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.
7 Contribution of Modern Contraceptive Use Relative to Postpartum Practices to Fertility Decline This volume has focused on contraceptive use, not fertility per se. However, the keen interest among the population community in contraceptive preva- lence in the higher use countries of Botswana, Kenya, and Zimbabwe stems from the possible implications for fertility decline in these and other coun- tries. In analyses of the variation in marital fertility in developing countries over the past two decades, contraceptive use has been the major determin- ing factor in the majority of cases. Traditional postpartum practices (i.e., intermediate or proximate fertility variables including fecundity, lactational amenorrhea, and postpartum abstinence) have played a secondary role in other parts of the developing world (Bongaarts, 1978~. However, in Africa, postpartum nonsusceptibility has been shown to be a major determinant of variations in fertility levels (Bongaarts et al., 1984~. Postpartum practices are closely related to cultural patterns and forms of social organization, particularly in Africa where regional differences with respect to these proximate determinants are very large (e.g., see Chapter 4; Caldwell, 1976; Page and Lesthaeghe, 1981~. In light of the importance of these postpartum practices to fertility, an exclusive focus on contraceptive practice in the context of sub-Saharan Africa would be inappropriate (see Appendix A). In this chapter, therefore, we address one final question: In Africa, 197
198 FACTORS AFFECTING CONTRACEPTIVE USE what has been the relative importance of modern contraceptive use versus postpartum infecundability in inhibiting fertility? Central to this question is the theory of a two-phased fertility transition in Africa, which we review before proceeding to an analysis of the available data on a regional level. THE TWO-PHASED FERTILITY TRANSITION The basic premise of this transition theory is that fertility decline in sub-Saharan Africa will occur in two phases, a notion entertained by several authors writing about fertility transition in general (e.g., Kocher, 1973; Easterlin, 19831. These authors envision the possibility of an initial fertility rise (the first phase) occurring prior to a fertility decline (the second phase). This initial rise would stem from socioeconomic development factors affecting the levels of natural fertility and the supply of children. A decline in subfecundity or secondary sterility resulting from improved health care (Romaniuk, 1968; Retel-Laurentin, 1974; Frank, 1983a,b; Larsen, 1989), and the shortened durations of breastfeeding and postpartum abstinence are often strongly associated with increased female education and urbanization (cf. Olusanya, 1969; Caldwell and Caldwell, 1977; Nag, 1980; Romanink, 1980; Adegbola et al., 1981; Gaisie, 1981; Mosley et al., 1982; Lesthaeghe et al., 1983~.2 Moreover, the initial increase in the supply of children is further enhanced by increased infant and child survival. With declining demand for children stimulated by some of the same socioeconomic changes, the second phase of the transition is set in motion. This phase is characterized by declining desired family size, increased knowledge of contraception, and subsequently increased use-effectiveness of fertility regulation. Reduced demand for children as a response to diminished child utility, increased costs of childrearing, and higher aspirations with respect to child quality are often associated with the same factors of socioeconomic development that produced the initial fertility rise (see Chapters 3 and 4~. In Africa, as in other parts of the world, female education is strongly linked with fertility (see Chapters 2 and 3~. From Chapter 6 it is evident that education is the strongest of the four contextual variables tested in d 1Because of the focus of this report on contraceptive use primarily among married women, we devote little attention to marriage patterns and their effects on fertility. For discussions of this topic, see Westoff, 1992; Jolly and Gribble, 1993; and the Working Group on the Social Dynamics of Adolescent Fertility, 1993. 2Although declines in the postpartum nonsusceptible period before the late 1970s are well documented, there is no study that systematically relates these declines to increases in fertility. However, it is the view of the working group that these declines, in addition to decreases in subfecundity, most likely led to increases in fertility.
CONTRIBUTION OF MODERN CONTRACEPTIVE USE TO FERTILITY DECLINE 199 determining modern contraceptive use. It is also associated with patterns of nuptiality and with postpartum practices in ways that are basic to African social organization, as discussed below.3 Regional Patterns of Nuptiality The ages at entry into a regular sexual union vary widely across sub- Saharan African regions.4 The proportions of single women aged 15-19 range from 10 to more than 90 percent. The corresponding singulate mean ages at marriage vary from about 16 to more than 21 years. The main determinants of ages at entry into a sexual union are also well known (cf. Goldman and Pebley, 1989; Lesthaeghe et al., 1989a). First, there are a number of factors that are intimately linked to patterns of social organization. The incidence of polygyny is often singled out as a prime factor leading to early marriage for women, because polygyny presupposes a large age difference between the spouses and hence the combination of late marriage for men and early marriage for women. Lesthaeghe et al. (1992) examined the relationship between the incidence of polygyny and the age pattern of entry into a sexual union for women for the regions covered by the World Fertility Survey (WFS) and the Demographic and Health Survey (DHS). They found that highly polygynous societies can exhibit both early and fairly late marriage for women, provided the large age gap between spouses is maintained. A second determinant of early mamage for women has been the Islamic influence on social organization. As argued by Goody (1973, 1976) and statistically supported by Lesthaeghe et al. (1989a), an early and more pro- found Islamization is associated with a preference for cousin-marr~age (en- dogamy) and a tighter social control on women via early first marriage or fast remarriage following divorce or widowhood. Lesthaeghe et al. (1992) found that the Sahelian Islamic societies contributed quite heavily to the set 3The fourth proximate determinant, abortion, is not treated herein, given the dearth of reli- able data on this topic. However, with possible exceptions in major urban areas among better- educated young women, abortion would appear to be fairly limited in African populations. 4Both the World Fertility Survey and the Demographic and Health Survey use a broad definition of marriage and include women reported as "having a partner" as being in a regular sexual union. Occasional premarital sexual relations are not taken into account by the defini- tion. The degree of tolerance of premarital sex varies considerably according to ethnic group and local custom. The norms range from a strong accentuation of premarital chastity to premarital sex and pregnancy being acceptable and occasionally desirable (Working Group on Social Dynamics of Adolescent Fertility, 1993).
200 FACTORS AFFECTING CONTRACEPTIVE USE of regions that had both a high incidence of polygyny and very low propor- tions single in the age group 15-19. A third factor involved is matrilineal kinship organization. This factor is generally associated with later marriage for women, largely because matrilineal societies have less polygyny than neighboring patrilineal or bilateral ones (cf. Lesthaeghe et al., 1988; Chapter 4~. Two additional factors have altered past practices: female schooling and urbanization. Both tend to be associated positively with the proportion of single women in the age group 15-19, either directly or through a nega- tive effect on polygyny. Lesthaeghe et al. (1992) found that the vast major- ity of regions with average schooling durations of four years or more (or with less than 50 percent illiterate women) have proportions of single women 15-19 in excess of 60 percent. At lower levels of education, however, there was evidence of several confounding factors: polygyny, the proportion Muslim, and the survival of traditional religions and syncretic churches. These factors shape the positive relationship between female schooling lev- els and ages at entry into first union at the aggregate level. Regional Patterns of Postpartum Infecundability Postpartum infecundability potentially affects fertility in any society. Yet it is of particular interest in sub-Saharan Africa because of the pro- longed periods of breastfeeding and sexual abstinence, which lead to peri- ods of postpartum nonsusceptibility that are much longer than in most parts of the developing world. The relationship of these postpartum variables to variations in fertility has been amply documented, as has the erosion of these practices throughout Africa during the twentieth century (Bongaarts, 1981; Caldwell and Caldwell, 1981; Ferry, 1981; Gaisie, 1981; Lesthaeghe et al., 1981; Schoenmaeckers et al., 19811. Table 7-1 presents data on the postpartum measures, as well as average length of schooling of women, for the countries and regions for which both WFS and DHS data are available. Clearly, there has been little further erosion of these means of child spacing between the survey dates for these regions. Even so, with regard to breastfeeding, the sub-Saharan populations examined here still have relatively long durations in comparison with many other regions of the world. For example, regions with aggregate schooling levels of five or more years have an average duration of breastfeeding of about 17 months, with none dipping below one year. The length of lactational amenorrhea in the regions at the time of the DHS varies between 9 and 20 months, and the mean duration for the regions with higher levels of schooling is still on the order of 11 months. The period of postpartum abstinence is considerably shorter than that of lacta- tional amenorrhea in three of the four countries shown in Table 7-1; it
CONTRIBUTION OF MODERN CONTRA CEPTI VE USE TO FERTILITY DEC f INK 201 varies in duration from an average of three months to as long as 27 months (e.g., Upper and Northern regions of Ghana). The outcome variable that directly matters for fertility is the length of the overall postpartum period of nonsusceptibility (PPNS). It is defined for each individual as the longest of either the duration of lactational amenorrhea or the length of postpartum abstinence. The "abstinence bonus," the differ- ence between the mean length of the overall PENS and the mean duration of lactational amenorrhea (Lesthaeghe, 1989b), is about three months when the average duration of postpartum abstinence equals that of lactational amenorrhea. However, the abstinence bonus rises rapidly if the mean of abstinence be- comes larger than the mean of amenorrhea. Abstinence bonuses of the order of five to ten months were common in the WFS regions of Cote d'Ivoire, Benin, Ghana, Nigeria, and Cameroon. In the DHS regions, absti- nence bonuses of four to ten months are found in Liberia, Togo, and Ondo State (Nigeria). In short, the fragmentary evidence for a small subset of regions sup- ports the thesis that the traditional mechanisms of child spacing may have remained fairly intact at the durations observed during the late 1970s. Therefore, gains in contraceptive use would have become the primary source of change in marital fertility during the 1980s. It is, however, still necessary to stress the conditional nature of this statement because the presumed absence of a further downward trend in the postpartum variables requires documentation for many other regions of Africa as well. Notwithstanding the evidence of "no further erosion" in the 1980s, fe- male education has been one of the prime factors involved in the shortening of birth intervals in sub-Saharan Africa during at least the last three de- cades, as demonstrated by a number of individual-level analyses (e.g., Olusanya, 1971; Caldwell and Caldwell, 1977; Page and Lesthaeghe., 1981; Mosley et al., 1982; Gaisie, 1984; Locoh, 1984; Tambashe, 1984; Mpiti and Kalule- Sabiti, 1985~. Thus, increased schooling levels for women have been par- tially responsible for the initial fertility increase during the first phase of the transition. At the regional level, similar negative relationships to education are found within the WFS and DHS samples. This holds for all postpartum variables, although we graph in Figure 7-1 the relationship for only one of the four: the length of the postpartum nonsusceptible period. Role of Education in the Two-Phased Transition The intricate link between education and the two-phased transition ap- peared in Cochrane's (1979, 1983) analysis of cross-sectional data gathered at the individual level in a variety of countries. As discussed in Chapter 3, Cochrane found that monotonically increasing relations of fertility with education
202 C) 4 - o m C) So o cq o ._ be So Cal U: 1 U) ._ Cal - Ct ._ So · ~, a; cc o o U. o . _ So Cal o CC ._ U: 0 ~ 1 ~ m u: EN 3 o ~ ~ Ct of ~ ~ ;^ 04 ~ o (L) o ~ ,. _ i= C , · ~ Cal o Cal ~ ·,- IS Cal ~ o ~ if, Cal o U: o 04 s . C) ~ ~ _ V, Cal ~ o sly ~ in, Cat a Cat 3 3 3 Cat 3 U. V: 3 Cal a Ct o . oo go ~ ~ o ~ ~ . ~. . . ~ oo ~c~ . ~ c~ ~ ~ c~ ~ ~ oo ~ Mo ~ ~ ~ ~ ~ ~o c~ ~ ~ 0 ~r~ ~ ~oo - ~ ~ ~ ~ ~ ~c~ ~ ~ cM ~ ~ ~o v) o _d _I ~ ~ ~ ~ r~ ~ o ~ ~ ~cM c~ ~ ~ ~ ~ ax~ oo ~ o ~ ~ ~ ~ ~ ~ - 4 ~ ~ ~ oo ~ ~ ~ ~ c~ ~ ~ - 4 ~ - ~ ~- ~ ~ 0 0 cn oo O 0N o0~ O r~ ~ ~ ~ o ~ ~ ~ ~ - ~ - ~ ~ ~ ~ 0 ~. cO ~C ~ ~ Z '~ ~ ~ 3 .S ~ v ~ ~ z ~v
203 ~ Do ~ onUS Cal ~ ~ . . . .. . . . o o ~ CMCal Vie ~ ~ ED OCal ~ \0 0 o o o Calo Cal o Do ~ ~ No~ ~ Go ~ ~ ~ ~_4 ~ _4 ~ _- ~ ~ o ~ Cal _4 I_ ~ ~, ~ ,, ~ Do ~ Vie U) ED ~, , Do ~ ~ ~ ~ ~ ~ ~ _. ~ ~ ~ ~ ~ ~ To ~ _ ~ ~ _. ~ ~ ~ ~ , ~ , , o of o ~ o ~ ~ o CM of Cal ~ Go ~ Go _. I_ '_ ~ Ct =) C s-= O2C . C,==o~o V) Cal · \=, . . C) ~ it: ON A: ._ ~ ~ 4 - '~ ( C, do ~ -Cal a' 3 am Ct on _ c: ~ o . ~ ; - o ~ Ct - Cal ~ o en o ~ A' o ~ - ~ 'e o ~ A, -= - ~ ~ o c) ~ ~ o 'e ~ . c: ~ ~ o At ~ ct of D Ct ~ ~ 1-' Cal ~ O ,= O O - I_ Ct ~ sir C,, .. U. O ,s: Ed - ._ C) Cal CO 53 o - Cal - CC IS O _ ~ :S ~ o { ·- ~4 O ~ C,) '- ~L) ~ lo: A> ~ C) 3 _ ~ Ct ~ ~ ..r, C) o V, ~ o o Id, - 5- C) o ~ ._ Ct ._ Ct o Cal _ o Cal c: o ._ Ce Cal ~ ~ ._ Cal 5: ~ ~ s ; - U: I- ~ Ct o _ ._ ._ ~ ~ Ct ._ ~ V, '_ ~ ~ c5 - (L) Ct ~ C.) 50 ~ o o ;^ sin
204 35 U) 4. to E . . ._, it_ . - .,. Q 'A 20 tn o in Q U. to FA CTORS AFFECTING CONTRA CEPTI VE USE 30 25 10 o o a o o o o o o o o ° o o o °@,° ~ go ~ No o ° o Cal o of oO° o o o o o r, o o o °° boo o o o ° o o o Oo oN o=o ~ 0 1 2 3 Education: years o 4 5 6 7 9 FIGURE 7-1 Relationship between the length of postpartum nonsusceptibility in months and the mean number of years of female education WFS and DHS regions (r= .53~. SOURCE: Demographic and Health Survey and World Fertility Survey reports. or inverted U-shaped relations were typical for areas in which female edu- cation acted more strongly in favor of increased natural fertility and re- duced child spacing than in favor of enhanced contraception. The cross- sectional effect of education on fertility is more negative at later than at earlier points in time, and more likely to be found in countries that have higher levels of urbanization, per capita income, and daily caloric intake (Cochrane, 1983~. These conclusions were largely borne out for sub-Saharan Africa in the contextual analysis of female education performed by Lesthaeghe et al. (1989a) for regions covered by the WFS. Through the calculation of the Bongaarts (1983) proximate determinants indices, these authors attempted to measure the effects of the overall nonsusceptible period and contracep- tive use on fertility. These indices were calculated for different educational groups within various age groups for each of the 33 WFS regions (Nigeria excluded). Overall child spacing was viewed as being negatively affected by education in these cross sections if the better-educated women (five or
CONTRIBUTION OF MODERN CONT~CEPTIVE USE TO FERTI~TY DECLINE 205 more years of schooling) did not offset the effect of reduced overall post- partum nonsusceptibility by increased contraception and would, therefore- other things being equal-have higher marital fertility than illiterate women. The same contrast was also examined for women with a few years of schooling (one to four years) compared to illiterate women. The conclusions of this study, which pertain predominantly to the late 1970s and capture many regions in western Africa during the first phase of the transition, can be summarized as follows: In areas where female education was low to begin with and Islam predominated, child spacing was reduced most by increases in education. Conversely, in less-Islamic regions, increased use of contra- ception was found to offset the effects of decreased postpartum nonsusceptibility. Hence, in the late 1970s, female education operated both as an indi- vidual and as a contextual variable, and the balance between child spacing via lactational amenorrhea and postpartum abstinence on the one hand and contraception on the other was conducive toward a positive fertility-educa- tion relationship in areas with the lowest scores for female education. However, one might expect that eventually as overall levels of education and contra- ceptive use increase, a negative relationship would hold. These findings for the late 1970s are in line with the general argument advanced by Caldwell (1980) that mass education, especially when it incorporates women, is ca- pable of triggering fertility transitions in developing countries. The effects, as indicated by WFS data, would begin to emerge once most women receive partial or full primary education and when illiteracy among them has been considerably reduced. Mass education, however, affects all components of reproduction, including those that initially produce an increase in the poten- tial supply of children. Similarly, mass education is an equally forceful agent in the reduction of infant and childhood mortality and may therefore account for much of the statistical relationship found between the pace of the early-life mortality reduction and declining fertility. CONTRACEPTION, NONSUSCEPTIBILITY, AND FERTILITY DECLINE We return to the question of the relative contribution of contraceptive use versus postpartum nonsusceptibility to fertility change. Specifically, using WFS and DHS regional data, we examine the association of female education with the combined effects of postpartum nonsusceptibility and contraception and with the trade-off between the two fertility-inhibiting factors. We first convert the mean length of the period of postpartum nonsusceptibility and the proportions of women using contraception to the Bongaarts indices
206 FACTORS AFFECTING CONTRACEPTIVE USE of nonsusceptibility (Ci) and of contraception (Cc) (see Bongaarts and Pot- ter, 1983~. The index Ci is defined asS Ci = 20/~18.5 + PPNS). The index of contraception is defined as Cc = 1 - (1.08ue) where u is the proportion of users among women aged 15-49 in a sexual union and e is the method-specific use-effectiveness.6 The two indices Ci and Cc indicate? respectively, what fraction of the total fecundity rate (TF), or the fertility level observed in the absence of these proximate determinants, remains after allowing for the fertility-reduc- ing effects of postpartum nonsusceptibility and contraception (given no change in the other proximate determinants of fertility). Because Ci and Cc gener- ally affect fertility only within sexual union, the total marital fertility rate (TMFR) can be written as TMFR = TF · Ci · Cc or, after taking logarithms to put the equation in an additive form, log TMFR = log TF + log Ci + log Cc. Because Ci and Cc range in value from O to 1, their logarithms have negative values. In the present application, we ignore the value of TF and adopt a ceteris Paribas clause with respect to the total fecundity rate across the regions. It should, however, be noted that levels of infecundity higher than normal have a major negative effect on TF. In the sample of regions used here, the central African zone of high levels of infecundity is not represented. The joint degree of fertility reduction stemming from postpartum nonsusceptibility and contraception can be represented as CiCc or (log Ci + SIf no breastfeeding and postpartum abstinence are practiced, the birth interval equals about 18.5 months, including waiting time to conception, time lost due to fetal loss, and gestation, plus 1.5 months for minimal postpartum amenorrhea. The index Ci is a simple ratio between such a minimal birth interval of 20 months and an interval prolonged by the observed period of postpartum nonsusceptibility. 6In our computations, e is set at 0.60 for the less efficient methods and at 0.97 for the more efficient ones. The coefficient 1.08 represents an adjustment for the fact that some couples do not use contraception if they know or believe that they are infecund.
CONTRIBUTION OF MODERN CONTRACEPTIVE USE TO FERTILITY DECLINE 207 log Cc). Because we are adding two negative values in the latter expres- sion, the degree of fertility reduction increases as the sum reaches larger negative values. The joint effect will be denoted as S: S = 1,OOO(log Ci + log Cc ~ . Alternatively, we also wish to determine to what extent the modern form of contraception is overtaking the child spacing via postpartum nonsusceptibility. We therefore define the difference D as D = 1,OOO(log Ci - log Cc ~ . As D approaches zero, contraception is catching up with the effect of postpartum infecundability, and when D reaches positive values, the fertil- ity-reducing effect of contraception outweighs that of lactational amenorrhea and postpartum abstinence. The plot of 5 versus D is presented in Figure 7-2. The curvilinear -150 -200 - - 250 - _% o O ~300 I_ -350 _ _ - . - C~ ~-400 - ~4 _ o -450 o -SOO -550 ~ - o ° ~ ° o ~ o or/ / o o O ha ret ° °CX i `, ,~ a o o o o Cot o \ o oo ~ o \ ~ o o \ o \o a\ o 11 ~ I I I -400 -300-200 - 100 0 100 200 1000 (log(Ci) - log(Co)) FIGURE 7-2 Relationship between the total degree of fertility reduction (S = verti- cal axis) and the difference between the share of postpartum nonsusceptibility and of contraception (D = horizontal axis)-WFS and DHS regions (r = .86~. SOURCE: Demographic and Health Survey and World Fertility Survey reports.
208 -150 -200 ~_ o -250 C' i_ On ~-300 -400 -450 -500 -550 FA CTORS AFFECTING CONTRACEPTIVE USE ooo o o o o ° f~ o o g 0 0° ° 8 0 o o o o cob o O O O. O O O ~ Go 0 0 \ o o o O 0\ o O. O O O o O O \ O o 1 1 1 1 4 5 6 7 ~9 l O ~ l 2 3 Education: years FIGURE 7-3 Relationship between the joint fertility-reducing effects of postpartum nonsusceptibility and contraception, and the average length of female education- WFS and DHS regions (r = .54~. SOURCE: Demographic and Health Survey and World Fertility Survey reports. relationship clearly indicates that a relatively high degree of fertility reduc- tion can be achieved via the exclusive action of lactational amenorrhea and long periods of postpartum abstinence (regions to the left in the figure). When periods of postpartum infecundability are shortened without adequate compensation through contraception, the overall degree of fertility reduc- tion obviously weakens (regions at the center of the figure). The second phase of the transition is observed in areas where modern contraception increases and produces again higher levels of overall fertility reduction (regions immediately to the left of zero in the figure). In the present sample of regions, there are only nine cases in which the effect of contraception is greater than that of postpartum nonsusceptibility (positive values of D). These regions are located in Zimbabwe (7), Botswana (1), and Kenya (1~. However, a few urban areas are close to this demarcation line: Khartoum, Kampala, and Nairobi. The same also holds for rural Botswana, the remain- der of Zimbabwe, and several other regions in Kenya. The relationships between female education and S and D, respectively, are shown in Figures 7-3 and 7-4. The relationship to the overall degree of fertility reduction (S) is weakest because high degrees of fertility reduction
0F ~ CO~^ ~ ~ ~F ~E ~2 200 iOO -100 -20Q -300 -400 o o o 0 0 0 0 o ~0 - o o ~ O o~ 0 ~ 0 ~o O 0 ~ ~ ~o ~0 ~° ~ w- 0 0 8° 8 0 0 o ° 0 o o o 0 0 o o I I 1 1 ~1 5 6 7 8 9 [ 1 ~1 0 1 2 3 EdUcatior: yeaPs fIGURE 7-4 Relabonship betweeD 1be growing impact of ~rdl~y reduction via conlracept1On over pos~tum DoDsusceptibiDty, and tbe average lengtb of ~male schooUng WFS and DHS regioDs (r = 77~. SOURCE: Demographic and Heal~ Survey and World FertHity Survey reports. can be obt~ned via eitber tbe ~adidonaDy long period of nonsusceptibiDty or modern contraceptioD. Tbo rel~10nship of ~male educ~ion to ~ is ~ronger bec~se 1be two components of ~ wo~ in opposHe ways in rela- tion to ~m~e educ~ioD. Cle~ly, tbe 1mpo~ance of contracephon rel~ive to pos~=um nonsusceptib1Uty 1ncreases w11b e~c~10n. However, it sbould be noted ag~n tb~ tbe sc~ter widens consider~ly ~r reg10ns with average ~male schooling levels of ~ur ye~s or more. Western Ahican regions with be~er scbooling levels bave lower degrees of compens~ion via mod- em coDtraception, ~bereas tbe eastem and soutbem A~C&D regions witb ~mU~ schooliDg levels ~e reaching or crossing tbe break-even poiDt O.e~ = 0~. 1BE [NCER1AIN FO10= Tbe momentum of iDcreased coDtraceptive use and tbe spread of contra- ceptioD to otber regions depend on a number of condidons. First the de- mand ~r children must be dechning as ~ consequence of ~miDisbed child utUity or increased costs of cbildren. Rising aspiraboDs ~itb respect to
210 FACTORS AFFECTING CONTRACEPTIVE USE child quality and rising educational costs could trigger a fertility decline (see Chapters 3 and 4~. Secondly, efficient forms of contraception must be available at affordable prices (see Chapter 5~. If, by contrast, there are a contraction of job opportunities, a breakdown of the schooling system, and a growing scarcity of family planning supplies and services, other scenarios may well develop. During the last decade, many sub-Saharan countries experienced declining foreign earnings, which have led to actual declines in primary school enrollment figures, although there is a continuing increase in the enrollment ratio of women relative to men (see Chapter 39. The engine of the demographic transition process, identified in this chapter as increased female education, has stalled in some regions. If in addition, the availability of family planning supplies and services declines, the original notion of a "crisis-led" fertility transition might be abandoned as couples are not able to meet their fertility goals. CONCLUSION It should be reiterated that the WFS and DIES samples of regions are not representative for the whole of sub-Saharan Africa. As a consequence, the results of the present study should be interpreted with care. The main points derived from the analyses can be summarized as fol- lows: . In the majority of regions studied, fertility is still controlled pre- dominantly through prolonged periods of postpartum nonsusceptibility. Only in a minority of regions in three countries (Zimbabwe, Botswana, Kenya) is there a comparable effect stemming from contraceptive use. · In countries having comparable data over time, there are no traces of a further decline in the components of the nonsusceptible period during the 1980s. Under such circumstances, rises in contraceptive use would carry their full effect. Western African regions still benefit from a considerable postpar- tum abstinence bonus, but continue to score low on contraceptive knowl- edge, small ideal family size, and actual contraceptive use. · Female education is particularly strongly associated with reduced duration of postpartum nonsusceptibility and increased use of contracep- tion. These relationships are, however, curvilinear. As noted in Chapter 6, contraceptive use greater than 10 to 15 percent emerges only in regions with average female schooling durations of four years or better. Further- more, the scatter widens considerably beyond this educational threshold. Among areas with higher levels of female education, there is a marked contrast with respect to contraceptive use between regions in Zimbabwe, .
CONTRIBUTION OF MODERN CONTRACEPTIVE USE TO FERTILITY DECLINE 211 Botswana, and Kenya with high levels of use and other regions in the sample with much lower levels. · Given recent economic reversals and associated curtailments or stag- nation of per capita expenditures on social services, especially education (see Chapter 3), the future is quite uncertain with respect to the prospects for both further increases in female education and improved family plan . . nlng services.
