National Academies Press: OpenBook

Factors Affecting Contraceptive Use in Sub-Saharan Africa (1993)

Chapter: 3 The Socioeconomic Context

« Previous: 2 Levels and Trends in Contraceptive Use
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 52
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 53
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 54
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 55
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 56
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 57
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 58
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 59
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 60
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 61
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 62
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 63
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 64
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 65
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 66
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 67
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 68
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 69
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 70
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 71
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 72
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 73
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 74
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 75
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 76
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 77
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 78
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 79
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 80
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 81
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 82
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 83
Suggested Citation:"3 The Socioeconomic Context." National Research Council. 1993. Factors Affecting Contraceptive Use in Sub-Saharan Africa. Washington, DC: The National Academies Press. doi: 10.17226/2209.
×
Page 84

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

The Socioeconomic Context Contraceptive use is the expression of individual desires to space or to limit births. Individual demands for birth spacing and limitation are them- selves shaped by the surrounding social, economic, and policy environment. If our aim is to understand contraceptive use in sub-Saharan Africa, we must look first to the determinants of the demand for children. This chapter opens the investigation with a consideration of the linkages between eco- nomic development and fertility. Our discussion will emphasize levels of income per capita, child mortality, educational attainment among adults, and the costs and benefits of child schooling. The overriding issue here and in Chapter 4 is whether, and under what conditions, sub-Saharan Africa is likely to join in the process of fertility decline in progress elsewhere. Recent demographic literature on Africa presents no consensus about the prospects for change; indeed, the two views that dominate the literature are strikingly opposed. In one, fertility decline is seen as the natural concomitant of economic development; the demo- graphic-economic linkages are thought to apply in sub-Saharan Africa much as they do elsewhere. Continued high fertility is therefore explained in terms of the factors that continue to inhibit economic development more generally. In the other perspective, however, sub-Saharan Africa is re- garded as being uniquely resistant to fertility decline. This resistance is said to be rooted in fundamental and very nearly immutable features of social organization, not easily swept away in the course of development. The two competing perspectives on change are perhaps best exemplified in 52

THE SOCIOECONOMIC CONTEXT 53 two publications-a report by the World Bank (1986) and an article by Caldwell and Caldwell (1987~. The World Bank report locates the reasons for high fertility in Africa in a set of conventional indices of economic development: low incomes per capita, high infant and child mortality rates; low levels of adult literacy; high proportions of the work force in agriculture; low percentages of the population living in urban areas; and continuing difficulties across the con- tinent in access to education, health, and family planning services. In none of these dimensions has much of Africa advanced to a threshold sufficient to induce fertility decline, according to the report. Thus, there should be nothing very surprising about high African fertility and low contraceptive use at present, but one should not expect these conditions to persist as the development indicators begin to improve. The alternative view, exemplified by the article by Caldwell and Caldwell, draws attention to the unique features of African fertility decision making. African societies are viewed as being wholly distinctive in respect to social organization. They are said to differ from the remainder of the developing world not merely in terms of the conventional categories of development, but more fundamentally in the nature of spouse relationships, household structure, kinship, culture, and religion. These elements have combined to provide a powerful and coherent rationale for high fertility, one that is without real counterpart elsewhere. Moreover, according to a later article by Caldwell and Caldwell (1988), the high-fertility rationale remains intact even in modern nation-states. The modern African state exerts only a su- perficial influence on social organization, due to difficulties in formulating population policies at the national level and to continuing deficiencies in local service delivery. The state commands very few of the tools needed to dismantle the high-fertility rationale. These two lines of argument, seemingly so divergent, in fact have a common point of origin in the historical record. The economic develop- ment of sub-Saharan Africa has been profoundly affected by its distinctive material conditions: principally, the relative abundance of land in compari- son to labor; soil that in much of the region yields meager dividends to intensified cultivation; and the very considerable uncertainties surrounding the standard of living and life itself, owing to vagaries of climate and the prevalence of disease (Locoh, 1991~. These material conditions are ex- pressed in economic relations and kinship structure, and have shaped as- pects of both culture and religion. They explain much of the high value placed on labor and fertility, as well as the importance accorded to net- works of mutual support, whether established through children, kin, or so- cial relations. They may also help to explain the comparative recency in Africa of territorially based bureaucratic structures, which provide the foun- dation for nation-states. Thus, in reconciling the views of the World Bank

54 FA CTORS AFFECTING CONTRA CEPTI VE USE (1986) and Caldwell and Caldwell (1987), what is at issue is not so much the historical legacy, which has affected both the level of economic devel- opment and social structure, as its relevance to African economies entering the decade of the 1990s. This chapter explores the view that high fertility in Africa can be ex- pla~ned principally by levels of development. (Chapter 4 will focus on the other view.) In an effort to weigh the evidence, we shall first review the major theoretical linkages between socioeconomic variables and fertility. These linkages are illustrated with cross-national data from developing countries including some in sub-Saharan Africa. We also summarize a body of evi- dence from household-level research on the socioeconomic determinants of fertility, with the aim of assessing whether fertility levels are indeed less sensitive to economic differentials in Africa than elsewhere. We then con- sider in more detail the evidence on the relation between fertility and child mortality and the prospects for increased survival, and the evidence on the role of schooling costs and benefits in the demand for children. Finally, we speculate as to the longer-run demographic consequences implicit in the current era of economic stagnation and structural adjustment. SOCIOECONOMIC DIFFERENTIALS OF FERTILITY National-Level Relationships In this section we briefly review the theoretical arguments linking fer- tility levels to a set of socioeconomic determinants: income per capita, child mortality, educational attainment among adults, and the costs and benefits of child schooling. Where appropriate, we illustrate the theoretical arguments with data drawn from recent cross-national samples of develop- ing countries. In so doing, we take per capita income as a convenient index of the level of economic development. The relationship between income and fertility has been the subject of much study. Evidence from many regions of the world that have experi- enced fertility declines shows that income growth is associated with lower fertility in the long run (World Bank, 1984~. However, economic theory suggests that fertility may rise with increases in income, because additional resources allow the family to afford more children. In the short term, this iThe data are taken from the sample of countries covered in the 1991 World Development Indicators (World Bank, 1991) and limited to countries with an income level per capita of $3,000 or less in 1989. The income cutoff is used because no sub-Saharan country exceeded this level in 1989. Gabon's per capita income in that year was estimated as $2,960; South Africa's was $2,470; and the next highest level of income per capita, $1,600, was in Botswana (estimates in 1989 U.S. dollars.)

THE SOCIOECONOMIC CONTEXT o Sub-Saharan Africa ~ Other LOCs an 90 ao 70 Q) 60 JJ ~50 a L) ~40 ~_ a, 30~ 20 10 O - + Zimbabwe ~ + a+ ~ + / ~ + + · it+ ~ Kenya + / ~ Botswana 4. 55 0 500 1000 1500 2000 2500 3000 GNP per Capita: 1939 $ FIGURE 3-1 Contraceptive prevalence versus per capita GNP. SOURCE: World Bank (1991 ) and Demographic and Health Survey data tapes for sub-Saharan Africa. positive relationship has been observed many times. The income-fertility relationship is complicated, mainly because changes in income are related to other aspects of economic development, such as educational opportuni- ties, the participation of women in the work force, accessibility of consumer goods, and the value of time. Thus, the effects of income on fertility are both direct and indirect, and income may exhibit both negative and positive associations with fertility (Mueller and Short, 1983~. However, it is gener- ally agreed that in the long term, income is negatively associated with fertility, because of the relationship of changes in income with changes in other factors that reduce the demand for children (World Bank, 19841. (See the section in this chapter on economic stagnation and adjustment for fur- ther discussion.) To set the stage for the theoretical discussion, Figure 3-1 shows the positive relationship of contraceptive prevalence (all methods) to gross na- tional product (GNP) per capita for more than 35 developing countries. The data include a small number of sub-Saharan African (SSA) countries with information on contraceptive prevalence as of 1990. In ascending order of incomes, the sub-Saharan group includes Burundi (9 percent prevalence),

56 FACTORS AFFECTING CONTRACEPTIVE USE Uganda (5), Mali (5), Kenya (27), Ghana (13), Togo (12), Senegal (12), Zimbabwe (43), and Botswana (33).2 Sub-Saharan countries taken as a whole clearly display lower levels of contraceptive prevalence than would be expected given their levels of development.3 The African countries compose the majority of countries in the sample with low incomes and prevalence rates. Kenya, Zimbabwe, and Botswana are outliers by compari- son with their sub-Saharan counterparts. Kenya and Zimbabwe have the prevalence rates expected given their levels of income, and Botswana, with the highest level of income of the sub-Saharan countries shown here, has a lower than expected prevalence rate (although still higher than all other sub-Saharan countnes, except Zimbabwe). Estimates of total fertility rates4 are available for a much larger sample of sub-Saharan countries. In what follows, we generally rely on fertility data to provide an indication of the level of demand for births or for surviv- ing children in the populations in question.5 Figure 3-2 depicts the relation- ship between total fertility rates (TFRs) and income per capita in 1989. As the figure shows, in comparison to developing countries outside the region, those in sub-Saharan Africa display higher TFRs than their income levels alone would suggest. 2Income-level data are not available for Liberia. 3In Figure 3-1 and several figures that follow, a regression line is used to help guide the eye through the data. In each case the regression was calculated by taking the variable displayed on the vertical axis as the dependent variable and that on the horizontal as the independent variable. Regressions were estimated by using the full sample comprised of all developing countries with per capita incomes of less than $3,000. Linear (y = a + bX) and log-linear (y = a + blnX) specifications were examined in each case; the better-fitting line is shown in the a. figures. 4The total fertility rate (TFR) is a measure of the average number of children that would be born to a woman who survives through her reproductive years. It is usually calculated on a period-specific or calendar-year basis by using the age-specific fertility rates in effect in the period in question. SFertility may vary for reasons having little to do with the demand for births or for surviving children. In parts of Central Africa, for example, primary and secondary sterility places important constraints on fertility, and in this case, levels of fertility are not easily interpreted in terms of the underlying demands. In other parts of Africa, observed fertility may be higher than desired because there is not perfect control over actual fertility (due to lack of access to efficient contraception, among other reasons). Despite these difficulties in interpretation, we think that fertility rates provide the best approximation of the demand for births. It is not obvious that the important postpartum proximate determinants, breastfeeding and postpartum abstinence, necessarily cause any difficulties in the interpretation placed on fertili- ty levels. To the degree that these behaviors reflect a demand for surviving children that is, to the degree that birth spacing is understood to be related to child survival in the populations in question the level of fertility is a good indicator of the underlying demand. If for some reason the connection between spacing and survival is not recognized, however, the interpreta- tion of fertility levels in terms of demand is more problematic.

THE SOCIOECONOMIC CONTEXT o Sub-Saharan Africa + Other LCDs 91 B cr 6 ~ L a' LL 57 a oo °`S, onto Q'O CX: of Oo :+ 0 4. -5_ _~, 4- + +' + ~ '__ 3 2 ~ 1 - O ~ - o 14. o 4- I I I I I T 0 500 1000 1500 2000 2500 3000 GNP per Capita: 1989 $ FIGURE 3-2 Total fertility rates by per capita GNP. SOURCE: World Bank (1991~. Figure 3-3 presents fertility rates and income among only the sub-Sa- haran countries with per capita incomes of less than $1,700 (excluding Gabon and South Africa). Within this group, we should distinguish two sets of countries: the five most populous countries of the region excluding South Africa, that is, Nigeria, Ethiopia, Zaire, Sudan, and Tanzania; and the three countnes, Kenya, Zimbabwe, and Botswana, thought to be the fore- runners of an African fertility transition.6 The set of largest countries ac- counts for some 51 percent of the total sub-Saharan population (as of 1989~; Nigeria alone accounts for 23 percent. Simply by virtue of their size, these 6South Afnca's 1989 population, some 35 million, would place it third in terms of size, on a par with Zaire. We have excluded South Africa for consideration here because (1) its per capita income ($2,470 in 1989) is much higher than the norm for the region; and (2) the considerable demographic and economic heterogeneity of its population leads to difficulties of interpretation . Sudan does not appear in some of our figures because the World Bank data sets that we used did not supply an estimate of its income per capita in 1989. The total fertility rate for Sudan was estimated to be 6.4 as of 1988, and an income estimate per capita for that year was $480 (World Bank, l990b).

58 . . a) 9 _ Q _ CJ a, CD 7 ~ o, 6 ~ 5 ~ .,. ._ .,, 4 a) 3 ~ 0 2 ~ 4 ~ O ~ FACTORS AFFECTING CONTRACEPTIVE USE o Ethiopia o o° Kenya 0 0 o. °lanz. /. ~ / Zaire ° 0 Nigeria o o ~ Zimbabwe o . Botswana 1 1 1 ' 0 250 500 750 GNP per Capita: 1989 $ 1000 1250 1500 FIGURE 3-3 Total fertility rates in sub-Saharan Africa (countries with GNP per capita less than $1,7001. SOURCE: World Bank (1991~. countries must dominate the demographic future of the region. Kenya's population is very nearly the size of Tanzania's. By contrast, Zimbabwe and Botswana collectively comprise only 7 percent of the region's popula- tion. Thus, the importance of the higher use countries rests not so much on their contribution to regional totals, as on the lessons they may embody regarding the determinants of change. Of the countries with higher prevalence, Figure 3-3 shows that both Botswana and Zimbabwe belong in the upper strata of sub-Saharan coun- tries in respect to income per capita. Kenya, by contrast, lies in the middle rank in terms of income and has a TFR comparable to that of Nigeria or Tanzania. Evidently, then, the interest that surrounds the case of Kenya has less to do with the present level of fertility than with its current contracep- tive prevalence and the fact of a recent downward change in fertility (Locoh, 19911. Mortality The linkages between fertility and mortality consist partly of what has been termed a "physiological effect," whereby a child's death truncates

THE SOCIOECONOMIC CONTEXT 59 breastfeeding, leaving the woman reexposed to the risk of conception. We expect this physiological effect to be more powerful in societies with longer periods of full breastfeeding, where the contraceptive effect of breastfeeding is itself stronger, and also where postpartum abstinence is linked to breastfeeding. Lloyd and Ivanov (1988), in reviewing Preston's (1975, 1978) work, sug- gest that in a population with long durations of breastfeeding and postpar- tum abstinence, a reduction of one infant death would be associated with some 0.25 to 0.30 fewer births over a typical reproductive lifetime. Thus, declines in mortality could be expected to produce declines in fertility, if on something less than a one-to-one basis.7 The behavioral linkages between mortality and fertility consist of what are termed insurance and replacement effects, the former having to do with the influence of perceived or anticipated mortality risks on decisions re- garding fertility; the latter, with the household's response to an actual child death. The insurance effect is thought to be evident in coefficients that relate community- or area-level indicators of mortality to individual-level information on fertility or contraceptive use. Very little is known about the relationship between the risks of mortal- ity facing adults and the decisions that adults make regarding fertility. We raise the point here because it will likely become an issue in the regions of Africa suffering from a high prevalence of acquired immune deficiency syndrome (AIDS; see the research agenda laid out in Grabble, 19921. A pronatalist reaction to the threat of AIDS is certainly plausible, especially as infant and child mortality due to mother-child transmission of human immunodeficiency virus (HIV) will increase in these regions. However, the social upheavals associated with an epidemic of the scale anticipated for Central and parts of southern Africa may call into question many previously established behavioral relationships. We see little basis for the extrapola- tion of existing demographic theory on mortality-fertility linkages to such unprecedented circumstances. Figure 3-4 shows the relationship between total fertility rates and child mortality rates, with the sub-Saharan countries highlighted. It can be seen that African countries conform to the statistical norm wherein high levels of 7If the decline in fertility is less than one to one, we expect the net reproduction rate (the average number of daughters born per woman in a cohort given certain mortality rates) to . . ~ . . . increase as lntant survivors. alp improves. 8Although we emphasize only the associations through which mortality affects fertility, it is important to recognize that causation can run in the other direction as well. Higher fertility may imply shorter intervals between births and greater proportions of births occurring in the portions of the maternal age span (under age 20 or over age 35) when children face higher mortality risks (Working Group on the Health Consequences of Contraceptive Use and Con- trolled Fertility, 1989).

60 a, CD ., FACTORS AFFECTING CONTRACEPTIVE USE 0 Sub-Saharan Africa 9 _ B 7 - 6- ~ 5 - +. j .,' 4- + c' 3_ >/- 2 - ~ ~ o O - ~ Other LCDs o O O 0 0 o o °° ° loo ° _&~° ~o to ~ O+ 0 + + ~ + o o 1 1 1 O 50 100 150 200 250 Child Mortality Rate: 1989 FIGURE 3-4 Total fertility rates versus child mortality rates. SOURCE: World Bank (l991~. fertility are associated with high levels of mortality. Similar correlations are evident across regions within individual African countries. For ex- ample, Montgomery and Kouame (1992) find a significant influence of area- level child survival rates on cumulative fertility in Cote d'Ivoire. Kelley and Nobbe (1990) report a high negative correlation between infant mortal- ity and contraceptive use in a set of Kenyan provinces. Educational Attainment of Adults Few relationships have been so exhaustively studied as that between a woman's educational attainment and her fertility. Yet the basis of the relationship remains far from understood (Cochrane, 1979, 1983), particu- larly in settings such as those characteristic of Africa. Economists have drawn attention to the link between education and labor market earnings. The higher a woman's educational attainment is, the greater is her potential wage. If time spent in work and time in child care are mutually exclusive, then wage measures one of the principal opportu- nity costs of childrearing. It follows that the higher is the price of time, the lower should be fertility.

THE SOCIOECONOMIC CONTEXT 100 90 ~P Be 70 60 50 40 - 30 20 10 O - 61 o Zimbabwe 0 ~ o Zaire O O Nigeria 0 0 o 0 0 0 0 - o . 0 0 Kenya Botswana . o o o o 1 1 1 1 1 1 1 . 0 250 500 750 1000 1250 1500 GNP per Capita: l98g $ FIGURE 3-5 Female literacy rates in sub-Saharan Africa (countries with GNP per capita less than $1,700~. SOURCE: World Bank (1991~. Figure 3-5 shows that higher levels of female literacy are associated with lower total fertility rates in sub-Saharan Africa, as elsewhere. How- ever, it is not at all clear that the chain of reasoning spelled out above can be applied to African economies. For most African women, work need not conflict directly with child care. Where a conflict exists, it may be resolved through the employment of low-cost substitutes for the mother's time in child care, such as care provided by relatives. The link to the opportunity cost argument is therefore likely to be weak. In an influential paper, Caldwell (1982) envisions a larger role for edu- cation than that considered in the simple economic model. In his view, education serves as a vehicle for the adoption of Western ideas regarding the family. It may encourage a more child-centered view of one's parental responsibilities. Education may alter the definition of what constitutes ac- ceptable child care, giving greater weight to the time spent by the mother with her child, compared to time given by mother substitutes. Education may also affect the distribution of authority within the household, so that educated women gain a measure of autonomy vis-a-vis their husbands, and couples vis-a-vis their elders. Little is known in any quantitative sense about these issues (see Chapter 4 on the subject of emotional nucleation),

62 FACTORS AFFECTING CONTRACEPTIVE USE but we suspect that adult educational attainment, and particularly schooling for women, may be a precondition for a decline in the demand for children. Schooling of Children: The Quantity-Quality Trade-Off In any number of societies that have gone on to experience fertility decline (for Thailand, see Knodel et al., 1987, 1990), changes in the per- ceived benefits and costs of child schooling have played a key role in the transition. The impetus for fertility change originates in the economic re- turns associated with schooling. In the course of modernization, an economy begins to display significant differentials in earnings by schooling level. Parents then come to regard schooling as an avenue to a better life for their children and as a human capital investment that may, over the long term, pay dividends to the parents themselves. Yet education is costly in terms of both direct costs and opportunity costs of forgone child labor; it generally remains too costly for parents to give each child the desired schooling and continue to bear the customary number of children. Some element of household expenditures must give way, and typically fertility falls as household in- vestments in education per child increase. The increase in child schooling, coupled with the decline in fertility, has been termed a quantity-quality demographic transition (Becker and Lewis, 1973; Willis, 1973; Caldwell, 1982~. Figure 3-6 gives an indication of the quantity-quality transition in de- veloping countries. Total fertility rates for 1989 are arrayed on the vertical axis the quantity dimension. The horizontal axis shows the quality dimen- sion, as expressed in 1988 primary school enrollment ratios. Viewed differ- ently, the vertical axis represents the principal determinant of the future rate of labor force growth, that is, fertility. The horizontal axis indicates the levels of human capital with which future labor market entrants will be equipped (albeit with a shorter time lag). A quantity-quality demographic transition is therefore fundamental to economic development, in that it im- plies a reduction in future labor force growth and an increase in human capital per worker. The trade-off should be understood to represent a systematic associa- tion between two variables, fertility and child schooling, both of which are determined by the preferences, costs, and financial constraints of the house- hold. Higher fertility does not in itself cause lower school enrollment, nor does greater enrollment cause lower fertility. Rather, each of these vari- ables taken individually reflects the full set of opportunities and constraints facing households. Nothing requires that the relationship between fertility and child schooling have a negative slope. Indeed, Figure 3-6 shows that for sub-Saharan countries, the negative relationship between fertility and primary schooling is far weaker, if it exists at all, than is the case among

THE SOCIOECONOMIC CONTEXT o Sub-Saharan Africa a, co _4 I_ ~2 - o O - 63 ~ Other LCOs oO + ~ + - o ~ O D ~ - o Q B° 0 :0 r . 1 1 1 80 i00 120 140 20 40 60 Gross Primary Enrollment Ratio: 198B % FIGURE 3-6 Total fertility rates versus primary school enrollment ratios. SOURCE: World Bank (1991~. Other developing countries. The reason for this difference in slopes must be sought in the socioeconomic background factors that jointly influence fertil- ity and schooling per child. Among such factors, a good deal of attention has been directed to policy- induced changes in the costs of schooling (discussed below). In some accounts, the magnitude of private schooling costs (tuition, fees, capital levies, etc.) is said to be integral to the quantity-quality transition. These cost increases could conceivably induce a quantity-quality transition, such that an African country previously in position A of Figure 3-6 would move to position B of higher enrollment and lower fertility, but this response is unlikely. More plausible responses to an increase in the price of education are represented in an enrollment reduction accompanied by little change in fertility (a movement from position A to position C), or perhaps a decrease in both fertility and child schooling (movement from A to D).9 These various responses to changing school costs have profoundly different impli- cations for future demographic and economic development. In short, it is 90ne would certainly expect an increase in the price of schooling to have a direct negative effect on enrollment. In addition, if quantity and quality are complements, an increase in the price would cause fertility to decline; otherwise it would increase.

64 FACTORS AFFECTING CONTRACEPTIVE USE far from clear that changes in the costs of schooling alone can engender a quantity-quality transition. The origins of the transition in Africa are likely to be found not simply in educational costs, but rather in a changing configuration of benefits and costs. The private benefitsi° accruing from an extra year of schooling consist of the extra earnings that year is expected to generate over a work- ing lifetime, suitably discounted and adjusted for unemployment and earn- in~s variability (World Bank, 1988~. If parents perceive little improvement in their children's earning potential with the level of schooling, they may be quite sensitive to marginal changes in the price of schooling. Faced with an increase in tuition or fees, they may simply respond by reducing human capital investments, as indicated in movement from point A to point C in Figure 3-6. If the earnings gradient is steep, on the other hand, parents may choose to respond to cost changes in part by making sacrifices in other dimensions of behavior, including that of fertility, as shown in the A to D response. However, one would not expect to observe a quantity-quality transition (A to B) unless the benefits of schooling themselves improve. In a sense, therefore, the benefits of schooling are critical. They serve as a fulcrum, giving the costs of schooling effective leverage against fertil- ity. If one wishes to make predictions about fertility change, the costs of educational investments cannot be considered in isolation from the antici- pated rectums. ---O ~ The Combined Relation of Per Capita Income, Mortality, Education, and Fertility Above we considered the bivariate relationships between fertility and per capita income, child mortality, and education, using national-level data. Here we investigate the relations in a multivariate regression framework. Two points are at issue in framing such a regression. First, is the level of 10We emphasize here the private benefits of schooling, recognizing the distinction between private and social returns, because the former are more likely to influence the investment decisions of parents. (Additional social benefits consist, for instance, in the well-documented effects of adult schooling on fertility and child mortality. Private benefits would also be calculated net of taxes on earnings. Social benefits consider pretax effects on productivity; a kite.`, issue in their calculation is the extent to which earnings differentials associated with education represent true differentials in productivity, as opposed to credentialism effects.) We do not discuss who in the African household makes such decisions regarding investments in child schooling, nor are we concerned in this chapter with mechanisms for spreading the costs of schooling among kin (see Chapter 4). 1lNote that schooling costs exert their leverage in large part because of the liquidity con- straints facing African households. The possibility of borrowing to finance educational invest- ments in children is available to very few households. Chapter 4 discusses the role of transfers among kin, including child fostering, in this regard. _,

THE SOCIOECONOMIC CONTEXT TABLE 3-1 Regression Analysis of 1989 Total Fertility Rate Parameter Model 1 Model 2 Constant (~-statistic) Income per capita, 1965 (log) Infant mortality rate, 1965 Primary enrollment ratio, 1965 Secondary enrollment ratio, 1965 Sub-Saharan Africa dummy variable SSA · income per capita SSA · infant mortality rate SSA · primary enrollment ratio SSA · secondary enrollment ratio it-squared F- statistic 4.569 (3.61) -0.85 (0.47) .011 (2.78) -.008 (1.52) -.025 (1~99) 1.310 (4.98) 3.206 (1.90) .206 (0.89) .015 (3.04) -.025 (2.63) -.006 (0.44) 5.037 (1.97) -.553 (1.39) -.011 (1.34) .021 (1.74) -.009 (0.13) .77.80 41.9825.75 NOTE: Number of observations = 68. 65 fertility in sub-Saharan Africa indeed higher than would be expected given its level of development? Second, is there any statistical support for the proposition of greater African resistance to fertility decline, as evidenced in distinctive regression coefficients on socioeconomic determinants of fertil- ity? Table 3-1 presents the results of two regression equations based on data from 68 developing countries around the world with per capita incomes in 1989 of less than $3,000. Model 1, shown in column 1, indicates that total fertility rates in sub-Saharan Africa are significantly higher than elsewhere, net of levels of income, infant mortality, and adult enrollment ratios atta~n- ment.~2 This difference in levels of fertility is evident in the positive coef- ficient on the SSA dummy variable, which distinguishes between sub-Sa 12The specification employs 1965 infant mortality rates, because more recent mortality fig- ures could as easily be the result of high fertility as its cause, as indicated in footnote 8. (The 1991 World Development Indicators data (World Bank, 1991) provide information at only two time points: 1965 and the most recent available date.) Lagged school enrollment levels are

66 FACTORS AFFECTING CONTRACEPTIVE USE haran countries and other developing countries; it amounts to some 1.3 children. Model 2, shown in column 2 of the table, considers whether the response to socioeconomic determinants is systematically different in sub- Saharan Africa by interacting the dummy variable for sub-Saharan Africa with the other explanatory variables. We see no evidence here that fertility rates in the region are any less responsive to incomes, mortality rates, or schooling, by comparison with other developing countries outside Africa. Among all regression coefficients on interactions between the SSA dummy variable and socioeconomic determinants, none achieves statistical signifi- cance. At least in the cross section, then, there is little to support the hypothesis that fertility levels in sub-Saharan Africa are unusually resistant to socioeconomic change. Admittedly, the 1991 World Development Indi- cators data (World Bank, 1991) are less than adequate for exploring models of change, so the definitive answer to the question of resistance to fertility decline remains to be given.~3 used to explore the distinction between primary and secondary school attainment, a distinction that is not measured by current levels of adult literacy. Various alternative specifications were considered for the regressions, including some that employed 1989 per capita incomes and 1985 literacy levels, and other specifications that used averages of 1965 and the most recent values. These results differed in details, but not in terms of the conclusions that we emphasize in the text. 13The difficulties in drawing time-series inferences from cross-sectional data are well known. If in the cross section for country i, time I, we have TFRit = with + hi + Sit, then change over time (denoted as a) in country i can be represented as ATFRi = ~iD + ^£i (1) (2) if the relationship between X and TFR did not change over time. The distinction between the cross-sectional model (equation 1) and the differences model (equation 2) does not consist in the parameters beta (,B) themselves, but rather in the ability to estimate these coefficients. Suppose that unmeasured country-specific factors embedded in ui are correlated with socioeco- nomic variables Xi'. If the ui are so correlated, they would invalidate estimation of the cross- sectional equation (1) and so would render suspect any of its implications about the magnitude of fertility change attributable to changes in socioeconomic circumstances X. (In other words, the regressions of Table 3-1 would be invalid, even as a guide to cross-sectional differentials.) The differences model (equation 2) is not vulnerable on this score, but requires much more data. Even if a full first-differences modeling exercise is beyond the capacities of the 1991 World Development Indicators data, simple correlations of ATFR with AX can shed some light on the proposition of resistance. Among all less developed countries, the 1965-1989 ATFR is correlated at: -.47 with the change in (the log of) per capita income; .16 with the change in infant mortality; -.15 with the change in primary enrollment; and -.56 with the change in secondary enrollment. (The signs of these correlations give an indication of the signs of the corresponding regression coefficients By.) Within sub-Saharan Africa, the correlations are -.30 with the change in per capita income; .26 with the change in mortality; -.28 with the change in primary enrollments; and -.63 with the change in secondary enrollments.

THE SOCIOECONOMIC CONTEXT o Sub-Saharan Africa a Other LDCs 9 _ 8 - 7 - 6 - 5 4 - 3 - 2 - 67 R anda Ethiopia .0 0 0 Kenya ~ O 0 ~ ~ O 0 '~ ~ ~m ° '<: Tanzania Nigeria O A/O ~ O Zaire /.ziabab~e 3 Botswana ~/ / A/ /^ South Africa rat ~ . 1 ~ 4 ~ 6 7 8 9 Predicted TFR FIGURE 3-7 Predicted and actual total fertility rates, 1989, regression results of Table 3-1, model 1. Figure 3-7 employs the regression results from column 1 to identify the sub-Saharan countries with pronounced differentials between their expected total fertility rates, given socioeconomic determinants, and their actual TFRs. As can be seen, the largest sub-Saharan countries display fertility levels that are quite similar to their predicted values; indeed, Nigeria, Zaire, and Tan- zania show modest shortfalls of actual fertility relative to predicted. Botswana and Zimbabwe exhibit somewhat lower fertility levels than predicted, whereas the Kenyan TFR remains above that suggested by its socioeconomic stand- ing. In summary, we find strong evidence that African fertility levels are higher than those obtaining elsewhere in the developing world, even net of socioeconomic variables, but there is little in these cross-sectional data to support the proposition of greater resistance to change. EIousehold-Level Research Household-level research can also be used to explore this proposition. Socioeconomic differentials in fertility at the household level in sub-Sa- haran Africa have been reviewed by Cochrane and Farid (1990) for a sample

68 FACTORS AFFECTING CONTRACEPTIVE USE of 10 countries based on World Fertility Survey (WFS) data. They give attention primarily to differences in total fertility rates according to urban- rural residence and women's education. The size of the differentials pro- vides evidence regarding the hypothesis of resistance to change: The smaller a socioeconomic differential at a given time, the greater is the implied resistance of fertility to changes in socioeconomic factors over time. (See footnote 13 on the dangers of using cross-sectional comparisons to draw such conclusions.) As is true elsewhere, urban residence in Africa is associated with lower total fertility rates. Comparing rural women to those in the major urban centers, Cochrane and Farid find a mean difference of 1.16 in the TFlls; the bulk of the difference is due to nuptiality (or timing of entry into childbearing) rather than marital fertility. Similar gaps in total fertility rates appear in the WFS data for Asia, but the differential by residence in Latin America is much larger, amounting to 2.6 children. In regard to women's education, Cochrane and Farid find a mean differ- ential in TFRs amounting to two children between women without educa- tion and those with seven or more years of schooling. Like the rural-urban differential, that for education in sub-Saharan Africa is similar to the gap in Asia but less pronounced than in Latin America, North Africa, or the Middle East. The differential in sub-Saharan Africa is again in large part the product of nuptiality. One interesting aspect of the educational differentials in fertility in Africa is an apparent nonlinearity in the effect of women's schooling. In the majority of the countries examined by Cochrane and Farid, women with one to three years of schooling, and in some cases those with four to six years, display higher fertility rates than do women without schooling. The relationship takes the form of an inverted U. with women in the middle schooling group having the highest fertility. These nonlinearities are thought to derive from the reductions in postpartum abstinence and breastfeeding that accompany higher levels of education. They may also have to do with the unmeasured effects of income or wealth on fertility that come to be correlated with women's schooling via marriage. Tambashe and Shapiro (1991) present additional evidence on educa- tional nonlinearities in a multivariate analysis for Kinshasa. In their sample, the difference in lifetime fertility between women with primary schooling (the highest-fertility group) and those with secondary schooling is on the order of one child. Women with no schooling generally have lower fertility than women with primary schooling and higher fertility than women with more than primary schooling. They show that education is positively asso- ciated with contraceptive use. There is little difference in length of breastfeeding for women with no schooling compared to women with primary schooling,

THE SOCIOECONOMIC CONTEXT 69 however, for women with at least primary schooling, education is associ- ated with shorter periods of breastfeeding. Two general conclusions can be drawn from this research. First, one should expect continuing advances in female schooling, particularly in later primary and secondary schooling, to exert an important antinatalist influ- ence. The effect of continued urbanization appears less powerful, but in the same direction. Second, the changes in fertility associated with these socio economic differentials are not always large, especially in comparison to those of Latin America,l4 and in this sense the proposition of greater Afri- can resistance to fertility decline could be said to receive support. But the socioeconomic differentials evident in Africa are nevertheless important and suggest a range of receptivity to further family limitation. The WFS-based research is somewhat compromised by a lack of infor- mation on income, which has the potential to distort findings regarding schooling. We know of only one study of African fertility with proper controls for income, that of Ainsworth (1990) for Cote d'Ivoire. Ainsworth finds that lifetime fertility significantly decreases with female education, holding income levels constant. There is no evidence of nonlinearity in the education effects, although this result may be due to the very low levels of female educational attainment in the sample. The income coefficients re- veal that cumulative fertility increases with household income. Ainsworth concludes that if income growth goes unaccompanied by increases in fe- male educational attainment, fertility will very likely rise in Cote d'Ivoire. These studies leave unclear precisely what behavioral mechanisms are at work in the link between female schooling and lower fertility. Certainly one aspect has to do with the nuptiality effects of education, but the nature of the education effect within African reproductive unions has yet to be elucidated. The connection usually emphasized in developed-country stud- ies; which has to do with the greater opportunity cost of time in child care for better-educated women, is of doubtful relevance to sub-Saharan Africa. Another possibility, perhaps more plausible in the African setting, is that better-educated women place greater emphasis on child quality, as opposed to child quantity. This is the view stated, albeit in different terms, by Caldwell (1982) in his discussion of schooling and the adoption of Western conceptions of the child-centered family. We now turn to a more detailed exploration of three areas that appear to be critical for future trends in fertility: progress in lowering child mortal 14It should be recognized that Latin America is highly socially stratified. Therefore, one might expect larger socioeconomic differentials than those found in Africa. Comparisons between Asia and Africa might be more appropriate, and the evidence from Cochrane and Farid (1990) suggests similar differentials.

70 FACTORS AFFECTING CONTRACEPTIVE USE ity, changes in the costs and benefits of child schooling, and the notion of a crisis-led transition. EVIDENCE ON CHANGES IN CHILD MORTALITY Hill's (1990, 1991a,b, 1993) findings regarding child mortality trends Document an impressive decline in African mortality over the period from World War II to the mid-1980s. As Hill (199lb) observes, in the 1950s most African countries displayed child mortality rates in the range of 160- 400 deaths per 1,000 live births, and in only a few countries was the rate lower than 220. By the mid-1980s the range of child mortality had shifted to 60-270 per 1,000, most countries then being found in the 120-220 range. This progress is remarkable by any standard; yet in all but a few African countries, child mortality remains very high. Although the gap is closing, a mortality differential between eastern and southern Africa, and western and middle Africa, is still in evidence. Child mortality has long been highest in the western and Sahelian regions, next highest in middle Africa, somewhat lower in eastern Africa, and lower still in southern Africa. However, recent data indicate continued blurring of these regional distinctions, particularly among midlevel-mortality countries (Hill, 1993~. These regional differentials in survivorship are not easily explainable in terms of standard development indicators such as education levels or incomes per capita; they may have to do with climate, long-stand- ing patterns of population settlement, or various epidemiological peculiari- ties (Hill, 1 991b). The empirical record for the 1980s is not yet complete, but it seems that despite the economic reversals of the period, most countries managed to sustain at least modest improvements in survivorship. Experiences have been enormously varied across the continent. The situation in what are now the lowest-mortality countries in Africa Botswana, Zimbabwe, and Kenya presents a marked contrast to that prevailing in the largest countries of the region Nigeria, Ethiopia, Zaire, Sudan, and Tanzania. In the former, child mortality rates have achieved the 50-100 per 1,000 range, a level compa- rable to much of Latin America and Southeast Asia. The improvements recorded by Kenya are especially noteworthy, given its still-modest level of income per capita. In the largest countries, by contrast, mortality remains high, and it seems that rather little progress has been made over the last two decades. Among the largest countries, uncertainty attaches to the mortality record of Nigeria, owing to doubts regarding the reliability of data collected in the World Fertility Survey of 1981-1982. Using the Demographic and Health Survey (DHS) results, which she believes to be of higher quality than the WFS, Hill (1993) tentatively concludes that child mortality has been rela

THE SOCIOECONOMIC CONTEXT 71 lively stable since the mid-1970s, with a probability of dying before age 5, sqo, ranging from 190 to 200.15 No recent data are available for Ethiopia, but it is doubtful that any substantial improvement could have been re- corded over the past decade and a half. Hill's estimate for Zaire, placing 5qo equal to 210-220 per 1,000, is based on a 1984 census; no figures are available covering the post-1984 years of deepening economic crisis (Hill, 1993~. Child mortality in the Sudan (northern) apparently leveled off at a 5qo of 150 per 1,000 in the mid-1980s with possibly some decline in the late 1980s (Hill, 199lb and 1993~. Little is known about recent mends in survivorship in Tanzania. In short, the degree of progress sustained by Africa's largest countries over the past decade is much in doubt; the fragments of data available suggest only modest improvements, if any. Elsewhere in Africa, child mortality rates continued to decline through the mid-1980s even in the face of difficult economic circumstances and cutbacks in government spending. Even the Sahelian countries (including Senegal, Mali, Burkina Faso, and The Gambia), beset by drought and eco- nomic stagnation, show appreciable and sometimes substantial improve- ments. The experience of Ghana, in which mortality either leveled off or increased from the late 1970s to the early 1980s in the face of difficult economic circumstances (Pinstrup-Andersen et al., 1987; Working Group on Demographic Effects of Economic and Social Reversals, 1993), and Uganda with its political turmoil appear to be exceptions. Taken together, Hill's findings can be summarized as follows: There is nothing automatic or self-susta~n~ng about improvements in African survivorship (although recent data indicate continued mortality decline for many coun- tnes during the last 20 years). Mortality improvement is likely in part the result of economic development and service delivery. Change can be slowed or suspended by periods of sociopolitical instability (Hill, 1991a) and accel- erated by continued investments in the social sector. In Hill's (199lb) view, investments in the health sector provided something of a short-term demographic cushion for a number of countries (notably in the Sahel, but also in Togo and Cote d'Ivoire) through the 1980s. For a time the accumu- lated investments in health services may continue to sustain improvements in survivorship, even in the face of spending retrenchments and macroeconomic adversity. Eventually, however, as in the case of Ghana, such protection must erode in the absence of any new investments in health. 15Both the DHS first country report for Nigeria (Nigeria, 1992) and Hill (1991a and 1993) caution that mortality estimates based on the 1981-1982 World Fertility Survey for Nigeria are likely to be biased downward. A conclusion of increase in Nigerian mortality, seemingly implied by the difference between Hill's best guess for 1978 and the DHS 1990 figures, is probably unwarranted.

. 72 FACTORS AFFECTING CONTRACEPTIVE USE Links to Fertility Although positive correlations between child mortality and fertility do not prove a causal relationship, because of the possibility of a common cause, we think the uncertain survival of children in Africa remains one of the strongest motivations for high fertility. Hill's research shows that at least among the largest countries of the region, mortality improvements would appear to have been slight. If even now, one in five births in Zaire or Nigeria fails to survive childhood, how persuasive could the logic of lower fertility be? The experiences of Kenya, Botswana, and Zimbabwe, by contrast, are as distinctive in respect to mortality levels and trend as they are in contraceptive use and fertility. It is difficult to escape the view that there must be a close connection. In judging the prospects for fertility decline elsewhere, as these pros- pects may be linked to mortality, one issue surfaces that has curiously received rather little study. It concerns the perceptions of mortality decline as viewed by different socioeconomic strata. Given the progress that has been made over time against child mortality, have the views of African parents indeed kept pace with the empirical realities? Is there an apprecia- tion of the extent of improvement in child survivorship? Or, perhaps, do perceptions and therefore fertility decisions respond only with a long lag? One study for Lagos (Adegbola et al., 1991) found that perceptions of mortality change varied considerably across socioeconomic groups. Re- spondents with completed primary schooling agreed that mortality had been reduced in the present generation compared to the past, whereas those with less schooling denied that any such change had taken place. Indeed, in the latter group, perceptions of long-term mortality increase were found to be as common as perceptions of decline. One wonders whether such percep- tions mirror the actual experience of mortality among socioeconomic sub- groups, or whether the educated are simply in possession of better informa- tion concerning the facts. EVIDENCE ON CHANGES IN TlIE QUANTITY-QUALITY TRADE-OFF There have been surprisingly few studies of the relationship between fertility and child schooling at the household level in Africa. DeLancey (1990) reviews the evidence from a handful of investigations based on data for the late 1960s to mid-1970s for Kenya, Botswana, and Sierra Leone. This early research suggests that the relationship between child schooling and fertility has been weak in Africa, and that where a significant associa- tion is uncovered, it tends to be positive rather than negative as is the case elsewhere. This positive relationship appears to have been due to income

THE SOCIOECONOMIC CONTEXT 73 effects, such that households with higher income can afford both to have more children and to provide each child with more education. Although recent quantitative studies are lacking, there is reason to be- lieve that changes in policy and economic conditions in the past decade have rewritten the terms of the quantity-quality trade-off in Africa. As discussed below, policy shifts in some countries have caused dramatic in- creases in the direct costs of schooling, particularly in the past decade. Focus group results from Nigeria and Niger (Wawer, 1990; Adegbola et al., 1991; Makinwa-Adebusoye, 1991; Okojie, 1991) suggest that parents are becoming increasingly conscious of the costs of schooling and cite these costs as being among the principal disadvantages faced by large families. Opportunity Costs of Schooling The opportunity cost of a year of schooling has to do with the value of the child's forgone labor. These costs will vary with the age of the child in question, with urban or rural residence, and with the economic circum- stances in which the household finds itself. For a child of primary school age, who resides in an urban environment and is a member of a compara- tively high-income household, the opportunity costs of schooling may well be trivial. But if the child is older or resides in a rural setting where his contribution to the family economy can be considerable, or if the household as a whole is in strained economic circumstances, the opportunity costs of school attendance can be considerable.~7 To the extent that economic conditions continue to favor the urban sector in sub-Saharan Afnca, where child labor may not have the signifi- cance that it does in the rural sector, and to the extent that income growth returns to the region, we would expect the opportunity costs of schooling to continue to decline in importance. i6Because the opportunity costs of schooling measure the value of child labor, they measure one aspect of the benefits of high fertility. A decline in the opportunity costs of schooling thus has two effects: a direct effect that presumably reduces the net benefits of high fertility, and a presumably positive influence, akin to a price reduction, on the demand for schooling. In other words, as the value of child labor decreases, the incentives for parents to have large families to provide labor also decline. In addition, schooling for children becomes a more attractive option, because the child's contribution at home is less valuable. 17There is another dimension of opportunity cost to consider. Okojie (1991) records a consistent complaint among Nigerian parents that schooling leaves their children harder to handle and more resistant to discipline. These findings echo the views of Caldwell (1982) on the extent to which child schooling may disturb the distribution of authority within the house- hold.

74 FACTORS AFFECTING CONTRACEPTIVE USE Direct Costs Although the experiences of the past decade are varied and difficult to quantify with any precision, the trend in the direct private costs of school- ing in the region appears to have been upward. The World Bank (1988) estimates that as of 1975-1980, private expenditures accounted for some 14 percent of total national spending on education in the Sudan, 23 percent in Tanzania, 31 percent in Zimbabwe, 48 percent in Sierra Leone, and 53 percent in Ghana. The private share of total spending has increased over time as governments place greater emphasis on the collection of fees and other charges for educational materials, and increasingly transfer responsi- bility for capital expenditures to local communities, which in turn pass them on to parents. For Nigeria, Makinwa-Adebusoye (1991) draws a sharp contrast be- tween present-day private costs of schooling and those of an earlier era when education (and health) services were highly subsidized. She presents examples drawn from focus group studies confirming that the perceived costs of education are high in both Lagos and Calabar. Okojie's (1991) results for Bendel and Kwara State also show that parents are conscious of a dramatically changed situation with respect to educational costs. For Kenya, Kelley and Nobbe (1990) and Robinson (1992) see increases in the costs of schooling as being associated in part with the educational reform of 1985 (which added an additional year to the primary curriculum) and with the introduction of new financing mechanisms that required par- ents to pay in direct proportion to the number of children enrolled. (Kenyan parents do not pay tuition fees; rather, they are responsible for school sup- plies, books, uniforms, and a share of capital costs.) Most of the capital costs for primary schools have been transferred from the national to the local level. Taken together, the private costs borne by Kenyan parents are said to be high, perhaps amounting to as much as 10 to 15 percent of household cash income per child. In Zimbabwe, Lucas (1991; citing Davies and Sanders) notes that pri- mary school fees were abolished in the early 1980s and then, in the face of a very rapid rise in enrollments, were replaced by various levies and capital charges (also see World Bank, 1988~. In Botswana, by contrast, public sector primary school fees were reduced through the 1970s and finally eliminated altogether in 1980. Secondary school fees (at state-supported schools, which enroll roughly two of every three secondary students) were eliminated in |8The World Bank (1988) does not provide a source for these figures, but they appear to include outlays on private educational institutions as well as the fees and capital levies borne by families with children in public sector institutions.

THE SOCIOECONOMIC CONTEXT 75 1988. Evidently the elimination of fees in Botswana was not counterbal- anced by increases in other charges. In summary, the general trend is that of an increase in the direct costs of schooling, although Botswana may provide an exception to the trend and the net change in Zimbabwe is unclear. Very little information is available regarding fees and other costs in the private educational sector, which is a significant presence at the secondary school level in a number of African countries (see World Bank, 1988~. Perceived Benefits of Schooling If the direct costs of schooling are indeed on the rise, will the perceived benefits of schooling in Africa be substantial enough to sustain a quantity- quality transition? The evidence on this critical issue is conflicting. By all accounts, African parents express a continued commitment to schooling and display a willingness to make sacrifices in other areas so that their children can be educated (see, among others, World Bank, 1988; Adegbola et al., 1991; Okojie, 1991; Makinwa-Adebusoye, 1991~. But there is great uncer- tainty regarding the economic benefits of schooling in African economies and therefore in the extent to which educational aspirations, however deeply felt, will be reflected in educational investments. The benefits are expressed principally in the gradient of earnings with respect to education; earnings, in turn, depend on the relative supplies of and demands for labor by educational level. The demand for educated labor depends in part on the growth of modern sector employment, which contin- ues to be slow in sub-Saharan Africa (Hansen, 1990; Vandemoortele, 1991~. The public sector component of employment will continue to be affected by cutbacks in government budgets. Vandemoortele (1991) finds a compres- sion of wage scales in the public sector and general declines over the past decade in the real earnings of civil servants. To the extent that educational investments in children are made in the expectation that they will eventu- ally secure employment in the public sector, these changes may reduce the private returns of schooling. However, as Vandemoortele points out, coun- tries in which public sector pay is deteriorating often show counterbalanc- ing increases in private sector earnings. The net effect on the anticipated payoff to schooling remains unclear. There is clear evidence of an erosion in the quality of schooling in sub- Saharan Africa, such that the economic gains gleaned from schooling by older cohorts may no longer be available to younger cohorts. The evidence regarding quality decline is drawn from a mix of statistical and impression- istic accounts. In a careful study of the Ghanaian wage sector, Glewwe (1991) finds that the returns to schooling evident in wage differentials have declined for more recent cohorts; indeed, the cognitive achievements of

76 FACTORS AFFECTING CONTRACEPTIVE USE recent cohorts per year of schooling are lower than those achieved by older cohorts.~9 Glewwe attributes this change to difficulties in sustaining school quality over the postindependence period of rapid expansion in primary and secondary enrollments (also see Schultz, 1987~. Among other factors influencing school quality, the World Bank (1988) draws attention to the very low level of public spending per pupil in Africa at the primary and secondary grades (if not at the tertiary). There is evi- dence that spending per pupil may have even declined in the 1980s (see below). In the important category of educational matenals, sub-Saharan governments make very small contributions. Public expenditures for educa- tional materials (World Bank, 1988) amount annually to less than 60 cents per pupil. The share of the primary education budget going to instructional materials is only 1.1 percent, compared to a 4.0 percent share among devel- oped countnes. (Recall that the responsibility for these expenditures is increasingly being shifted to parents.) All these factors are consistent with the relatively poor performance of African students on standardized tests of achievement in math, reading, and the sciences, even by comparison to their counterparts in other low- or middle-income countries (World Bank, 1988~. Moreover, the impression of decline in school quality in Africa is not con- fined to statistical accounts. It is also a widely shared perception among African parents and a common theme in conversation. To sum up, we regard the prospects for a quantity-quality demographic transition in the region as being decidedly mixed.20 Recent accounts taller from focus group studies (e.g., Adegbola et al., 1991; Okojie, 1991) suggest that changes in the costs of schooling are perceived to add significantly to the financial burdens of large families. In time, these relative price effects |9Recent studies show that with native ability held constant, cognitive achievement explains a large part of educational differentials in earnings. That is, an extra year of schooling is rewarded in the labor market primarily because it produces a change in the cognitive skills that are valued in the market. Among wage earners in Nairobi and Dar es Salaam, Boissiere et al. (1985) find strong evidence that cognitive achievement is a more powerful determinant of earnings than years of schooling per se; achievement is itself affected by years of schooling as well as by innate ability. Glewwe's (1991) findings for Ghana are similar. In summarizing the results for Kenya and Tanzania, the World Bank (1988) suggests that an earnings differential of 25 percent for secondary relative to primary schooling is attributable to the improvement in cognitive skills produced by secondary school. The above is not to deny the importance of credentialism and other factors in determining the private returns to education in Africa. But tests for "diploma effects" (Boissiere et al., 1985; Glewwe, 1991) show that the effects are weaker than the influence of cognitive achieve- ment, save perhaps in the public sector. Credentialism may have its largest effect on earnings at the outset of the work career. 20The picture is further muddied when child fostering and spouse decision making are con- sidered in relation to the quantity-quality transition (see Chapter 4).

THE SOCIOECONOMIC CONTEXT 77 may help to bring about a fertility transition. But we cannot assume that African parents will continue to make sacrifices to invest in human capital if the payoff to that investment is in doubt. Indeed, in some countries, primary school enrollment rates fell in the 1980s, as discussed below. Policy- driven changes in the direct costs of education can have little long-run effect on fertility if the economic benefits of schooling cannot be sustained and improved. ECONOMIC STAGNATION AND ADJUSTMENT: EFFECTS ON FERTILITY The 1980s and early 1990s witnessed a contraction of incomes across sub-Saharan Africa, such that many countries were left only slightly better- off in terms of average incomes than they had been in the early 1960s (Vandemoortele, 1991~. During the decade a number of countries responded to economic stagnation and crisis with structural adjustment programs. These programs sought to redirect government spending and to reduce the size of the government sector; in addition, they were often designed explicitly to influence the relative prices of food, education, and health care. Thus, the past decade has been one of profound change not only in income levels, but also in relative prices and policies determining access to social services. In an influential paper, Lesthaeghe (1989a) speculated that the economic re- versals and turmoil of the 1980s in sub-Saharan Africa might hold the seeds of what he termed a "crisis-led" demographic transition. In one sense, the idea of "crisis" breaks no new ground, in that it simply revisits many of the socioeconomic factors that have already been discussed above. The difference is that these factors are brought together in a particular configuration, and in addition, the concept introduces a distinc- tion between short-term demographic effects at work in the crisis period itself and the potential longer-term consequences. Thus, comparatively temporary income contractions may have effects on fertility decisions that depend on expectations regarding longer-run income growth. Government retrench- ments, intended to ameliorate short-term problems in the balance of pay- ments, may have a long-run effect on access to education and health ser- vices. And a temporary receptivity to family limitation on the part of households under economic pressure may translate, over time, into a greater acceptability of family planning in more normal circumstances. There is little doubt about the depth of economic crisis in many African countries, although experiences are perhaps more varied than might have been thought. In some countries the term "stagnation" better describes the situation during the past decade than does "crisis." Figure 3-8 depicts the changes in income per capita between 1975 and 1987 in the five largest countries of the sub-Saharan region. Population-weighted averages for sub

78 1200 1000 800 600 400 Boo FACTORS AFFECTING CONTRACEPTIVE USE GNP per Capla, Dow \ \ \ O 1975 1977 1979 1981 Year ~ Zaire - End + Ted ~ So ~ Nag 1983 1985 1987 ~ Sit AInca FIGURE 3-8 Per capita GNP, 1975 to 1987 (1980 dollars). SOURCE: World Bank 199Oa). Saharan Africa. A steady deterioration in income levels can be observed in the region from the high point in 1981 Among the larger countries, the collapse of incomes in Nigeria is particularly marked, but Zaire has also experienced a long decline. By contrast, among the three countries with higher prevalence, Botswana has exhibited a reasonably steady advance in income over the decade, and Kenya and Zimbabwe have displayed fluctuat- ing incomes with little apparent deterioration (see Figure 3-9~. In considering the longer-run prospects for demographic transition, it is important to separate the effects of income on fertility from the effects of changes in relative prices. As long as fertility remains a normal good (one whose consumption increases with income) in the economic sense of the term, one would expect income contraction to be accompanied by a fertility decline, or at least by a pause in family building. One would equally well expect a return to high fertility as income levels improve. Thus, if current economic circumstances are to set off a demographic transition in sub- Saharan Africa, the origins of the transition must lie in a transformation of both incomes and relative prices, which is precisely the argument employed above in reference to the quantity-quality transition. The implications of relative price changes induced by structural adjust

THE SOCIOECONOMIC CONTEXT t800 1600 1400 1200 1000 8009 600 400 200 o 1975 1977 1979 1981 GNP per Capla, Dow , , , . , . . - Year - Kenya ~Botswana * ~rr~we 79 1983 1985 1~7 FIGURE 3-9 Per capita GNP in Botswana, Kenya, and Zimbabwe (1980 dollars). SOURCE: World Bank (199Oa). ment for fertility are less than clear-cut (World Bank, l990c). Food price increases, which drive down the standard of living for urban consumers, may well improve living standards among rural producers (indeed, this is the prime motivation for adjustment policies that remove artificial food price ceilings). If conditions in agriculture improve, might that not increase the derived demand for child labor and exert a pronatalist influence on rural fertility? Would improvements in agricultural earnings reduce the disper- sion in earnings levels according to schooling, further dampening the pros- pects for a quantity-quality transition? Do cutbacks in public sector salaries and employment, however well justified with respect to longer-term eco- nomic efficiency, further depress the private returns to schooling and thereby delay the quantity-quality transition? The task of documenting the distribution of benefits and costs associ- ated with economic reversals and adjustment policies is just under way in sub-Saharan Africa (World Bank, l990c; Working Group on the Demo- graphic Effects of Economic and Social Reversals, 1993~. Nothing defini- tive can be said about these complex matters, and given the many interlock- ing markets that are involved, economic theory alone can provide no clear guidance. However, the current economic situation, which combines in- come contraction and various increases in childrearing costs, may well have rendered acceptable, for the first time, the notion of family limitation. Such

80 FA CTORS AFFECTING CONTRACEPTIVE USE a development is certainly consistent with the views expressed in the results of qualitative studies in Nigeria (Caldwell and Caldwell, 1987; Adegbola et al., 1991; Makinwa-Adebusoye, 1991; Okojie, 1991), in which Nigerian parents seem to comprehend rather fully the negative consequences of high fertility in the new regime of income and prices. The negative aspect, however, is that macroeconomic austerity may reduce the capacities of gov- ernments to improve access to education, family planning, and health ser- vices. It is therefore important to assess what has happened to public investments and access to services over the past decade. Access to Education and Health How have the social sectors, including education and health, fared in an era of general retrenchment in government spending? Hicks (1991; also see Hicks and Kubisch, 1984) finds that in periods of overall reduction in gov- ernment expenditures in developing countries, recurrent expenditures on the combined social sectors (including all spending on health, education, hous- ing, and other social services) tend to be well protected in comparison to the productive, infrastructural, and general public sectors. That is, the social sectors tend to experience less-than-proportionate cuts in budget as total government budgets decline. Capital expenditures tend to be cut much more severely than do recurrent expenditures, and these cuts tend to be in roughly the same proportion across all sectors. Thus, perhaps contrary to expectation, the policy response to economic crisis and adjustment has typi- cally been to compromise in the dimension of physical capital investments. However, even if the education and health sectors have been able to protect their shares of government budgets, they have not been able to escape cutbacks on a per capita basis.22 Figure 3-10 shows the trends in per capita central government spending on education, both for the largest coun- tries in the region and for sub-Saharan Africa as a whole (for the latter, results are weighted by population). There is an unmistakable downward 2iThe sample used by Hicks (1991) includes, but is not limited to, countries from sub- Saharan Africa. The productive sector includes expenditures for industrial and agricultural development; infrastructure includes power, transportation, and communications; and general public services encompass administration, police, and the judicial system. 22In their analysis of the 1979-1983 period in sub-Saharan Africa, Pinstrup-Andersen et al. (1987) note a stagnation or decline in per capita expenditures on health and education, with the decline being more apparent in respect to health expenditures. They note considerable diversi- ty among countries in the extent to which the health and education sectors maintained their budget shares. Like Hicks (1991) and Hicks and Kubisch (1984), they find that capital expen- ditures tended to be cut more than recurrent expenses, particularly in the economic services sector.

THE SOCIOECONOMIC CONTEXT 25 20 15 103 Expenditure per Capita, Dollars O i ~i 1 980 - 1 ~ 1981 1982 1983 1984 1985 1 ~1~7 Year - Niger a ° Tanzania ~Zaire 81 Ethiopia ~ Sudan ~ Sub Saharan Avenge FIGURE 3-10 Central government educational expenditure per capita. SOURCE: World Bank ( 1990a). trend in per capita education spending for the region; among the large coun- tries this trend is mirrored in the experiences of Zaire, Tanzania, and Nige- ria. Figure 3-11 considers health expenditures per capita; here, there is less evidence of a systematic downward trend. However, the very low levels of per capita expenditure on health in sub-Saharan Africa certainly did not improve. Consequences for lIuman Capital Investment With regard to primary enrollments, the 1980s appear to have marked the end of an era of expansion in sub-Saharan Africa. Primary enrollment ratios, which reached their high point of 74 percent in 1981 (population weighted), gradually slipped back to less than 70 percent by the end of the decade (United Nations Educational, Scientific and Cultural Organization, 19901. Among the largest countries, Nigeria, Zaire, and Tanzania experi- enced sharp declines in enrollments from about 90 percent in the early 1980s to 65-75 percent in the latter part of the decade. These declines are no doubt due in part to income effects and in part to policy-driven increases in the private costs of schooling described above. Let us emphasize again that the more responsive the demand is for education with respect to price, the less likely are the prospects for a quantity-quality transition in fertility. Moreover, the primary-age cohorts of the mid-1980s are now entering their

82 7 64 4 3 2 1 , O L 1980 1981 1982 1983 1984 1985 1 ~1~7 FACTORS AFFECTING CONTRACEPTIVE USE ExpendRum ~ C - = ~1~ Year - Sudan + NbeNa ~ Rhbpla ~Tot Are Sub Seaman Avenge FIGURE 3-11 Central government health expenditures per capita. SOURCE: World Bank (199Oa). childbearing years. They will be without the benefits of the schooling levels that they might otherwise have enjoyed in the absence of an eco . . . nomlc crisis. The situation with respect to secondary schooling is very different. Enrollment ratios at this level continued to advance through the decade, although Zaire again and Tanzania present an exception to the general trend, with Zaire showing a dramatic decline and Tanzania showing little change. The advance in secondary schooling by Nigeria, from less than 20 percent in 1980 to about 24 percent in 1987, is especially noteworthy, given the decline in its primary enrollments over the period (United Nations Educa- tional, Scientific and Cultural Organization, 1990~. Also of interest is the continuing increase in the enrollment ratios of women relative to men. Although males continue to outnumber females at both the primary and the secondary levels, the enrollment gap between them is being reduced progressively (United Nations Educational, Scientific and Cultural Organization, 1990~. Summary The diverse experiences among countries show that there is no neces- sary connection between macroeconomic stagnation and crisis, and cutbacks

THE SOCIOECONOMIC CONTEXT 83 in access to services (Pinstrup-Andersen et al., 1987~. In Kenya, for ex- ample, the period from 1979 to 1982 saw an annual decline in per capita gross domestic product on the order of 1 percent per year, yet government expenditures on health and education continued to increase. Cornia and Stewart (1987) cite the cases of Botswana and Zimbabwe, which met their episodes of severe drought in the early 1980s with well-designed drought relief programs that targeted health care and children's supplementary feed- ing programs. Health expenditures increased, in general, even as economic conditions deteriorated; primary school enrollments also continued to rise. Moreover, the recent study by another working group of the Panel on the Population Dynamics of Sub-Saharan Africa, the Working Group on Demo- graphic Effects of Economic and Social Reversals (1993), indicates that the effects of economic reversals have varied considerably from country to country. CONCLUSION The socioeconomic record offers few general lessons regarding fertility transition that can be applied across the whole of sub-Saharan Africa. Cer- tainly the steady increase in female educational attainment, albeit from very low levels in some regions, may remove one of the props supporting high fertility. But in other respects the likelihood of fertility decline is very much a country-specific matter. Although mortality has fallen substantially in many sub-Saharan countries, the decline is not universal. Most of the largest countries still exhibit relatively high levels of child mortality. In Nigeria, the largest country, little progress, if any, has been made in the last decade and a half. Changes in the costs of living, and specifically the costs of education, may well hold the key to fertility trends in some countries, but as we have argued, the responses to higher education costs depend on the nature of the benefits to schooling. The economic crises of the 1980s may have opened the door to the acceptability of family limitation, but if income growth resumes without structural change, fertility decline need not follow. The experiences of sub-Saharan countries have been so varied in re- spect to socioeconomic development, and governments so heterogeneous in their social sector policies, that no general forecast regarding fertility and contraceptive use should be made. Enough has been said to indicate that the three countries that may be the forerunners of fertility change, Kenya, Botswana, and Zimbabwe, are distinctive with respect to policy and socio- economic setting. Among the larger countries, Nigeria shows some indica- tions of sensitivity to the costs of schooling, but its decline in primary enrollments is worrisome. Ethiopia and Sudan remain in sociopolitical turmoil, which will necessarily limit the reach of policy; Zaire gives evi- dence of severe deterioration over the past decade with worsening condi 4

84 FA CTORS AFFECTING CONTRACEPTIVE USE lions since 1990. None of these large countries displays the steady ad- vances against mortality characteristic of the higher use countries, and it is doubtful that major fertility declines at a national level can be initiated without such improvement. Yet even in the larger countries, economic stagnation, and structural adjustment may have brought into relief previously latent demands for fam- ily limitation among important subgroups of the population. In an atmo- sphere of austerity, policies addressed to these subgroups may find a newly receptive audience.

Next: 4 The Household, Kinship, and Community Context »
Factors Affecting Contraceptive Use in Sub-Saharan Africa Get This Book
×
Buy Paperback | $50.00
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

This book discusses current trends in contraceptive use, socioeconomic and program variables that affect the demand for and supply of children, and the relationship of increased contraceptive use to recent fertility declines.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!