The following is an excerpt from:

M. Omar Rahman and Jane Menken. Forthcoming. Reproductive health in the developing world. In Robert Black and Michael Merson (editors), International Health. Gaithersburg MD: Aspen Publishers, Inc.

In order to understand the different types of fertility transitions that have taken place, we need understanding of the determinants of fertility and fertility change in different contexts. There is an extensive literature that examines the impact of socioeconomic factors on desired family size (c.f. Bankole and Westoff, 1995; Rutstein, 1998, Bulatao and Lee, 1983). Much of the discussion is centered on the costs and benefits of children and the notion that couples desire additional children as long as the benefits are greater than the costs. These benefits and costs are, in turn, determined by a range of factors, some of which are structural (e.g. wages, rates of return on investments, opportunity costs), and some which are attitudinal (changes in values and expectations). Improvements in the educational status of women, for example, are thought to decrease desired family size because they increase the potential wages that women can earn and thus raise the opportunity costs of childbearing and childrearing. Education may, in addition, lead to attitudinal change about quantity-quality tradeoffs in numbers of children, e.g. having fewer children so that greater investment in the education of each child is feasible.

Implicit in this theoretical framework is the idea that couples weigh a variety of alternatives, with childbearing being just one of the possible behavioral choices available. Other structural factors include changes such as increasing landlessness, which decreases the benefits of the labor provided by children and thereby tends to reduce family sizes. More recent research emphasizes attitudinal change. It posits that values and expectations can change as a result of outside influences. Thus exposure to messages in which small families are treated as a marker for modernity may motivate couples to reduce their desired family sizes even in the absence of any changes in the structural costs and benefits of children. This remains a controversial topic.

Although this chapter focuses on developing countries, in almost none of which has fertility declined to replacement level, it is worth noting that the developed world, especially Europe, is concerned about its very low fertility and population decline. For Europe as a whole, TFR declined below 1.9 before 1980 and has continued to decline. It is expected that the entire continent will have a negative growth rate for 2000 (United Nations, 1998). Understanding what maintains below replacement fertility and what causes it to increase is an important issue for the developed world.

How do people control their fertility?

In addition to considering why people control their fertility, we need to understand how people actually do so. It is useful first to consider the proximate determinants that lead to variation in fertility in the absence of deliberate family planning (c.f. Bongaarts, 1978; Bongaarts and Potter, 1983; Sheps and Menken, 1973; Menken and Kuhn, 1996). These proximate determinants can be divided into those that affect the reproductive span and those that influence the intervals between successive births within that span. As shown in Figure 4, the effective reproductive span exists within boundaries set by both the biological and the social reproductive spans. The former, the biological span, is the time during which a woman is capable of childbearing because she has the biological capacity to ovulate and to carry a pregnancy to a live birth. It is usually marked by menarche and menopause, but first ovulation may occur well after menarche and last ovulation precedes menopause. However, in no society do women devote their full biological span to reproduction. Were they to do so, according to Bongaarts (1978), women who survive to sterility would bear over 15 children on average. This figure is well beyond the maximum ever recorded for any population. Every society has social controls on initiation and cessation of sexual activity. We will refer to entry into sexual activity as marriage and cessation as marriage dissolution. We use these terms as social markers rather than representing legal ceremonies and arrangements of the state. Specifically, marriage dissolution can occur through breakup of the relationship or through widowhood. The social reproductive span is, therefore, the interval between initiation and cessation of sexual activity. The effective reproductive span is the overlap of the biological and the social spans. It begins with the later of menarche and marriage and ends with the earliest of sterility, death, and cessation of sexual activity. In many societies, this effective reproductive span is interrupted by time between successive unions or by temporary separation of spouses. Within the effective reproductive span, the pace of childbearing is determined by the lengths of the successive intervals between births (B1, B2, etc. in Figure 4). We will first discuss birth intervals in the absence of deliberate family planning. The birth interval may be divided into several segments:

* the postpartum period after a birth until both ovulation and sexual relations resume;

* the time to conception;

* additional time due to fetal loss through spontaneous abortion; and

* the pregnancy leading to the next live birth.

Fertility in the absence of contraception and induced abortion

The postpartum period ends when both ovulation and intercourse have resumed. It is largely determined by the duration and intensity of breastfeeding and by postpartum taboos against intercourse by a nursing mother. Women who do not breastfeed usually menstruate for the first time about two months after the birth (Salber et al. 1966), whereas frequent, intense breastfeeding can postpone average time of ovulation to over 20 months (c.f. Wood et al., 1985). Some populations, particularly in sub-Saharan Africa, have traditionally had taboos against intercourse that can increase the postpartum period beyond the resumption of ovulation, but these practices are rare outside this region and the observance is believed to be decreasing. Therefore, breastfeeding not only provides the child with nourishment but also, depending on the pattern of suckling, can postpone the return of ovulation for many months. Breastfeeding exerts this effect through a maternal response to suckling that suppresses the secretion of gonadotrophins. The classic studies of McNeilly (c.f. 1996) and his coworkers have shown, for women in Edinburgh, that if the frequency of suckling is maintained above five times a day and the duration is maintained above 65 minutes a day, amenorrhea will often be the consequence. Others have found that nightfeeds are particularly important in maintaining amenorrhea (Jones, 1988). In addition, demographic studies suggest that the duration of lactational amenorrhea increases with the age of the woman (Wood, 1994, p. 355). The effects of breastfeeding patterns are so important that they are the major factor in explaining differences in fertility among populations in which no family planning was practiced.

The time to conception depends upon the monthly probability of conception in the absence of birth control, and can vary among populations, by age, and according to the frequency and pattern of intercourse. The monthly probability of conception, known as fecundability, is extremely difficult to measure, and available estimates differ in part because of the methods used to determine that a conception has occurred. If early fetal loss occurs before the woman is aware she is pregnant, then the estimates of conception are biased downward. According to Wood (1994), the best recent study of early fetal loss is by Wilcox et al. (1988), who followed a group of women aged 20-35 and collected blood and urine samples regularly. There were 198 pregnancies detected by assays of these samples; 43, or about 22 percent, were lost before the woman realized she was pregnant and before clinical diagnosis. Usually, however, fecundability has been measured by accepting a woman's report of her pregnancy. This measure of apparent fecundability has yielded values that range from about .10 to .30 for relatively young women (Menken, 1975; Woods, 1994, p. 289). The waiting time to conception is, on average, the inverse of fecundability, so that this time is, for younger women, 3-10 months on average. In many cases, fecundability has been estimated from reported waiting times to conception using a variety of mathematical models (some of which are summarized by Wood, 1994 pp.281-292).

Fecundability depends in part on a couple's frequency and pattern of intercourse, which determines the likelihood that coitus will occur during the woman's fertile period. Both those wishing to conceive a wanted child and those hoping to avoid pregnancy without the use of hormonal or barrier contraceptives depend on knowledge of the woman's cycle to time their sexual activity and thereby change their probability of conception. Fecundability declines with age, although there is increasing evidence that the decline does not take place until the late 30s, on average, if patterns of intercourse remain unchanged. Lactation has a fertility-reducing effect even after a woman has resumed menstruation and ovulation (c.f. Woods, 1994, pp. 338-343' John et al., 1987). Apparently, continued suckling beyond ovulation reduces fecundability through a response that interferes with the functioning of the corpus luteum. Although fertilization can occur, the corpus luteum may not produce sufficient progesterone for the pregnancy to continue (McNeilly, 1996).

Spontaneous abortion occurs frequently; the rate, as already described, depends upon how early the pregnancy is detected. From the time conceptions are recognizable by virtue of late menses, about 24 percent end in spontaneous abortion (French and Bierman, 1962). Even higher proportions of fertilized ovum do not lead to live births. Wilcox et al. (1988), in the study cited earlier in which urine specimens were collected from which early pregnancy could be detected, found that 31% of the pregnancies ended in fetal loss. Rates of spontaneous abortion increase with age of the woman (c.f. Wood, 1994, p. 250). The time added to the birth interval by a recognized fetal loss is the sum of the time from pregnancy to the next ovulation (usually estimated to be just over 3 months on average, since the vast majority of spontaneous abortions occur very early in pregnancy) plus the time to the next conception. There is little evidence of much variation among populations in the rates of spontaneous abortion.

Gestation leading to a live birth does not vary much, usually lasting between 35 and 40 weeks, with little variation between population groups (Wood, 1994, p. 207). In some populations with outstanding care of premature infants, gestation may be somewhat shorter on average, but we know of no good data demonstrating a decline in the average duration of gestation.

A crosscutting issue is that of infertility and sterility. There is concern, especially in parts of Africa, that infertility and early sterility are affecting fertility. The effect may be through absolute sterility that causes the effective reproductive span to end early, or through decreased fecundability and/or increased risk of spontaneous abortion (c.f. Larsen, 1993, 1995, 1996). In most cases, however, the concern is with early sterility, much of which is believed to be due to sexually transmitted infections.

Thus, the main reasons that populations not practicing family planning vary in fertility are differences in the effective reproductive span, the postpartum period, and the duration of the postpartum period, with variation in time to conception playing a lesser role.

Deliberate control of fertility

People can deliberately reduce their fertility 1) by reducing the effective reproductive span through postponement of marriage or interrupted marriage or by sterilization that ends reproductive capacity early, 2) by use of contraception (which increases the time to conception); and 3) by induced abortion (which increases the time added to the birth interval by pregnancies that do not lead to a live birth). Family planning programs can promote both the motivation to reduce fertility and the means to do so. Many governments encourage or enforce later marriage explicitly through changes in the legal age of marriage and implicitly through programs that foster female education. Family planning programs have traditionally focused on education regarding methods of fertility control, motivation to reduce the number of wanted children, and provision of family planning methods themselves. These methods include 1) promotion of breastfeeding, both for the health of the infant and to prolong the postpartum period; 2) contraception, intended to prolong the time to conception; and, except where there is opposition for religious reasons, 3) abortion, which increases the time added to the birth interval.

Although specific family planning methods will be discussed in greater detail in the section on family planning programs, it seems appropriate to consider two important general issues here. 1) Why is it that populations in which the desired number of children is low still have high proportions of unintended births? And 2) why is reliance on abortion an inefficient approach to family planning?

Contraception, even when highly effective, may still not prevent all unintended pregnancies. A simple calculation makes this problem clear. Suppose a woman is using a highly effective method, one that reduces her probability of conceiving to about 1/1000 per month. She begins using it at age 30 and wants no more children before she reaches menopause at age 45. We can calculate the probability that she has no pregnancy in each of 13 lunar months over the next 15 years - or 195 lunar months. The probability of succeeding (not getting pregnant) is .999 each month. The probability of not getting pregnant in 195 months is .999.195, which equals .38. In other words, she only has a 38% chance of avoiding pregnancy for 15 years. Among women like her, 62% will have at least one unintended pregnancy in that time period. For this reason, even women who are very serious users of contraception are at high risk of unintended pregnancy. Family planning program and health planners, in developing their programs, usually emphasize unmet need for contraception or problems in the use of various methods. They, however, also need to be aware that even perfectly used contraception with very low monthly or annual failure rates carries with it this high long-term risk of unintended pregnancy.

The second point concerns induced abortion. In a population that relies primarily on induced abortion to reduce fertility, if a woman becomes pregnant unintentionally, she may choose to have an abortion. About three months after conception, she again begins ovulating and is capable of conceiving. Suppose her time to conception averages 10 months. Then 13 months after the first abortion, she is again pregnant and must have another abortion if she is not to have an unwanted birth. And 13 months later, again an abortion, and so on. Preventing a birth for 15 years may require that many abortions. It is, therefore, not surprising that many women in Eastern European countries that relied primarily on abortion reported numbers of abortions in the double digits. Women in the former Soviet Union, for example, are believed to have 6 or more abortions on average over the course of their lifetime (David, 1992). Abortion, is, however, extremely effective as a backup to effective contraception. A woman who has an abortion and subsequently uses extremely effective contraception is unlikely to have more than 1 or 2 unintended pregnancies -- but our previous analysis shows that she may have these one or two.

For these reasons, it is not surprising that sterilization, the one method that has a failure rate near zero, is so widely selected by women and couples who want no additional children. In the United States in 1995, it was the modal method selected by couples where the woman was over 30 (Weeks, 1999, p. 171).

The effect on fertility of the proximate determinants: Bongaarts' indices

Bongaarts (1978) developed a set of indices to measure the effect on fertility of some of these proximate determinants. Because he is in the process of revising them indices (Bongaarts, 1999), the derivation of these indices is not presented here. They are based on the assumption that there is some maximum potential fertility, TF, for women. This figure is usually estimated to be just over 15 children.

Ci The index of postpartum infecundity varies from 0 to 1. It represents the proportion of potential fertility, TF, remaining when the average postpartum period of the population of interest is taken into account. Therefore, Ci = 1 if the population does not breastfeed at all. The fertility reducing effect of postpartum infecundity is (1-Ci).

CA The index of abortion is the proportion of TF, after postpartum infecundity is first taken into account, remaining when the effect of induced abortion in reducing live births is taken into account. Spontaneous abortions are included in the original estimate of TF, since they are treated as a purely biological occurrence. Few countries have sufficient information available on abortion to make reasonable estimates of CA, so it usually must be disregarding in application.

CC The index of contraception is the proportion of TF, after the effects of postpartum infecundity and induced abortion are taken into account, remaining after contraceptive use is considered; and, finally,

Cm The index of marriage is the proportion of TF, after the first three factors are considered, remaining when the particular marriage or sexual union pattern is taken into account.

Thus, in the Bongaarts decomposition of the Total Fertility Rate,

TFR = TF x Ci x CA x CC x Cm.

It should be noted that both CC and Cm contain adjustments for infertility and/or sterility. In the first, the adjustment takes into account infertility sterility and assumes no use of contraception by infertile and sterile couples. In the second, there is a weighting factor, in that non-marriage has a greater effect on fertility reduction when the woman is young, e.g., the effect of non-marriage is much greater for a 25 year-old than a 42 year-old.

Table 2 presents these indices, except for CA, which is assumed to be one because of lack of data, for a number of populations around 1970 and for several historical populations (Bongaarts and Potter, 1983). The major impact of breastfeeding can be seen through the values of the index Ci in countries that, at the period in question, used little contraception. All of the South and East Asian countries, as well as Kenya, have indices that do not exceed 0.67; thus, their potential fertility is reduced by at least a third by long postpartum periods. In fact, the long breastfeeding practiced in Bangladesh and Indonesia at that period reduced their fertility to only about half its potential. In Europe, the demographic transition to low levels of fertility was caused, to a great extent, by very late marriage and a relatively high degree of non-marriage. The index of marriage is far lower, on average, for the developed countries around 1970 than for the developing countries. But two historical populations shown had indexes of marriage under .45, indicating that non-marriage reduced their potential fertility by at least 55%. By 1970, fertility in all the developed countries shown was reduced by contraceptive use to no more 30% of its potential level.

Table 2. Estimates of Total Fertility Rate and Bongaarts' Proximate Determinants Indices*

Source: Estimates based on Bongaarts and Potter, 1983, Table 4.2

Note: * Each index represents the proportion of potential fertility remaining after the particular factor is taken into account in the order: postpartum infecundity, marriage, contraceptive use.

Thus, in developed countries, breastfeeding has little effect on fertility, but non-marriage and use of contraception reduce the TFR to relatively low levels. In the 1970's, contraceptive use had little impact on fertility in the developing countries included in this table; lower fertility was achieved in some of these countries through long-term breastfeeding. Populations that had very high TFRs achieved them through a combination of high indexes of marriage and breastfeeding and little or no contraception.