higher. The TFR (and corresponding rapid growth in the size represents the average lifetime births per of successive birth cohorts). These micro- and woman implied by the age-specific fertility macro-level demographic features have con- rates prevailing in one historical period. There sequences that have been identified in a large are micro- and macro-level demographic con- body of research. The key conclusions from comitants of a high TFR. At the micro level, that research are summarized here. they include a relatively high incidence of Assessing the causal impact of high fertility births of order five and above, a relatively high is an analytical challenge because fertility is, to fraction of women experiencing pregnancies a greater or lesser extent, a choice, that is, it is of order five and above, and a greater likeli- endogenous. Covariation of fertility with other hood of short inter-pregnancy intervals. At outcomes—health, social, and economic—may the macro level, the main demographic fea- reflect deliberately chosen trade-offs rather than Box 1 | Main Points on the Implications of High Fertility Child health: The risk of mortality in infancy and early childhood is greater for higher-order births and closely-spaced births, and when the mother is over age 40. Maternal health: The risk of maternal mortality is greater at higher parities, and younger and older ages. Moreover, fertility decline reduces the lifetime risk of maternal death simply by reducing the average num- ber of pregnancies each woman experiences. Child schooling: Children from large families attain less schooling. And successively larger birth cohorts—a feature of high fertility societies—detract from the quality of schooling by diluting the expenditure per pupil. Economic growth: An exogenous drop in fertility raises productive output in the long-run. And the associa- tion between population growth and economic growth has become more negative since the 1980s. Demographic dividend: Fertility decline assists economic growth via favorable changes in the age-struc- ture—the “demographic dividend” of a larger concentration of the population in the working ages, thereby increasing per capita productivity. The “demographic dividend” contributed substantially to economic growth in East Asia and Latin America in the period since 1960. Natural environment: High fertility (and the resulting population growth) is a direct and proximate cause of looming shortages of fresh water in many countries. Population growth has also contributed to global warming—the contribution may be as much as one-third—and fertility reduction via expanded family plan- ning services is among the more cost-effective strategies for restraining global warming. 4 a straightforward causal effect of fertility. Most infant (0–1 year), early childhood (1–4 years), economic analyses adopt a multi-period model and under-five (0–5 years). Birth orders 2 in which investment by adults in children and and 3 show the lowest rates. By comparison, other items is driven by expected returns in the at orders 7+ neonatal mortality is 43 percent same and later periods (e.g., in old age). Both higher and early childhood mortality is 11 number and quality of children are arguments percent higher. in the utility function, i.e. adults must decide Maternal mortality is also more likely at how much to invest in number and in quality. higher pregnancy orders. Some of the best evi- The empirical research summarized below is dence comes from the surveillance system data sensitive to this analytical challenge, but re- in Matlab thana, Bangladesh. These data re- search designs that support valid identification veal that women with five or more pregnan- of causal effects are generally infeasible. cies have a significantly higher risk of dying A key conclusion from more recent work due to maternal causes. Women at pregnancy on the linkages between population and de- orders five and six suffer roughly 50 percent velopment is the importance of being spe- higher mortality. This differential persisted cific about the channels through which the even as mortality declined from high levels in linkages work. At the macro level, the impact the 1970s to much lower levels in the 2000s of high fertility on other outcomes could be (Chen et al. 1974; DaVanzo et al. 2004). There channeled through the size of the population is a further, less noticed return from avoiding (implications for the natural environment), high fertility: since pregnancy is an absolute re- the rate of population growth (implications quirement for maternal mortality, fewer preg- for budgets), or the age distribution of the nancies lowers the lifetime risk (Campbell and population (implications for economic pro- Graham 2006). This is one reason why a re- ductivity). At the household and individual cent modeling exercise for India concludes that level, high fertility means not only a large family planning would be the most effective number of births by the end of most women’s intervention for reducing pregnancy-related reproductive careers, but also typically a high mortality (Goldie et al. 2010). incidence of pregnancies at young ages, of un- In high-fertility regimes, short inter-preg- planned and unwanted pregnancies, and of nancy intervals occur more often than in low- closely-spaced pregnancies, all of which can af- fertility regimes. (The inter-pregnancy interval fect household and individual welfare. is the period between delivery and the next conception.) Applying multivariate analysis Consequences for Health (Child to DHS data from 52 developing countries, and Maternal) Rutstein (2008) shows that the optimal inter- Children from higher-order births are known pregnancy interval from a health standpoint to be at greater risk of dying during infancy is 36–47 months: adjusted mortality risk ra- and early childhood. One comparative anal- tios for shorter intervals are always substantially ysis (Mahy 2003) of Demographic and Health higher. The population-attributable risk for Survey (DHS) data examines risk of death under-five mortality for avoiding conceptions during four intervals: neonatal (0–4 weeks), at less than 24 months after a birth is 0.13 (i.e., Determinants and Consequences of High Fertility | A Synopsis of the Evidence 5 if there were no conceptions within 24 months encourage reproductive-age couples to make of delivery, under-five deaths would fall by 13 this choice as it is an effort to reduce the prev- percent). The analogous population attribut- alence of large sibling-sets that are obstacles to able risk for avoiding inter-pregnancy intervals the schooling of their members. of less than 36 months is 0.25. These results are A large literature looks at the impact of consistent with empirical evidence that short fertility (number of siblings) on schooling out- inter-pregnancy interval is a putative indepen- comes in developing countries (see thorough dent risk factor for low birth weight (less than reviews in Lloyd 1994, Kelley 1996, Lloyd 2.5 kg), preterm birth (less than 37 weeks ges- 2005). The literature consists largely of mul- tation), and small size for gestational age. tivariate analyses of household-level data. The In contrast to its effect on perinatal out- large majority of these studies find that chil- comes, there is little empirical evidence of the dren from large families attain less schooling, inter-pregnancy interval’s effect on maternal an outcome usually attributed to resource di- health. An analysis of almost half a million lution (i.e., less financial and time investment Latin American women (Conde-Agudelo and per child). Belizan 2000) indicates that short inter-preg- Unusual insight into the effect of fertility nancy intervals (less than six months) are as- on child schooling is afforded by long-term sociated with higher risks of maternal death, analyses of the family planning experiment anemia, third trimester bleeding, premature conducted in Matlab thana in Bangladesh. rupture of the membranes, and puerperal en- The treatment area experienced lower fertility dometritis. But a more recent literature review than the control area for several decades, pre- (Conde-Agudelo et al. 2007) does not confirm sumably because of the provision of family this differential, which in any case refers to a planning services. Joshi and Schultz (2007) small fraction of intervals. show that children in the treatment area com- plete more schooling—roughly one-half stan- Consequences for Human Capital dard deviation more for boys and one-third Investment standard deviation more for girls (not statisti- Child health is one critical human capital in- cally significant). These results are largely free vestment; the research summarized above sug- of the endogeneity bias that damages most gests that high fertility per se places children at other research on this topic. higher health risk. The impact of high fertility Another set of interrelations between fer- on a second critical human capital investment, tility and schooling operates at the macro formal schooling, is considered next. This level. In high-fertility societies, the age struc- topic is bedeviled by the likely endogeneity of ture of the population is young and, in par- fertility decisions: under the quantity-quality ticular, the fraction of the population that is trade-off model originally articulated by Gary school-age will be relatively large. Moreover, Becker, parents consciously decide to have each birth cohort is larger than the previous fewer children in order to invest more per cohort. Lam and Marteleto (2008) show that child, with investment in schooling salient. these macro-level demographic features per- Policy to lower fertility is as much an effort to sist for several decades after family sizes have 6 begun to shrink at the micro level. This lends that an exogenous drop in fertility raises pro- even more importance to the question of ductive output in the long run. Note that this whether the population’s demographic struc- research examines the effect of the overall level ture in itself affects educational attainment. of fertility, and implicitly of the overall popu- The usual supposition is that relatively large lation growth rate, rather than growth rates (and growing) child cohorts exert downward of different age-strata of the population as has pressure on schooling expenditure per child become common in the research literature of (World Bank 1984). But the macro-level evi- the past 15 years (i.e., the concept of “demo- dence provides only mixed support. Kelley graphic dividend”—see below). (2001) reviews the literature and notes that Barro’s work has been followed by a flurry several well-conducted cross-country studies of empirical analyses in the past 15 years after estimate that relative cohort size has no impact a period of relatively little research on this on the share of national budgets allocated to topic in the late 1980s and early 1990s. One schooling. Most of the studies Kelley reviews, major conclusion that emerges from this re- however, do not consider schooling outcomes. cent literature is that the association between Several country-specific studies are suggestive. population growth and economic growth For example, Lam and Marteleto (2005) show has become more negative since the 1980s that declines in the growth rate of its school- (Headey and Hodge 2009). A second conclu- age population partly explain Brazil’s large in- sion is that resource dilution effects are con- crease in school enrollment in the 1990s. sequential (an inference from the fact that the effect of population growth on economic Consequences for Economic growth diminishes if one controls for invest- Growth ment or savings). In general there is a negative correlation be- Economists now recognize that to assess tween fertility and economic growth. This the effect of fertility (and concomitant pop- simple correlation, however, cannot be re- ulation growth rates) on economic growth garded as revealing the true causal relation- one must take account of the population’s age ship between fertility and economic growth. structure. Three age-strata are distinguished: Barro (1991) provides a theoretical frame- children (pre-working-age); working-age work for incorporating fertility (or population adults; and the elderly. This categorization is growth) in models of economic growth. In ad- applied to a stylized yet typical fertility transi- dition, his neoclassical growth model contains, tion that unfolds in three phases. In Phase I, as basic arguments, human capital investment fertility is high, and therefore the population is and technological change. young (i.e., a relatively large fraction are chil- Barro (1991, 1997) tests the model using dren); where mortality has declined, this makes panel data (1960–1990) from 100 countries the population even younger on average. In and finds that fertility has a negative impact Phase II, fertility has begun to decline, re- on productive output, reflecting expenditure sulting in successively smaller birth cohorts on child-rearing rather than production of and a bulge in the population in the working goods (income generation). Barro concludes ages (due to high fertility in the past). In Phase Determinants and Consequences of High Fertility | A Synopsis of the Evidence 7 III, lower fertility has persisted for decades and anisms may wane. This in turn generates an the population becomes markedly older (i.e., a incentive for individuals to accumulate assets relatively large fraction are elderly). that they may draw on once they retire from The notion of a “demographic dividend” the labor force. The consequence is higher sav- follows from this typical historical pattern of ings rates, which, all else being equal, pro- fertility transition. In fact, two “dividends” can duce increased economic growth. This is the be identified (Lee and Mason 2006). In Phase “second dividend” (Bloom et al. 2003, Lee II there is growth, sometimes rapid, in the and Mason 2006). In contrast to the first divi- fraction of the population that is working-age; dend, the second dividend is not a direct con- or, equivalently, there is a decline in the depen- sequence of fertility decline and it can last dency ratio (the ratio of children and elderly indefinitely. to those of working-age). Everything else being Most research during the past decade equal, this relatively excessive weighting of the has taken into account age-structure, often working-age population results in increased by examining the impact of the growth rates productivity per capita for the population as of age-strata (children, working-age, elderly) a whole and, therefore, in increased economic rather than the growth rate of the popula- growth (Bloom and Williamson 1998, Bloom tion as a whole. Kelley and Schmidt (2005) et al. 2003, Lee and Mason 2006). (Whether attempt a synthesis of the demographic im- output per worker also grows is less certain; see pact on economic growth. Their main point Kelley and Schmidt (2005).) This phase ends is that this impact is multifaceted. Certain in- when sustained lower fertility leads to a rela- fluences are negative, others are positive; some tively smaller labor force and a return to higher are felt immediately, some with lags of 10–20 dependency ratios. The increased productivity years (or longer). In this vein, Kelley and per capita during this phase is the “first divi- Schmidt specify effects of fertility through dend.” Note that this dividend cannot be re- dependency ratios and through population alized without a decline from high fertility; density and size (demographic features that formal demographic models demonstrate that change relatively slowly) as well as through high fertility societies are necessarily character- the growth rate of the youth population and ized by high youth dependency. Note also that the working-age population (demographic the opportunity to take advantage of this divi- features that change relatively rapidly). The dend is fleeting, although in some countries it empirical analysis uses data for 86 countries may extend for as long as five decades; hence for the period 1960–1995. It reveals that, the term “demographic window.” among the demographic variables, declines A second dividend can be induced by the in youth dependency—a direct result of fer- aging of the population if this age-structure tility decline—have had the strongest influ- change generates an incentive to save. As the ence on output per capita. This result holds elderly become more numerous in relative for all continents with the exception of Africa, terms and if individuals recognize this demo- where youth dependency has remained high graphic fact, confidence that adequate old-age throughout the second half of the 20th cen- support will be provided by state or kin mech- tury due to persistent high fertility. 8 Demographic dividends are not auto- and national studies, such as Heilig’s (1997) matic. Among the conditions that improve the assessment of land-use change in China. High prospects of realizing a dividend are human fertility and population growth is an over- capital investment (i.e., a healthy and edu- arching factor whose effects on the natural en- cated labor force) and government policy that vironment may be profound but difficult to creates a favorable environment for financial calculate with precision. In the following para- investments and encourages household sav- graphs, what is known about the impact of ings (Bloom et al. 2003). In the absence of high fertility and concomitant rapid popula- such conditions, growth in the working-age tion growth on various aspects of the natural population may lead to high unemployment environment are reviewed in turn. and attendant social ills. Bloom and Can- Undoubtedly, population growth leads ning (2006) use the East Asian Tigers and Ire- to changes in land-use patterns: rural areas be- land to underscore how prior investments in come more intensively farmed, grazed, or schooling can foster larger dividends during logged, while at the same time urban growth the demographic window. In Ireland and in absorbs formerly rural areas. Population in- East Asia, the exceptional demographic divi- crease has led to reduced forest cover in Costa dend can almost certainly be attributed in Rica (Rosero-Bixby and Palloni 1998), Ec- part—perhaps in large part—to prior public uador (Pan et al. 2007), and Brazil (Vanwey investments in schooling, especially at the sec- et al. 2007). There is evidence of the same oc- ondary level. currence in Africa and Asia (Carr et al. 2005). But the net effect of population growth and Consequences for the Natural population density on deforestation appears to Environment be relatively weak (Angelsen and Kaimowitz The research base is less conclusive regarding 1999), and deforestation is a land-use change the impact high fertility has on the natural that is not unambiguously harmful, as it de- environment than it is regarding the impact pends on the alternative uses to which the on economic growth. In part this reflects the land is put. Desertification is more clearly an relative infancy of systematic research on this unwelcome development, but for this the de- issue. In fact this is a cluster of issues, since termining role of demographic factors has not various aspects of the natural environment, been ascertained even in vulnerable regions such as land, air, fresh water, biodiversity, and such as Sahelian Africa, North Africa, West global warming, must be distinguished. It is Asia, and South Asia. There are numerous ex- also clear that the effect of fertility and other amples of deliberate efforts to improve land- demographic factors on the natural environ- use practices, organized locally or nationally; ment is heavily conditioned by institutional arguably, increasing population density is an factors such as land-tenure regulations and incentive to engage in such efforts (as initially agricultural practices and by consumption posited by Esther Boserup (1965)). But agri- patterns, and that the effect varies markedly cultural intensification also has limits. Analysis across regions and even between localities. of data from 37 high-fertility countries in Af- These are the conclusions from many global rica reveals a significant relationship between Determinants and Consequences of High Fertility | A Synopsis of the Evidence 9 population pressure, shortening of fallow, soil person daily needs about one liter of water erosion, and soil nutrient depletion (Drechsel for drinking, and more than 1,600 liters are et al. 2001), a set of interrelations described required to produce the grain to feed each on a larger scale by Pimentel and Pimentel person daily (Pimentel and Wen 2004). Some (2006). of the high-fertility countries (Yemen, Af- Degradation in the quality of air is pri- ghanistan, Sahelian Africa) are located in arid marily an urban phenomenon (industrial regions, many of which already suffer from emissions, motor vehicle emissions), although water scarcity. More generally, while in 2000 agricultural practices (biomass burning) can there were 31 countries with populations to- also contribute. Poor air quality in large cities taling 508 million experiencing water stress or in developing countries is well established scarcity, it is expected that by 2025, 48 coun- and can be attributed in part to the growth tries with a combined population of 3 billion of these cities, which in turn is a function of will experience water stress or scarcity (Ber- overall national population growth. Among nstein 2002). This will include the two most the more comprehensive analyses is Nagdeve’s populous South Asian countries, India and (2007) assessment of the impact of population Pakistan. To be sure, available fresh water can growth on air pollution in India. Increased air be used more efficiently: in most countries, pollution in urban areas is attributed to popu- large amounts of fresh water are wasted in ag- lation growth and consumption patterns, and ricultural, industrial, and domestic practices. in rural areas to use of fuel wood, crop resi- Even so, the finite amount of fresh water and dues, animal dung, and low-quality coal. Bio- its uneven distribution around the globe set mass burning in rural areas is a major cause of undeniable constraints. There is little doubt air pollution; indeed, Gustafsson et al. (2009) that high fertility (and the resulting popula- show that biomass burning for both cooking tion growth) is a direct and proximate cause of and agricultural purposes accounts for al- current and looming shortages of fresh water most two-thirds of the carbonaceous aero- in many countries. sols causing brown clouds over South Asia. Declines in biodiversity have multiple di- But these practices that degrade the quality of rect and indirect causes. The latter include air are not inextricably linked to population global warming, which itself is due in part growth, and fertility decline per se in all like- to population growth (see next paragraph). lihood makes a rather small contribution to Population growth can have more direct im- long-term improvement in air quality. pacts on biodiversity, chiefly through changes Numerous studies confirm that popula- in land-use patterns (as discussed above) but tion growth exacerbates the challenge of pro- also because of increased direct contact be- viding adequate fresh water to sustain human tween humans and plant and animal species. life. Although two-thirds of the earth’s surface Luck (2007) conducts a meta-analysis of 85 is water, the supply of fresh water is limited studies (encompassing 401 analyses) of the re- and finite. Hence as the world’s population lationship between human population density grows, the average amount of renewable fresh- and biodiversity. The clear conclusion is that water available to each person declines. Each an increase in population density leads to an 10 increase in the number of threatened and en- a substantial portion of the growth in green- dangered species that have specific environ- house gases can be attributed to popula- mental requirements (as do most species). It tion growth (Dietz and Rosa 1997, York et follows that fertility decline can contribute to al. 2003). In an early assessment, Bongaarts the protection of biodiversity (Chu 2008). (1992) concludes that as much as 35 percent No aspect of the ongoing changes in the of the increase in the emission of greenhouse natural environment garners as much atten- gases has been due to population growth. tion as global warming, attention that is fully Even so, going forward the potential contri- justified by the numerous and diverse re- bution of fertility is probably modest: Bird- percussions that are forecast, including im- sall (2001) estimates that feasible reductions pacts on fresh water, biodiversity, agricultural in fertility in developing countries will reduce production, human health, and human set- global warming through 2050 by only 10 tlement patterns. Global warming is due pri- percent as compared to an unchanging-fer- marily to the burning of fossil fuels, which tility scenario. But this modest contribution is releases carbon dioxide (CO2) and other quite cost-effective as compared to alternative greenhouse gases into the atmosphere. strategies for restraining global warming, as Growth in the emission of greenhouse gases demonstrated in recent calculations by Wire during the past century has run far ahead of (2009), and therefore serves as an important global human population growth. Even so, rationale for further fertility decline. Determinants and Consequences of High Fertility | A Synopsis of the Evidence 11 Determinants of High Fertility V igorous scholarly investigation of the de- whelming body of research. This synopsis fo- terminants of fertility in low-income set- cuses on concepts and empirical findings that tings extends back to the 1960s, and this are especially germane to the task of formu- has produced a rich theoretical and empir- lating strategies for reducing fertility in con- ical literature. For the purpose of developing temporary high-fertility societies, which are policies and programs to reduce fertility in predominantly located in Sub-Saharan Africa. the remaining high fertility societies, one can The Easterlin Synthesis Framework (East- reasonably protest that while the amount of erlin 1975) provides the conceptual frame- knowledge about fertility determinants is ex- work for this review. At issue is whether and tensive it is undifferentiated. It is a challenge when couples are prepared to exercise delib- to distill a few key lessons from this over- erate control over their childbearing via fer- Box 2 | Main Points on the Determinants of High Fertility High demand for children: The demand for children is high in most of the remaining high fertility countries (especially in Central and West Africa). Unmet need for family planning: And yet many of the high fertility countries have moderate to high levels of unmet need for family planning—the prevalence typically ranges from one-fifth to one-third of married women. Age at first union: Age at first union is relatively young in most high fertility societies (less than age 20 on average). Several years delay would contribute to fertility decline, and it would have other health and socio- economic benefits. Mortality: Improved child survival is perhaps the most powerful stimulant of fertility decline. In contrast, increased mortality due to the HIV-pandemic is having minimal overall impact on rates of fertility and popu- lation growth. Education: Formal schooling is second only to mortality as a determinant of fertility. Income: By contrast, income is a relatively weak predictor of fertility decline, net of mortality and education. Poor economic performance is not in itself an obstacle to fertility decline. Obstacles to contraception: Non-access obstacles (cultural, social, psychic) appear to be robust in some settings but are not well quantified. Family planning services: The evidence on access obstacles is less ambiguous: in diverse settings expanded provision of family planning services has had an impact on fertility, typically 10%–25% net reduction in fertility. 12 tility regulation behavior (contraception, tion policies may be intended to affect either induced abortion). This includes limiting the motivation to regulate, by influencing the childbearing to a small number, that is, less demand for children, or the cost of regula- than four children. Fertility regulation be- tion, whereas family planning programs are havior is posited as a direct function of two designed mainly to reduce the cost of regula- constructs, namely motivation to regulate and tion. At the same time, there has been a lively cost of regulation. Motivation to regulate in debate about whether programs can also affect turn is determined by the demand for children the demand for children (see Freedman 1997). (e.g. desired number of children) in relation to the current supply of children; when the cur- Supply of Children rent supply matches or exceeds the demand A fundamental direct determinant of the pace for children, there is motivation to take ac- of childbearing after the onset of the biological tions to avoid becoming pregnant. Note that capacity to reproduce is the age at first birth, motivation is driven primarily by the demand which in turn is typically heavily determined by for children but is also affected by biological the age at entrance to a formal union. An in- factors, themselves conditioned by social and verse association between age at first union and cultural factors, and that these biological fac- lifetime number of births is one of the most tors affect the pace of childbearing (i.e. supply established relationships in the research litera- of children) once a woman becomes physi- ture (Bongaarts 1982). Contemporary high- cally capable of conceiving. The more rapid fertility countries are characterized by early age the pace, the more likely a woman will at any at first union and a resulting early age at first given moment have a stock of children that birth. DHS data reveal that in those countries matches or exceeds her desired number. Age where the TFR is 5.0 or higher, the median age at first birth and inter-pregnancy intervals (as at first marriage averages 17.7 years, almost two determined by postpartum behaviors) are di- years younger than in moderate fertility coun- rect determinants of the supply of children. tries and two-and-a-half years younger than in Cost of regulation is broadly defined to include low-fertility countries. While age at first union not only costs of accessing family planning appears to be increasing in the high- fertility services (financial, time) but other social and countries, the pace is relatively slow. Were the psychic costs, including concern about detri- average age at onset of childbearing to increase mental health side-effects due to contraceptive in the high-fertility countries, almost certainly methods. reduction in the overall fertility rate (TFR) Child survival rates can influence the mo- would follow. Because the ages at first union tivation to regulate by affecting both the stock and first birth are young in these societies, of living children and the demand for chil- there is much scope for fertility reduction due dren. Economic and social factors bear on to this mechanism. both the motivation to regulate fertility, pri- Inter-pregnancy intervals are determined marily through the demand for children, and by behaviors such as breastfeeding (the primary the cost of regulation; the former has received determinant of the length of postpartum amen- far more attention in the literature. Popula- orrhea) and coital frequency (especially post- Determinants and Consequences of High Fertility | A Synopsis of the Evidence 13 partum abstinence). These behaviors have a one of the countries with a TFR below 5.0. substantial effect on individual-level reproduc- Would an improvement in child survival in tive patterns as well as societal levels of fertility the high-fertility countries, a desirable health (Bongaarts 1982). But they do not provide op- outcome in its own right, be a force towards portunities for interventions to reduce fertility fertility decline? in contemporary high-fertility countries be- Mortality decline and fertility decline are cause long durations of breastfeeding and post- entwined in classical demographic transition partum abstinence are the norm in most of theory, with mortality decline leading and mo- these countries. Indeed, a concern is that ero- tivating subsequent fertility decline. From a sion of the extended postpartum durations of societal perspective, lower fertility seems an breastfeeding and abstinence will, all else being inevitable, though often delayed, response to equal, lead to an increase in fertility. lower mortality; otherwise population will More generally, it is usually assumed grow relentlessly. Taking this argument to its that substantial fertility decline does not logical conclusion, Cleland (2001) argues that come about due to changes in the spacing of mortality decline is the necessary and sufficient births—i.e., a lengthening of interbirth inter- condition for fertility decline. However, there vals—but, rather, will only occur when there is no supra-individual mechanism to guarantee is widespread adoption of behaviors, usually a fertility response to mortality decline. contraception, intended to terminate child- At the individual level, a complex of bi- bearing after a certain number of living chil- ological and behavioral factors ties indi- dren has been attained (Van De Walle 1992). vidual fertility to infant and child mortality. Timaeus and Moultrie (2008) have chal- Physiologically, the death of an infant affects lenged this presumption, demonstrating in subsequent fertility by leading to a sudden ter- analysis of survey data from South Africa that mination of breastfeeding, triggering resump- postponement of births has contributed sub- tion of menses and ovulation, and leaving stantially to the country’s fertility decline. the woman exposed to the risk of conceiving Whether the same could be replicated else- again. For this reason alone, improved child where in Africa is uncertain; if this potential survival should reduce the number of live exists, it has direct implications for the for- births. Complementing this physiological re- mulation of population policy and programs sponse, three volitional responses have been in the large number of high- fertility countries suggested: a replacement response, an insur- that are African. ance response, and a quality-quantity tradeoff. First, parents may attempt to replace a child Motivation: Mortality Change who dies young in an effort to attain a desired The high fertility countries are characterized number of children. Second, they may pro- by relatively poor child survival. According tect their childbearing goals against possible to United Nations estimates for the period deaths by having extra children, an insurance 2000–2005, in two-thirds of these countries response. Third, and perhaps most impor- the infant mortality rate exceeded 100 deaths tant, as survival prospects improve, parents are per 1,000 births, a level observed in only more likely to invest time and money in their 14 children, leading to a quality-quantity trad- fertility decline in some African settings, for eoff. There is empirical evidence that all four example accounting for 12 percent of the fer- of these mechanisms (physiological and voli- tility decline in South Africa (Camlin et al. tional) are operative, although there is debate 2004) and one-quarter of the fertility de- about their relative magnitudes. cline in Zimbabwe since the 1980s (Zaba et More germane to this report is how these al. 2003). Still, meaningful fertility decline in individual-level responses aggregate into fer- the high-fertility countries cannot come about tility decline. A recent country-level econo- via this mechanism alone: Zaba and Gregson metric analysis for the period 1960–2000 (1998) and Lewis et al. (2004) both estimate (Angeles 2010) concludes that mortality that each percentage-point increase in HIV changes have a large impact on fertility, in- prevalence among adult females is linked to a deed can account for a major part of observed 0.4 percent reduction in fertility, which im- fertility decline in the period since 1960 and plies that a one-birth reduction from a TFR substantially more than can be attributed to of 5.0 (20 percent reduction) would require growth in GDP per capita and urbanization. a 50 percentage-point increase in HIV preva- It appears that fertility decline lags mortality lence, an entirely far-fetched explosion of the decline by 10–20 years on average. This re- pandemic. search provides some basis for expecting that Are there also fertility responses among improved child survival in the high-fertility uninfected women, driven perhaps by concerns countries will motivate fertility decline. Con- about HIV infection? This is an active topic sistent with this expectation, the African of research, and even the direction of the ef- countries with substantial fertility declines in fect (if it exists) is uncertain. Some changes in the 1980s and 1990s were also countries that sexual behavior can be documented in partic- enjoyed some success in reducing infant and ular settings, especially a delay in sexual debut child mortality, namely Botswana, Ghana, among women and a reduction in number of and Kenya. sexual partners due to HIV/AIDS-related con- Because most of the high-fertility coun- cerns (see Gregson et al. 2009 and papers cited tries are African and because the HIV/AIDS therein). Whereas the former change is related pandemic is most severe in this region, the to lower fertility, the impact of the latter is un- question naturally arises whether the pan- clear. Other changes, such as a delay in first demic has any bearing on the course of fer- marriage, can have effects difficult to sort out, tility transition. This question can be refined such as greater premarital activity or greater by considering separately HIV-positive and marital stability. An increase in condom use HIV-negative women. The former are known also has problematic effects, increasing contra- to have diminished fertility—estimates range ceptive protection only if it does not substi- from 10 percent to 50 percent—due pri- tute for more effective hormonal contraceptive marily to higher fetal loss (Gray et al. 1998, methods. That the evidence is so cloudy sug- Zaba and Gregson 1998, Lewis et al. 2004). gests there is no reason to believe that the This diminished fertility among the HIV- HIV/AIDS pandemic will motivate substantial positive has made a large contribution to the fertility decline. Determinants and Consequences of High Fertility | A Synopsis of the Evidence 15 Demand for Children: Economic gradually than fertility. Angeles (2010) esti- and Social Determinants mates a significant net effect of urbanization A vast literature examines the effect on fertility on fertility decline that is smaller in magni- of economic and social factors, from both a tude than the effects of mortality decline and macro-level and a micro-level perspective. educational increase but larger in magnitude The most prominent factors are income (e.g., than the effect of income growth. GDP per capita), urbanization, and educa- With very few exceptions, research in de- tional attainment. veloping countries reveals an inverse relation- At the country level, measures of income ship between the amount of formal schooling and fertility are strongly associated cross-sec- and fertility. In cross-sectional analyses, edu- tionally (Schultz 2006), but the same asso- cation indicators are often the strongest single ciation is not evident in analyses of fertility correlates of fertility at both the macro level change. In Angeles’ (2010) regression analysis and the micro level. (Reviews of the empir- of fertility decline in the period 1960–2000, ical research include Cochrane 1979, Castro for example, GDP per capita has the expected Martin 1995, Jejeebhoy 1995, and Bledsoe negative coefficient but its estimated effect is et al. 1999.) There has been less explicit anal- far weaker than mortality and education. One ysis of the contribution of educational change possible explanation for this relatively weak es- to fertility decline, but the existing research timated effect is that the true effect of income shows robust effects. In Angeles’ analysis, for growth on fertility is heterogeneous, raising example, educational change is second only the demand for children in some subgroups to mortality decline as a predictor of fertility and lowering it in others (as economic theory decline. would predict). It is also possible that Angeles’ Income, urbanization, and formal and other analyses have shortchanged the ef- schooling are key items in a bundle of fac- fect of income by not fully accounting for in- tors that together comprise “socioeconomic direct effects (through other determinants development” as conventionally understood. such as mortality and education). But even al- A common theme in the research literature lowing for some under-accounting for the full of the 1980s and 1990s, epitomized by Bon- causal effect of income, the bulk of the empir- gaarts and Watkins (1996), was that socio- ical evidence suggests that income growth per economic development is weakly linked to se is not essential for fertility decline. fertility decline, both in the European past Fertility is almost always lower in urban and in contemporary developing countries, as compared to rural areas. So too is the de- and indeed might even be secondary to “ide- mand for children (e.g., desired number of ational change” and social diffusion (of fer- births) lower in urban areas. The urban-rural tility attitudes, of knowledge of modern differential typically persists with controls for contraception). But the past decade has wit- confounding variables such as educational at- nessed a re-establishment of the fundamental tainment. Urbanization figures less promi- determining role of socioeconomic factors nently in empirical analyses of fertility decline, such as schooling, and the fact that fertility in part because generally it changes far more remains high in most countries that rank low 16 on development indicators, chiefly African in general the demand for children was not as countries, has been an important reason for high (Mauldin 1965, Lightbourne 1985). this revival of earlier theory of the causes of Successive DHS surveys (from 1990 to fertility decline. Bryant (2007) demonstrates 2008) reveal change in the demand for chil- that the relationship of conventional develop- dren in Eastern and Southern African coun- ment indicators to fertility decline is strong tries but little change in Central or Western and, further, that key results in Bongaarts and African countries, especially the Sahelian Watkins (1996) are methodological artifacts. countries of Western Africa (Casterline 2009). A key concept is unmet need for family plan- Demand for Children: Current ning. This describes the condition of wanting Patterns in High Fertility Countries to avoid pregnancy (temporarily or indefi- Scholars and policy-makers examine fertility nitely) but not practicing family planning. declines in the past for lessons that can be ap- That is, the concept of unmet need juxtaposes plied to the remaining high-fertility societies. reproductive desires and behaviors; it is non- But the relevance of past experience, chiefly use of contraception conditional on a desire Asian and Latin American, to contemporary to postpone or terminate childbearing. De- high-fertility societies, which are concentrated mographic survey data, such as the DHS, pro- in Sub-Saharan Africa, is far from certain. The vide estimates of the prevalence of unmet need prominent demographer John Caldwell has among women of reproductive age. Note that repeatedly cautioned that Asian experience unmet need has been adopted as an indicator might not be applicable to Sub-Saharan Af- for MDG 5. rica (Caldwell and Caldwell 1988, Caldwell While DHS data document high demand et al. 1992). One indication of the distinc- for children in the high-fertility African so- tiveness of African reproductive regimes is the cieties, the same surveys also reveal levels of high demand for children. DHS data show unmet need for family planning that are at that the mean desired number of children ex- least modest and in some countries relatively ceeds 4.0 in all high-fertility African countries high (DHS 2010) (i.e. ranging from 15 per- and sometimes is greater than 6.0 (Westoff cent to 40 percent of women currently in 2010). Similarly, among women who have al- union). This indicates that some fertility de- ready given birth to four children, the fraction cline could be achieved without changes in stating a preference to have no further births is fertility demand. This opportunity notwith- less than 50 percent in most of the high- fer- standing, it should be recognized that sub- tility African countries. Demand for children is stantial decline in the demand for children especially high in Western and Central Africa. is probably a prerequisite if most of the re- While the comprehensive standardized survey maining high-fertility countries are to experi- record provided by the DHS for contempo- ence a decline in fertility to a low level. In this rary African societies is not available for Asian respect, their fertility declines may well differ and Latin American societies at the onset of in character from the fertility declines that oc- their fertility declines in the 1950s–1970s, the curred in Asia and Latin America during the few pieces of empirical evidence suggest that period from the 1960s to the present; in these Determinants and Consequences of High Fertility | A Synopsis of the Evidence 17 declines, changes in the demand for chil- it would be surprising if such obstacles are not dren were rather modest (Casterline 2010). of some significance. While the Asian and Latin American de- There is far more research into access ob- clines consisted largely of the realization of stacles (e.g., financial costs, time costs). Indeed, existing demand for small families (Feyisetan family planning programs, the intervention and Casterline 2000, Casterline 2010), fu- most clearly identified with organized attempts ture declines in the high-fertility countries will to reduce fertility, have as their first goal the re- necessarily more nearly resemble the demand- duction of access costs. The effect on fertility of driven decline postulated by Pritchett (1994). family planning programs has been investigated It follows that if decline in the demand for in multiple settings over the course of five de- children is a requirement, then fertility decline cades. A useful summary of their performance in the remaining high-fertility societies will be is offered by Robinson and Ross (2007), who especially sensitive to changes in factors such draw on an extensive earlier literature and the as mortality, schooling, and urbanization that experience of practitioners and analysts familiar are known to be strong correlates of fertility de- with many separate programs. They assemble mand. The inference is that these factors will studies of 22 countries from each of the major be even more decisive for fertility decline in the regions. Their conclusion is that for the most remaining high-fertility societies than they were part these programs have had a net impact on for fertility declines in the past. But change in fertility that ranges from 6 percent (weak pro- these factors is not easily accomplished, and grams) to 32 percent (strong programs). hence additional policy levers must be sought. The best evidence concerning the ef- One scenario, described below, is that increased fect that enhanced provision of family plan- capacity to control fertility (due to easier ac- ning services has on reproductive behavior cess to family planning services) will itself drive is provided by field experiments. There is a down the demand for children. long list of locations in Asia and Africa where family planning experiments, most of them Costs of Fertility Regulation short-term, have been conducted. In most Non-access obstacles to contraceptive use— instances, the treatment of enhanced family the social and psychic factors that figure planning services yielded more contracep- heavily in anecdotal accounts of unmet need tive use, lower fertility, or both. The most for contraception—are rarely investigated with thoroughly analyzed field experiment was rigor. Where they have been given due con- conducted in Matlab thana, Bangladesh. An sideration, they have been shown to explain intense family planning effort there began a considerable portion of the unexplained in 1977 and led to steadily rising contracep- unmet need (see review in Casterline and tive prevalence in the treatment areas that Sinding 2000). Social barriers (e.g., husbands, far outpaced the increase in the control areas in-laws) and fear of health side-effects pre- (Phillips et al. 1988) and also produced sus- dominate in some societies. At present there is tained fertility decline (Phillips et al. 1988, almost no systematic research on the non-ac- 1996). More recently, Joshi and Schultz cess obstacles in the high fertility societies, but (2007) have returned to the data from the 18 Matlab experiment and, after carefully taking above that the remaining high-fertility soci- into account initial differences between treat- eties are characterized by high average desired ment and control areas, calculate that the family size. Possibly the relatively exceptional number of surviving children (reflecting both demand for children in these societies will be fertility and child survival) was 18 percent more responsive to an increased ability to limit lower in the treatment area after five years family size via modern contraception than evi- and remained 10 percent lower 14 years after dently was the case in Asian and Latin Amer- that. ican societies in the past. An unresolved issue, possibly of critical What is the quality of family planning importance to fertility decline in the high fer- programs in the high-fertility countries? tility countries, is whether enhanced provision Country-specific “program effort” has been of family planning services affects fertility de- assessed periodically since the early 1970s mand. A plausible argument is that increased using informed observers in each country capacity to exercise control over reproduc- (Ross et al. 2007). Figure 2 shows the trend tion lowers the demand for children. That is, in the program effort scores (on a scale from 0 making smaller family size more feasible also to 100) averaged for five major regions. Pro- makes it more desirable (a self-efficacy ef- gram effort improved substantially in all re- fect). The authoritative review on the subject gions over the three-decade period. The figure (Freedman 1997) concludes that the empir- shows that program effort scores in Sub-Sa- ical record does not provide consistent sup- haran Africa, where most of the high- fertility port for this argument. This is a discouraging countries are located, have lagged in the An- conclusion in light of the evidence reviewed glophone countries and more so in the Fran- Figure 2 | Family Planning Program Effort Scores for Major Developing Regions (percent maximum) 70 60 Effort score (0–100) 50 40 30 20 10 0 1960 1970 1980 1990 2000 East, South, and Southeast Asia Anglophone Sub-Saharan Africa Central Asia Middle East and North Africa Latin America and Caribbean Francophone Sub-Saharan Africa Source: Ross, John, John Stover, and D. Adelaja. 2007. “Family Planning Programs in 2004: New Assessments in a Changing Environment.” International Family Planning Perspectives 33:22–30. Determinants and Consequences of High Fertility | A Synopsis of the Evidence 19 cophone countries. Even so, as of 2004 the socioeconomic context is on balance less fa- average program effort score in Africa matches vorable in the high-fertility countries in Sub- the level in Asia and Latin American during Saharan Africa. Experimental interventions their periods of most rapid fertility decline that have succeeded (such as the Navrongo in the 1980s and 1990s. Whether this can experiment in northern Ghana, analyzed in be taken as a favorable portent for future in- Debpuur et al. 2002) have entailed more in- creases in contraceptive use (and decline in tensive effort than national programs can fertility) is uncertain because the health and mount. 20 Policy Options T he synopsis of the research literature pre- tive regime, in particular whether the demand sented above, and especially the review of for children is high or low and whether or the major determinants of fertility (pre- not there is substantial unmet need for family vious section), provides a framework for con- planning. sidering policy options. Over the years a range One might conclude from the evidence of policy options for promoting fertility de- presented above that the first order of business cline have been suggested, encompassing not in most of the high-fertility countries is the only family planning programs but also human implementation of policies that reduce the de- capital investment (health, schooling) not to mand for children. In such settings improved mention interventions that would promote access to reproductive health services may be gender equity and empower women. Many of insufficient; instead there must also be multi- the policies are desirable on multiple grounds sectoral interventions that will, among other and not only for their fertility consequences. outcomes, reduce the demand for children. The underlying fertility determinant tar- But effective interventions that are also afford- geted by the policies varies. Some measures, able within current resource constraints are such as education, are directed at reducing difficult to identify. Substantial improvement the demand for children. Others, such as en- in child survival and mass schooling (with at- couragement of later start of childbearing, in- tendant labor force opportunity upon gradu- fluence fertility by reducing exposure to the ation) would probably drive down desired risk of conception. Finally, family planning family size, but collectively these are extremely services are designed to address the costs of expensive. Developing alternative, more af- regulating fertility, and hence might be ex- fordable policies to reduce the demand for pected first of all to reduce unwanted fertility. children is of high priority, and no doubt will Judging from the cumulative experience of require some imagination. the past five decades, effective implementation Most immediately, providing family plan- of any one of these measures can contribute ning services remains a relatively inexpensive to fertility decline However, the high-fertility and targeted fertility reduction policy. While countries that have been the focus of this re- the demand for children is high in these coun- view have tried few if any of these interven- tries, there is also unmet need for family plan- tions. Family planning is one intervention ning, as indicated by survey data and also by many have tried, though often with indif- the high incidence of induced abortion in ferent commitment and insufficient resources some countries. Addressing this unmet need The most effective interventions will be those is desirable on health grounds and imperative that are tailored to the nature of the reproduc- if women and men are to have the ability to Determinants and Consequences of High Fertility | A Synopsis of the Evidence 21 make voluntary and informed decisions about high fertility countries, and it begins with fertility. Addressing the unmet need would relatively affordable investment in reproduc- also translate into fertility reduction – de- tive health services that address unmet need mographic survey data indicate potential for for family planning while also improving ma- small to modest reduction in most of the high ternal and child health. fertility countries. Policy and programs must also be respon- Whether the demand for children will be sive to the marked inequalities in reproduc- downwardly responsive to improved capacity tive health outcomes that are endemic in most to exercise fertility control remains an un- of these countries. The demand for children tested possibility. Information campaigns and is generally higher among the poor, as noted population education that deliberately target in the literature review above. Unmet need high desired family size could reinforce this is a more complicated matter, because it is a possibility and are clearly mandated in this set function of both the demand for children and of countries, and in fact are already integral to contraceptive behavior. In the early stages of most family planning programs. They might fertility decline, the decline in desired fertility also accelerate fertility decline where it is al- often out-paces the increase in contraceptive ready underway. Explicit and visible political practice among the middle and upper strata commitment at the highest level comes with of society, resulting in higher unmet among no financial cost and would reinforce the in- those who are wealthier and better educated. formation and education messages. Together, But one might presume that these strata pos- these efforts could set off a virtuous circle sess resources to close the gap between repro- in which initial fertility limitation gener- ductive desires and behaviors, albeit with a ates incentives for further fertility limitation, lag. After the early stages of fertility decline, thereby reducing the demand for children. however, the more common predicament A dynamic of this sort probably accounts for that emerges is higher unmet need among the rapid fertility declines witnessed in many the poor, and this can become a chronic con- Asian and Latin American countries in re- dition that is not readily alleviated without cent decades (Casterline 2001). When this public provision of reproductive health ser- dynamic is operative, the public cost of effec- vices. Hence equity considerations become an- tive fertility policy can fall dramatically. This other important rationale for investment in stands as the most promising scenario for the such services. 22 References Angeles, Luis. 2010. “Demographic Transi- in Critical Perspectives on Schooling and tions: Analyzing the Effects of Mortality Fertility in the Developing World, edited on Fertility.” Journal of Population Eco- by C.H. Bledsoe, J.B. Casterline, J. A. nomics 23: 99–120. Johnson-Kuhn, and J. G. Haagas. Wash- Angelsen, Arild and David Kaimowitz. 1999. ington, DC: National Academies Press. “Rethinking the Causes of Deforestation: Bloom, David E., and David Canning. 2006. Lessons From Economic Models.” World “Demographic Challenges, Fiscal Sustain- Bank Research Observer 14: 73–98. ability and Economic Growth,” Program Barro, Robert J. 1997. Determinants of Eco- on the Global Demography of Aging nomic Growth: A Cross-Country Empirical Working Paper 8, Harvard School of Study. Cambridge, Massachusetts: MIT Public Health, Boston. Press. Bloom, David E., David Canning, and Jaypee Barro, Robert J. 1991. “Economic Growth in Sevilla. 2003. The Demographic Divi- a Cross-Section of Countries.” Quarterly dend: A New Perspective on the Economic Journal of Economics 106: 407–443. Consequences of Population Change. Santa Barro, Robert J., and Gary S. Becker. 1989. Monica, California: RAND. “Fertility Choice in Model of Economic Bloom, David E., and Jeffrey G. Williamson. Growth.” Econometrica 57: 481–501. 1998. “Demographic Transitions and Becker, Gary S. 1981. A Treatise on the Family. Economic Miracles in Emerging Asia.” Cambridge, Massachusetts: Harvard Uni- World Bank Economic Review 12: 419– versity Press. 455. Bernstein, Stan. 2002. “Freshwater and Bongaarts, John. 1992. “Population Growth Human Population: A Global Perspec- and Global Warming.” Population and De- tive.” In K. Krchnak (ed.), Human Popu- velopment Review 18: 299–319. lation and Freshwater Resources: US Cases Bongaarts, John. 1982. “The Fertility-Inhib- and International Perspectives. New Haven: iting Effects of the Intermediate Fertility Yale University. Variables.” Studies in Family Planning 13: Birdsall, Nancy. 2001. “Another Look at Pop- 179–189. ulation and Global Warming,” World Bongaarts, John, and Susan C. Watkins. 1996. Bank Policy Research Working Paper “Social Interactions and Contemporary WPS 1020, World Bank, Washington, Fertility Transitions.” Population and De- DC. velopment Review 22: 639–682. Bledsoe, Caroline H., Jennifer Johnson-Kuhn, Boserup, Esther. 1965. The Conditions of and John G. Haaga. 1999. “Introduction” Agricultural Growth: The Economics of Determinants and Consequences of High Fertility | A Synopsis of the Evidence 23 Agrarian Change Under Population Pres- A Fresh Assessment.” Paper presented at sure. London: Allen and Unwin. the Annual Meeting of the Population As- Bryant, John. 2007. “Theories of Fertility sociation of America, Dallas, Texas, April Decline and the Evidence From Devel- 2010. opment Indicators.” Population and Devel- Casterline, John B., and Steven W. Sinding. opment Review 33: 101–127. 2000. “Unmet Need for Family Planning Caldwell, John C., and Pat Caldwell. 1988. in Developing Countries and Implications “Is the Asian Family Planning Program for Population Policy.” Population and De- Model Suited to Africa?” Studies in Family velopment Review 26: 691–723. Planning 19: 19–28. Castro Martín, Teresa. 1995. “Women’s Edu- Caldwell, John C., I. O. Orubuloye, and Pat cation and Fertility: Results From 26 De- Caldwell. 1992. “Fertility Decline in Af- mographic and Health Surveys.” Studies in rica: A New Type of Transition?” Pop- Family Planning 26: 187–202. ulation and Development Review 18: Chen, Lincoln C., M. C. Gesche, S. Ahmed, A. 211–242. I. Chowdhury, and W. H. Mosley. 1974. Campbell, Oona, and Wendy Graham. 2006. “Maternal Mortality in Rural Bangladesh.” “Strategies for Reducing Maternal Mor- Studies in Family Planning 5: 334–341. tality: Getting on With What Works.” Chu, C. Y. Cyrus. 2008. “Biodiversity Decline Lancet 368: 1284–1299. and Population Externalities.” Journal of Camlin, C., Michel Garenne, and Tom Population Economics 21: 173–181. Moultrie. 2004. “Fertility Trend and Pat- Cleland, John. 2001. “The Effects of Im- tern in a Rural Area of South Africa in the proved Survival on Fertility: A Reas- Context of HIV/AIDS.” African Journal of sessment.” Population and Development Reproductive Health 8: 38–54. Review 27 (Suppl.): 60–92. Carr, David, Laurel Suter, and Alisson Barb- Cochrane, Susan H. 1979. Fertility and Edu- ieri. 2005. “Proximate Population Factors cation: What Do We Really Know? Wash- and Deforestation in Tropical Agricultural ington, DC: Johns Hopkins University Frontiers.” Population and Environment Press. 27: 89–113. Conde-Agudelo, A., and J. M. Belizán. 2000. Casterline, John B. 2001. “Introduction.” In “Maternal Morbidity and Mortality As- Diffusion Processes and Fertility Transition: sociated With Interpregnancy Interval: Selected Perspectives, ed. J.B. Casterline. Cross-Sectional Study.” British Medical Washington, D.C.: National Academy Journal 321: 1255–9. Press. Conde-Agudelo, A., A. Rosas-Bermúdez, and Casterline, John B. 2009. “Fertility Desires A. C. Kafury-Goeta. 2007. “Effects of and the Prospects for Fertility Decline in Birth Spacing on Maternal Health: A Sys- Africa.” Manuscript, Department of Soci- tematic Review.” American Journal of Ob- ology, Ohio State University. stetrics and Gynecology 196: 297–308. Casterline, John B. 2010. “Wanted Fertility, DaVanzo, Julie, Abdur Razzaque, Mizanur Unwanted Fertility, and Fertility Decline: Rahman, Lauren Hale, Kapil Ahmed, 24 Mehrab Ali Khan, Golam Mustafa, and phine Hu. 2010. “Alternative Strategies Kaniz Gausia 200. “The Effects of Birth to Reduce Maternal Mortality in India: A Spacing on Infant and Child Mortality, Cost-Effectiveness Analysis.” Plos Medicin- Pregnancy Outcomes, and Maternal Mor- ePLoS Med 7(4): e1000264. doi:10.1371/ bidity and Mortality in Matlab, Ban- journal.pmed.1000264. gladesh ” RAND Labor and Population Gray, Ronald H., Maria J. Wawer, David Ser- Working Paper Series, RAND, Santa wadda, Nelson Sewankambo, Chuanjun Monica. Li, Frederick Wabwire-Mangen, Lynn Debpuur, Cornelius, James F. Phillips, Eliza- Paxton, Noah Kiwanuka, Godfrey Kigozi, beth F. Jackson, Alex Nazzar, and Pierre Joseph Konde-Lule, Thomas C. Quinn, Ngom. 2002. “The Impact of the Nav- Charlotte A. Gaydos, Denise McNairn. rongo Project on Contraceptive Knowl- 1998. “Population-Based Study of Fer- edge and Use, Reproductive Preferences, tility in Women With HIV-1 Infection in and Fertility.” Studies in Family Planning Uganda.” Lancet 351: 98–103. 33: 141–164. Gregson, Simon, J. Todd, and Basia Zabia. Demographic and Health Surveys [DHS]. 2009. “Sexual Behavior Change in Coun- 2010. http://www.statcompiler.com/ tries With Generalised HIV Epidemics? Dietz, Thomas, and Eugene A. Rosa. 1997. Evidence From Population-Based Cohort “Effects of Population and Affluence on Studies in Sub-Saharan Africa.” Sexually CO2 Emissions.” Proceedings of the Na- Transmitted Infections 85(Suppl. 1): I1–I2. tional Academy of Sciences 94: 175–179. Gustafsson, O., M. Krusa, and Z. Zencak. Drechsel, Pay, Lucy Gyiele, Dagmar Kunze, 2009. “Brown Clouds Over South Asia: and Olufunke Cofie. 2001. “Population Biomass or Fossil Fuel Combustion?” Sci- Density, Soil Nutrient Depletion, and ence 323: 495–498. Economic Growth in Sub-Saharan Af- Headey, Derek D., and Andrew Hodge. 2009. rica.” Ecological Economics 38: 251–258. “The Effect of Population Growth on Easterlin, Richard A. 1975. “An Economic Economic Growth: A Meta-Regression Framework for Fertility Analysis.” Studies Analysis of the Macroeconomic Litera- in Family Planning 6: 54–63. ture.” Population and Development Review Feyisetan, Bamikale, and John B. Casterline. 35: 221–248. 2000. “Fertility Preferences and Contra- Heilig, Gerhard K. 1997. “Anthropogenic ceptive Change in Developing Countries.” Factors in Land-Use Change in China.” International Family Planning Perspectives Population and Development Review 23: 26: 100–109. 139–168. Freedman, Ronald. 1997. “Do Family Plan- Jejeebhoy, Shireen J. 1995. Women’s Education, ning Programs Affect Fertility Preferences? Autonomy, and Reproductive Behaviour: A Literature Review.” Studies in Family Experience From Developing Countries. Ox- Planning 28: 1–13. ford: Clarendon Press. Goldie, Sue, Steve Sweet, Natalie Carvalho, Joshi, Shareen, and T. Paul Schultz. 2007. Uma Chandra Mouli Natchu, and Del- “Family Planning as an Investment in Determinants and Consequences of High Fertility | A Synopsis of the Evidence 25 Development: Evaluation of a Program’s Lewis, J., C. Ronsmans, Alex Ezeh, and Simon Consequences in Matlab, Bangladesh,” Gregson. 2004. “The Population Impact Center Discussion Paper 951, Economic of HIV on Fertility in Sub-Saharan Af- Growth Center, Yale University. rica.” AIDS 18: S35–S43. Kelley, Allen C. 2001. “The Population De- Lightbourne, Robert E. 1985. “Individual bate in Historical Perspective: Revisionism Preferences and Fertility Behaviour.” In Revised.” In Population Matters: Demo- Reproductive Change in Developing Coun- graphic Change, Economic Growth, and tries: Insights From the World Fertility Poverty in the Developing World, ed. N. Survey, ed. J. Cleland and J. Hobcraft. Birdsall, A. C. Kelley, and S. W. Sinding. Oxford: Oxford University Press. Oxford: Oxford University Press. Lloyd, Cynthia B. 2005. Growing Up Global. Kelley, Allen C. 1996. “The Consequences of Washington, DC: National Academies Rapid Population Growth on Human Re- Press. source Development: The Case of Educa- Lloyd, Cynthia B. 1994. “Investing in the Next tion.” In The Impact of Population Growth Generation: The Implications of High Fer- on Well-Being in Developing Countries, ed. tility at the Level of the Family. In Popula- D. A. Ahlburg, A. C. Kelley, and K. O. tion and Development: Old Debates, New Mason. Berlin: Springer. Conclusions, ed. R. Cassen. Washington, Kelley, Allen C., and Robert M. Schmidt. DC: Overseas Development Council. 2005. “Evolution of Recent Economic-De- Luck, Gary W. 2007. “A Review of the Re- mographic Modeling: A Synthesis.” Journal lationship Between Human Population of Population Economics 18: 275–300. Density and Biodiversity.” Biological Re- Lam, David, and Leticia Marteleto. 2008. views 84: 607–645. “Stages of Demographic Transition From Mahy, Mary. 2003. “Childhood Mortality in a Child’s Perspective: Family Size, Cohort the Developing World: A Review of Evi- Size, and Children’s Resources.” Popula- dence From the Demographic and Health tion and Development Review 34: 225– Surveys,” DHS Comparative Report 4, 252. ORC Macro, Calverton, MD. Lam, David, and Leticia Marteleto. 2005. Mauldin, W. Parker. 1965. “Fertility Studies: “Small Families and Large Cohorts: The Knowledge, Attitudes, and Practice.” Impact of Demographic Transition on Studies in Family Planning 1: 1–10. Schooling in Brazil.” In The Changing Mensch, Barbara, Susheela Singh, and John Transition to Adulthood in Developing Casterline. 2005. “Trends in the Timing Countries: Selected Studies, ed. C. B. of First Marriage Among Men and Lloyd, J. R. Behrman, N. P. Stromquist, Women in the Developing World.” In and B. Cohen. Washington, DC: Na- The Changing Transitions to Adulthood tional Academies Press. in Developing Countries: Selected Studies, Lee, Ronald, and Andrew Mason. 2006. ed. C. B. Lloyd, J. R. Behrman, N. P. “What Is the Demographic Dividend?” Stromquist, and Barney Cohen. Wash- Finance and Development 43: 16–17. ington, DC: National Academies Press. 26 Nagdeve, D. A. 2007. “Population Growth ed. J. Phillips and J. Ross. Oxford: Clar- and Environmental Degradation in endon Press. India.” Asia Pacific Journal on Environment Robinson, Warren C., and John A. Ross. 2007. and Development 14: 41–63. The Global Family Planning Revolution: Pan, W., David Carr, Alisson Barbieri, Richard Three Decades of Population Policies and Bilsborrow, and C. Suchindran. 2007. Programs. Washington, DC: World Bank. “Forest Clearing in the Ecuadorian Am- Rosero-Bixby, Luis, and Alberto Palloni. 1998. azon: A Study of Patterns Over Space and “Population and Deforestation in Costa Time.” Population Research and Policy Re- Rica.” Population and Environment 20: view 26: 635–659. 149–178. Phillips, James F., M. Bazle Hossain, and Ross, John, John Stover, and D. Adelaja. Mary Arends-Kuenning. 1996. “The 2007. “Family Planning Programs in Long-Term Demographic Role of Com- 2004: New Assessments in a Changing munity-Based Family Planning in Rural Environment.” International Family Plan- Bangladesh.” Studies in Family Planning ning Perspectives 33:22–30. 27: 204–219. Rutstein, Shea O. 2008. “Further Evidence of Phillips, James F., Ruth Simmons, Michael the Effects of Preceding Birth Intervals on A. Koenig, and J. Chakraborty. 1988. Neonatal, Infant, and Under-Five-Years “Determinants of Reproductive Change Mortality and Nutritional Status in De- in a Traditional Society: Evidence From veloping Countries: Evidence From the Matlab, Bangladesh. Studies in Family Demographic and Health Surveys,” DHS Planning 19: 313–334. Working Paper 41, ORC Macro, Cal- Pimentel, David, and M. Pimentel. 2006. verton, MD. “Global Environmental Resources Versus Schultz, T. Paul. 2006. “Fertility and Income.” World Population Growth.” Ecological In Understanding Poverty, ed. A. V. Ba- Economics 59: 195–198. nerjee, R. Benabou, and D. Mookherjee. Pimentel, David, and D. Wen. 2004. “Agri- New York: Oxford University Press. culture in China: World Population, Food Timaeus, Ian M., and Tom A. Moultrie. 2008. and Environment.” In Dare To Dream: Vi- “On Postponement and Birth Intervals.” sion of 2050 Agriculture in China, ed. T. Population and Development Review 34: C. Tso. Beijing: China Agriculture Uni- 483–510. versity Press. Van De Walle, Etienne. 1992. “Fertility Tran- Pritchett, Lant H. 1994. “Desired Fertility and sition, Conscious Choice, and Numeracy.” the Impact of Population Policies.” Popula- Demography 29: 487–502. tion and Development Review 20:1–55. VanWey, Leah, K., Alvaro O. D’Antona, and Robinson, Warren C., and John Cleland. Eduardo S. Brondízio. 2007. “Household 1992. “The Influence of Contraceptive Demographic Change and Land Use/Land Costs on the Demand for Children.” In Cover Change in the Brazilian Amazon.” Family Planning Programmes and Fertility, Population and Environment 28: 163–185. Determinants and Consequences of High Fertility | A Synopsis of the Evidence 27 United Nations. 2009. World Population Pros- Development. New York: Oxford Univer- pects: The 2007 Revision. New York: Popu- sity Press. lation Division, Department of Economic York, Richard, Eugene A. Rosa, and Thomas and Social Affairs, United Nations. Dietz. 2003. “Footprints on the Earth: Westoff, Charles F. 2010. “Desired Number of The Environmental Consequences of Mo- Children: 2000–2008,” DHS Compara- dernity.” American Sociological Review 68: tive Report 25, ICF Macro, Calverton, 279–300. MD. Zaba, Basia, and Simon Gregson. 1998. “Mea- Wire, Thomas. 2009. Reducing Future Carbon suring the Impact of HIV on Fertility in Emissions by Investing in Family Plan- Africa.” AIDS 12 (Suppl.): S41–S50. ning: A Cost-Benefit Analysis. London: Op- Zaba, Basia, N. Terceira, P. Mason, and Simon timum Population Trust, London School Gregson. 2003. “The Contribution of of Economics, Operational Research. HIV To Fertility Decline in Rural Zim- World Bank. 1984. World Development Report babwe, 1985–2000.” Population Studies 1984: Population Change and Economic 57: 149–164. THE WORLD BANK 1818 H Street, N.W. Washington, DC 20433