Prospects for Fertility Decline in Africa

Abstract

In the early 1990s, the Committee on Population of the National Academy of Sciences organized an ambitious overview of the demography of sub-Saharan Africa, resulting in seven volumes containing a literature review, fresh empirical investigation, and assessment of likely future trends in phenomena ranging from reproductive behavior to health/mortality to adolescent transitions (Panel on the Population Dynamics of Sub-Saharan Africa 1993). As a contribution to this large project, a demographic analysis of levels and trends in fertility was conducted by Cohen (1993). He concluded that there was indisputable evidence of marked decline in fertility in a few countries (Botswana, Kenya, Zimbabwe) and, more importantly, strong suggestions of the onset of fertility decline in many others. His updated assessment five years later was more sweeping: “A fairly widespread decline in fertility is currently underway across Africa” (Cohen 1998, p. 1445). This was by no means an idiosyncratic reading of the empirical evidence available at the time. Cleland, Onuoha, and Timæus (1994) arrived at essentially the same conclusion, perceiving substantial decline in the same three countries singled out by Cohen and “modest decline” in four other countries, including Nigeria and Senegal in West Africa. Kirk and Pillet (1998) saw evidence of fertility decline in two-thirds of twenty-three African countries with recent demographic survey data. To be sure, both Cohen and Cleland, Onuoha, and Timæus offered important qualifiers. Cohen speculated that sub-Saharan African societies might diverge from the experience of other major regions in showing a resistance to aggregate fertility falling below four births per woman (echoing Lesthaeghe's (1989) prescient assessment), and Cleland, Onuoha, and Timæus pointed to evidence that raised doubts about whether circumstances in West African societies were conducive to substantial fertility decline anytime soon. However, when the Committee on Population of the National Academy of Sciences in 2000, as part of a broader assessment of global population dynamics (Bongaarts and Bulatao 2000), again considered prospects for fertility decline in sub-Saharan Africa, qualifiers were dutifully noted but the overall stance was a confident expectation that most of sub-Saharan Africa was on the cusp of a rapid and pervasive revolution in reproductive behavior resembling what had occurred in Asia and Latin America in the preceding four decades. This optimism about the likelihood of rapid fertility decline in sub-Saharan Africa, common among scholars in the 1990s as reflected in the successive National Academy of Sciences reports in 1993 and 2000, was dashed by the empirical evidence (mainly DHS estimates) that began to arrive in the late 1990s and steadily accumulated through the 2000s. Declines that had begun in the 1980s evidently stalled in the 1990s (Bongaarts 2008), the most notable case being Kenya (Westoff and Cross 2006), the country whose rapid decline through the 1980s into the early 1990s was regarded as the vanguard of many such declines expected to follow soon thereafter. In most countries in the region, fertility declines that commenced in the 1980s and 1990s were proceeding at a slow pace at best, and in a few countries in Middle and West Africa there was reason to doubt whether fertility decline was even underway (see Gerland, Biddlecom, and Kantorová, this volume). By the 2010s, demographers recognized that national fertility declines of the size and speed common in Asia and Latin America in the 1970s and 1980s—fertility declining from high to low levels in the span of three decades or less—were not yet a common feature of the demography of sub-Saharan Africa. The biennial population estimates and projections produced by the United Nations Population Division from 2002 through the most recent in 2015 have, as a general rule, each time revised upward the estimated fertility rates for the immediate and future quinquenniums for most sub-Saharan African countries, indicative of the continued failure of fertility declines in the region to accelerate. In response to these facts on the ground, during the past decade scholarship on fertility decline in sub-Saharan Africa has been preoccupied with the distinctiveness of its historical trajectory. Straightforward comparisons—such as those provided in this volume by Bongaarts and by Casterline and Agyei-Mensah—demonstrate that Africa stands apart from the other major regions. By now there is near-consensus that, from an historical comparative perspective, when it comes to fertility decline Africa is different. But there is nothing approaching consensus on the sources of this difference. To this end, past literature that stressed the distinctive nature of African reproductive regimes, itself expressive of deep-seated features of African cultural and social systems (Lorimer 1954; Lesthaeghe 1989; Caldwell and Caldwell 1987; Caldwell, Orubuloye, and Caldwell 1992; Ware 1994), has been revisited and applied to the abundant new demographic survey data on childbearing patterns in the region (Johnson-Hanks 2007; Timæus and Moultrie 2008). Mbacké (this volume) points to lingering effects of the colonial experience, which through various direct and indirect channels reinforced high fertility. He and others have attributed the slow pace of fertility decline to lags in socioeconomic development (Goujon, Lutz, and KC 2015; Bongaarts this volume) and/or poor governance and the resulting overwhelming uncertainty of day-to-day life in African societies, including the absence of reliable social and health services (Moultrie and Timæus 2015). The problématique of fertility decline in sub-Saharan Africa—the late onset of decline, the slow pace of decline, and debate about the most effective strategies for accelerating decline—was the motivation for the Committee on Population of the National Academy of Sciences to once again consider this topic by soliciting research papers that were presented at a workshop held in June 2015. The scientific goals of the workshop were to analyze recent fertility trends in sub-Saharan Africa and their implications, including the importance of country-specific factors and public policies, and to assess the prospects for fertility change in the region. This edited volume is derived from the papers for that workshop. Five contributions to this volume together provide a rich portrait of fertility transition in sub-Saharan African up to the present. Three of these (Gerland, Biddlecom, and Kantorová; Bongaarts; Eloundou-Enyegue, Giroux, and Tenikue) are analyses of fertility mainly, whereas two others (Casterline and Agyei-Mensah; Hertrich) focus on fundamental determinants of fertility (fertility desires and nuptiality, respectively). A further feature of two of the contributions (Gerland, Biddlecom, and Kantorová; Casterline and Agyei-Mensah) is assessment of future fertility decline scenarios via simulation exercises. Because fertility decline has been late and slow in sub-Saharan Africa, the region is virtually certain to experience substantial population growth during the twenty-first century. As Gerland, Biddlecom, and Kantorová report, the region began the present century with a total population of less than one billion. Under the 2015 set of projections by the United Nations Population Division, at the end of this century the total population of the region will be roughly 4 billion, that is, a fourfold increase during the century. This is the UN's medium variant, and for sub-Saharan Africa the uncertainty surrounding this projection is large, especially in the latter half of the twenty-first century. Indeed, Gerland, Biddlecom, and Kantorová note that when a country has high fertility, as is the case for most countries in sub-Saharan Africa at present, the uncertainty is especially large because its demographic future depends so heavily on the timing and pace of fertility decline. Global experience during the past five decades demonstrates how much countries can vary in the timing and pace of decline (the chapter contains a concise summary of Africa's experience to date). Gerland, Biddlecom, and Kantorová pose a hypothetical: suppose the pace of fertility declines in sub-Saharan Africa beginning at present were to mimic the pace of decline in a set of countries that, after a slow start (like Africa to this point), accelerated to a rapid pace that was sustained until fertility fell to moderate levels (TFR < 4.0). Twenty-one countries fall into this set, including South Africa. Under this scenario of faster fertility decline, the population of sub-Saharan Africa at the end of the twenty-first century would be 3.2 billion rather than 4 billion, that is, about one-quarter less population growth during the century. How plausible is this scenario? To this point there are few indications that fertility declines throughout the region are on the verge of marked acceleration. But the twenty-one countries that Gerland, Biddlecom, and Kantorová use for their simulation exercise are an eclectic set on most every dimension—culture, political history, economic development. Evidently, rapid fertility decline can occur in a wide range of societal settings—this argues against pessimism about the capacity of African societies to achieve rapid fertility decline. In short, the late and slow start to fertility decline notwithstanding, Gerland, Biddlecom, and Kantorová suggest that it is premature to conclude that ultimately African fertility declines will prove to be “exceptional” from an historical comparative perspective. Bongaarts conducts an empirical analysis of the association at the country level between fertility, mortality, and major social and economic factors (urbanization, GDP per capita, formal schooling), leading to two principal conclusions. First, at the onset of fertility declines in sub-Saharan Africa, mortality levels were higher and socioeconomic development levels were lower than was the case on average in Asia or Latin America. Second, at any given level of development, fertility is higher in sub-Saharan Africa than in the other regions. This second finding, which Bongaarts attributes to “traditional pronatalist social, economic, and cultural practices” (citing Caldwell, Orubuloye, and Caldwell 1992), applies especially to fertility desires and contraceptive use. Bongaarts shows that at any given level of conventional macro-level determinants of fertility—mortality, income per capita, urbanization, education—fertility is on average higher in sub-Saharan Africa, a roughly one-birth differential that he terms the “Africa effect.” Bongaarts submits evidence in support of two explanations for late and slow fertility decline in Africa that are often treated as in competition although they are by no means incompatible: (i) the region's slow progress in health and socioeconomic development and (ii) pronatalism intrinsic to African social and cultural systems. The two preceding contributions consider national-level data only, with no attention to sub-national differentiation. In the existing research literature there are also numerous country-specific analyses of fertility differentials, often relying on multiple regression analysis. As compared to these two analytical approaches, the chapter by Eloundou-Enyegue, Giroux, and Tenikue makes an innovative contribution in characterizing national fertility declines in terms of increase or decrease in the variation in fertility among sub-national groups. This is in effect a national-level analysis in which sub-national differences are the central concern. The extent to which the decline in fertility occurs unevenly according to societal strata (urban–rural residence, social class) is a topic of fundamental significance on both scientific and policy grounds, among other reasons because of its implications for differential gains from fertility decline. Eloundou-Enyegue, Giroux, and Tenikue ask whether there might be a “demographic Kuznets curve,” referring to Kuznets's (1955) hypothesis that income inequality first grows and then declines as economies industrialize. The authors translate the hypothesis from income inequality to fertility inequality, the latter according to educational attainment. At issue is the evolution of educational differentials in fertility over the course of fertility transition. The empirical results partially confirm the existence of a demographic Kuznets curve: in sub-Saharan Africa, educational differentials increase in the early stages of fertility decline. But the authors find little evidence of turn-around in the later stages of decline. Of course, to this point sub-Saharan Africa offers almost no late-transition experience, as the authors concede. This is a promising line of investigation for which surely the story is half-told at best. As noted above, Bongaarts's chapter shows that fertility desires in Africa are high, conditional on other structural factors (demographic or socioeconomic), as compared to fertility desires in other regions, reinforcing the same point in Bongaarts and Casterline (2013). While fertility desires in Africa are high, they are hardly unchanging, as Casterline and Agyei-Mensah (this volume) document through an analysis of trends in both the ideal number of children and parity-specific desires not to have any further children. Indeed, with the exception of a few Sahelian West African countries, recent survey data indicate sharp departures from the pronatalism evident in survey data prior to the mid-1990s. Taking a prospective view, the authors ask what mix of changes in fertility desires and rates of implementation of these desires would have the largest impact on fertility. This is a hypothetical, addressed via macro-simulation. What emerges from the simulation is that neither increased desire to limit family size nor increased effectiveness in avoiding unwanted births (i.e. lower unwanted fertility rates) alone would reduce fertility from present levels to a low level (i.e. three births per woman or less). Rather, for substantial fertility decline to occur, both changes must occur concurrently. It is an important conclusion for policy purposes: meaningful fertility decline in sub-Saharan Africa is dependent on both reduced demand for children and improved birth control. In all societies, levels and trends in fertility are in part a function of levels and trends in union formation and union stability, for the simple reason that sexual exposure is on average higher among women in formal unions. This is true despite the fact that union status is an imperfect proxy for sexual exposure in most societies. The imperfect correspondence between union status and the intensity of sexual exposure is especially the case in African societies, where premarital sexual activity is relatively common, the transition into marriage is a multi-stage process that often unfolds over several years, and polygynous unions are common. Accordingly, the earliest demographic analyses of the association between nuptiality and fertility in sub-Saharan Africa were cognizant of the complexity and diversity of marriage systems in this region, in turn reflective of the complexity and diversity of kinship systems (van de Walle 1968). Hertrich (this volume) is mindful of all this, but makes the case for the meaningfulness of assessing at the country level the association between fertility change and nuptiality change (specifically, timing of entry to first union). Her investigation draws on an impressive assemblage of marital status distributions in censuses and surveys from 1960 to the present. An intriguing picture emerges. An increase in the average age at first union has preceded the onset of fertility decline in most African societies, but the timing is such that the mechanical contribution of nuptiality change to fertility decline up to now appears limited. More consistent with the empirical patterns is a model in which both phenomena—nuptiality and fertility—have changed in response to the same social and economic forces. Hertrich notes, however, that the average age at first union remains young in many African societies, especially in Middle and West Africa, and therefore there is potential for nuptiality change to have impacts on fertility rates of the magnitude that characterized many fertility declines in Asia and the Arab Region. While a substantial increase in the average age at first union—or, more radically, a sharp growth in the fraction of women who never marry and largely forgo sexual relations—could produce meaningful reductions in fertility, there is little basis for expecting either change to occur soon. Either change, and especially high rates of never married, would represent a sharp departure from prevailing social relations and kinship systems in African societies. A far more promising route to low fertility is increased reliance on contraception, either to space births or to terminate childbearing. Accordingly, most discourse about reducing fertility rates in sub-Saharan Africa is devoted to strategies for raising contraceptive prevalence, and this is the focus of most investments and programmatic interventions whose underlying rationale is to reduce fertility. Corresponding to this high priority on contraception, a number of the papers at the June 2015 NAS workshop considered contraception from either a behavioral or a programmatic standpoint. Three chapters in this volume conduct multi-country analysis of contraception (with one of these considering induced abortion as well). Singh, Bankole, and Darroch evaluate the current impact of contraception and induced abortion on national fertility rates in sub-Saharan Africa. Via macro-simulations, they estimate that the number of births per 1000 women of reproductive age (i.e. the General Fertility Rate), which was 159 births per 1000 in 2014, would have been 31 births higher in the absence of modern contraception and almost 77 births higher in the absence of both contraception (modern and traditional methods) and induced abortion. That is, ceteris paribus, current birth control practices reduce fertility by one-third as against what would otherwise be the case. Further macro-simulations show the impact of various changes in contraceptive method mix; not surprisingly, the fertility impact of contraception would be even larger if current users of ineffective methods switched to highly effective methods (long-acting reversible methods in particular). Greater prevalence of contraception is not far-fetched when one considers, as Singh, Bankole, and Darroch report, that almost one-quarter of pregnancies are unintended (80 out of 220 pregnancies per 1000 women per annum, an estimate obtained by combining survey reports of unintended births with indirect estimates of induced abortions). The upshot is that there is considerable scope for fertility reduction even without changes in fertility desires—that is, simply by satisfying current unmet need for contraception. Demographic survey data—the DHS beginning in the 1980s, and more recently PMA2020—measure contraceptive behavior in some detail along with important concomitant behaviors (sexual activity, postpartum behaviors). Tsui, Brown, and Li use this deep reservoir of information to examine multiple facets of contraception. Trends in contraception are presented separately for married and sexually active unmarried women. For selected countries, method-specific trends are shown. Discontinuation rates are calculated where the data permit (twenty countries), and these are exceptionally high (over 30 percent within twelve months in eleven of the twenty countries). Finally, adoption of contraception postpartum is examined in six purposively selected countries. This analysis is preceded by a concise recounting of the stances of African governments toward family planning during the past two decades, a history described in more detail by May (this volume) and also touched on by Mbacké (this volume). Tsui, Brown, and Li conclude with reflections on some salient programmatic issues, several of which are contentious: the role of external assistance, the potential contribution of community-based distribution of information and services, and competition for resources with programs targeting other health outcomes (including HIV/AIDS). The chapter offers a rich and variegated portrait of current patterns of contraceptive use, as well as insightful reflection on existing and emerging programmatic challenges. The contributions by Singh, Bankole, and Darroch and by Tsui, Brown, and Li point to the prevalence of methods of contraception that are assumed to be less effective in preventing pregnancy. Rossier and Corker offer a comprehensive analysis of this phenomenon for twenty-nine countries, using the most recent DHS survey in each country. Two methods are singled out: withdrawal and periodic abstinence. Rossier and Corker reluctantly label these “traditional contraception”—reluctantly because it is unclear whether these were prominent features of past reproductive regimes in Africa and, moreover, there is sound evidence of the growing popularity of these methods among educated urban women, that is, the most modern sectors of these societies. The survey data indicate that in most countries a small minority of users are using a traditional method, with the exception of a few countries in Middle Africa, specifically Republic of the Congo and Democratic Republic of the Congo (in both countries rates of traditional use essentially match rates of modern use). Rossier and Corker posit two reasons why women may choose one of these traditional methods rather than a modern method: high costs of obtaining modern methods (geographic access, financial cost) or innate appeal of traditional methods (religious considerations, health concerns). Regression analysis of the survey data lends support to both hypotheses, and it seems likely that both reasons are operative for one or another group of women. This chapter stands as the most thorough examination to date of the practice of traditional contraception in sub-Saharan Africa. The contributions by Singh, Bankole, and Darroch; Tsui, Brown, and Li; and Rossier and Corker are all multi-country analyses that span the entire continent, and the first two also contain historical depth. All three chapters offer summary evaluations that inevitably cannot represent the diversity of contraceptive patterns in the region. The authors are fully aware of this limitation—indeed a major theme of Tsui, Brown, and Li is the heterogeneity of society and demography in sub-Saharan Africa. As a sharp contrast to these multi-country analyses relying on national survey data, the data analyzed in the chapter by Beguy et al. is confined to a few urban slums in Nairobi. In these urban settlements, the African Population and Health Research Center (APHRC) has been collecting data on health attitudes and behaviors, including reproductive attitudes and behaviors, since the late 1990s. Comparison of surveys conducted in 2000 and 2012 reveals a substantial increase in contraceptive prevalence, from 34 percent to 53 percent of married women, a nineteen-point increase in little more than one decade. Beguy et al. ask how this sharp increase can be explained, and more specifically whether it can be attributed to a shift in characteristics of the population toward a demographic and socioeconomic profile more conducive to contraceptive use or, instead, to an increase in contraception within demographic and socioeconomic strata. This leads the authors to a regression decomposition that leaves no doubt that the increased contraceptive prevalence was the result of increased rates of use within population strata. Particularly striking is the large contribution to the overall increase in prevalence of higher rates of use among women who express a desire to postpone the next child or not have another child. Increase in contraceptive prevalence among these women can be regarded as satisfaction of unmet need for contraception, and Beguy et al. show how fundamental this was to the overall increase in contraceptive prevalence. How, then, to explain the satisfaction of unmet need? While acknowledging that they have not conducted a rigorous evaluation, the authors propose that the results support a large net contribution of the Kenya Urban Reproductive Health Initiative (Tupange). Tupange, designed to provide women in poor urban settlements with family planning counseling and services, was active in these communities starting in 2009. What fraction of the nineteen-point increase in contraceptive prevalence between 2000 and 2012 occurred after the initiation of Tupange is not known. The argument that fertility decline is beneficial for economic growth received its first rigorous articulation nearly six decades ago by Coale and Hoover (Coale and Hoover 1958). Only during the past two decades, however, has the more specific effect of fertility decline on the economy termed the “demographic dividend” been fully appreciated. One consequence of fertility decline is a transitional period during which the age structure of the population is weighted toward the prime working ages (e.g. 20–64), thereby giving a boost to economic growth ceteris paribus (Bloom, Canning, and Sevilla 2003). The more rapid the fertility decline, the more intense the shift in the age structure toward the working ages (but also the shorter the transitional period during which this age-structure effect is operative). The demographic dividend is now recognized as one of the principal returns from fertility decline; indeed, according to some analyses it was a major driver of economic growth in East Asia beginning in the 1970s (Mason 2001). Rather suddenly during the past decade, the opportunity to realize a demographic dividend became the principal rationale for policies and programs to accelerate fertility decline in sub-Saharan Africa, an evolution in population policy discourse described by both May and Mbacké. Two chapters consider the prospects that fertility decline will yield a demographic dividend in sub-Saharan Africa. Karra, Canning, and Wilde specify a large number of channels through which fertility decline can affect economic growth and represent each channel in a macro-simulation model. The model is applied to Nigeria under scenarios of slow and rapid fertility decline in order to ascertain the potential economic gain from more rapid decline. This is an extension of earlier work (Ashraf, Weil, and Wilde 2013), with the important addition of feedback effects of fertility decline on human capital formation (health, schooling). The macro-simulation results indicate a substantial economic impact of more rapid fertility decline, on the order of a doubling of per capita income in the period 2010–2070. Rapid fertility decline also has positive ramifications for related outcomes including health, schooling, and female labor force participation. The payoffs from more rapid decline are strikingly large and perhaps at the outer boundaries of what is realistic: the difference between the low and high fertility scenarios is fixed at one birth per woman from 2020 onward. Cleland and Machiyama offer a far more guarded appraisal of the prospects that African societies will realize a demographic dividend. The formal economic logic underlying the demographic dividend argument is not questioned; instead Cleland and Machiyama stress that the size of the dividend is conditional on non-demographic factors that on balance are highly unfavorable. These include inadequate public-sector resources for human capital investment (especially schooling), the slow pace of improvements in agricultural productivity, and, especially, the dim outlook for absorption into the labor force of the large cohorts reaching adulthood during the next few decades. The argument is, in effect, that Africa approaches the multi-decade period when age structures may be conducive to economic growth with many factors aligned against the demographic dividend. By cataloguing the obstacles to realizing the dividend, Cleland and Machiyama correct the temptation to regard the dividend as a virtually guaranteed return from fertility decline. All of the chapters already discussed in this Introduction offer insights that can inform population policies and programs. Several offer explicit assessment of policy and program factors, most notably Beguy et al.’s chapter on contraceptive change in Nairobi slums. Three further chapters in this volume are devoted largely or entirely to population policies and programs. Askew, Maggwa, and Obare compare the joint evolution of fertility and population policy in Ghana and Kenya. The choice is strategic: one country each from East and West Africa, each country a fertility-decline forerunner in its region, and each with a succession of well-executed national demographic surveys conducted every five years beginning in the late 1980s. Few African countries have such well-documented fertility declines as those in Ghana and Kenya, and few have drawn as much attention from scholars and policymakers alike. The large amount of attention can be attributed to the striking historical trajectories fertility has followed in both countries: rapid decline commencing in the 1980s, regarded throughout the 1990s as exemplars of declines that might occur in other countries; but then an abrupt stalling of decline in the late 1990s, sobering lessons that achieving fertility decline in sub-Saharan Africa might be more difficult than had been assumed. The stall was more pronounced in Kenya, where the total fertility rate (TFR) declined by almost 3.5 births (8.1 to 4.7) in the fifteen-year period from the late 1970s to the early 1990s, an astonishingly rapid decline. By comparison, Gha