Introduction Data and Methods

2. Theoretical Considerations

2.1 A Brief Review of Suggested Individual-Level Effects of Education

There are several plausible reasons why women with, for example, some secondary education usually display a lower fertility than the uneducated. To summarize very briefly, and without pretending to produce a complete list of mechanisms, fertility desires have been thought to be influenced by the individual woman's education because of: i) the high opportunity costs of childbearing involved in some types of work that may be offered to the better-educated women, ii) the cash expenses and children's reduced contribution to domestic and agricultural work as a result of children's schooling, which tends to be encouraged by educated mothers, iii) the reduced need for children as old age security, or to support the woman even as a relatively young widow, when the family's wealth allows other kinds of savings, or when the woman is able to earn a living on her own and even set something of that aside for the future, iv) the higher prevalence of nucleated families, which may reduce fertility partly because childbearing costs to a larger extent must be covered by parents, v) a stronger desire to spend more time caring for a child and to invest more in each child, not only in terms of education, vi) stronger preferences for consumer goods or other sources of satisfaction, and vii) a lower infant and child mortality, influencing desires through replacement and insurance effects. These fertility-inhibiting effects may be set off against viii) a possibly stimulating impact of a higher purchasing power resulting from educated women's own work or their marriage into a relatively rich family [Note 1].

Mortality has a bearing also on the supply side. Besides supply and regulation are likely to be influenced by education, one way or the other, because of ix) the higher age at marriage among the better educated, x) their knowledge about and accept for modern contraception, and their ability to use it efficiently, as well as their more efficient use of traditional methods because of better knowledge about their own body, xi) the erosion of traditional norms about post-partum sexual abstinence and breastfeeding that is supposed to go hand in hand with education, and xii) their higher fecundity because of better health or treatment for sexually transmitted diseases. As widely known, the fertility-stimulating effects have actually been dominant at a moderate educational level in many countries, and in particular in Africa during the 1970s and early 1980s.

One reason why education may operate through these channels is that schooling generally (disregarding variations in curriculum, resources and teacher attitudes) makes the women able to read and write, increases their knowledge about the outside world, and provides them with certain practical and theoretical skills that enhance their productivity [Note 2]. In addition, women's position relative to men may be involved. While their `economic autonomy', `physical autonomy' and `decision-making autonomy' (using terms from [Jejeebhoy 1995]) are likely to depend to a large extent on community norms and rules and institutional structures, there may also be individual variations determined by individual factors, such as education and each family's interpretation of the cultural proscriptions (see e.g. [Niraula and Morgan 1996]). In other words, her own current position compared to men, and that she expects for the future and for her daughters, are probably determined both by gender attitudes and structures in society and such individual characteristics as her own education. If she has an education, she may, for example, be allowed by the family to work outside the house and more often be heard in discussions with husband or in-laws [Note 3]. This will add to the effect of her literacy and skills and possibly reduce fertility desires through such factors as opportunity costs, old age security concerns and child mortality.

Women's status may also operate through channels other than those listed above. For example, in situations where the husband wants more children than the wife, a strengthening of women's decision-making autonomy will reduce fertility [Note 4]. Besides, when a wife is considered more of an equal than a subordinate, the couple may communicate better about contraception, and the husband may see more clearly how childbearing burdens the wife. On the other hand, such a relationship may stimulate sexual activity. As a third example, women who themselves have a relatively inferior position relative to men may not only consider the childbearing costs generally low and the rewards high, but expect boys to be even more useful than girls [Note 5]. This will enhance fertility desires in settings where fertility is not extremely high.

While such status effects seem theoretically plausible, however, the quantitative empirical evidence is still quite weak.

2.2 The Possible Importance of Others' Education

Several of the causally intermediate factors mentioned above may depend not only on the woman's own education, but also on that of other women. As pointed out by, for example, Montgomery and Casterline [Montgomery and Casterline 1996], other women may exert an effect because of social learning, social influence and more indirect mechanisms. The individual woman may learn directly from others by communication and observation. It is not only factual knowledge that is likely to be transmitted, but also attitudes as well as understanding of possible consequences of different actions. Bongaarts and Watkins [Bongaarts and Watkins 1996] have stressed that this learning may include interpretation in light of current local conditions and the individual situation. There may also be a more passive imitation of behaviour (`social influence') without any (active digestion of) new knowledge, driven by a desire to attract other people's approval. A more indirect mechanism is that others' ideas, resources or behaviour can influence society and social institutions and thereby also behaviour more generally.

In principle, these influential `others' may be close neighbours, other women in the village or city, or even women in other parts of the country, but the district level has been in focus of this study, because of lack of adequate data at a lower level. Such aggregate effects may be positive as well as negative, and may differ between women who themselves have little education and those who are better-educated.

To be more specific, one possibility is that uneducated women may have more knowledge of contraception and more modern views about its acceptability if they live in a society where many women have attended school for some years than if they live elsewhere. They may also simply tend to imitate the more widespread use of contraception among the better-educated. Moreover, their preference structure and their practice of breastfeeding and post-partum abstinence may be influenced by aggregate education.

Other people's education may be of importance also to those who themselves have more education, although for slightly different reasons. A diffusion of factual knowledge of contraception is, for example, less relevant, but there may be a more efficient interpretation of the ideas and attitudes the better-educated have been exposed to at school, through reading or because of interaction with people elsewhere if there are more women to share these experiences with.

An argument along similar lines has been advanced by Caldwell [Caldwell 1980], who suggested that `mass education' will make everyone more conscious about the need to educate their children, and that it will strengthen Western middle class values, with more emphasis on individual rights than on the duty to a larger family network. His hypothesis is that an increase in the proportion who are literate, from a very low level, is the most crucial change, whereas the average length of education among the literate is of less importance. Apparently, he did not consider the differentials in such macro effects, but it seems most plausible that the (large majority of) uneducated would be most influenced.

Moreover, opportunity costs of childbearing may depend on other people's education. Generally, having a high education will increase a woman's chance of finding a relatively well-paid job (where she cannot bring her children with her). However, under the assumption of a fixed supply of such jobs, a high proportion of well-educated women in the community will decrease her chance. To elaborate on this, even women with a quite short education may be able to get into this niche in the labour market when few have an education, whereas their situation would be little different from that of the uneducated when the average education is higher.

In addition to its influence through such a competition effect, aggregate education is linked to the supply side. When more women are educated, the attitude towards women's work in the modern sector is likely to change, and more jobs that are attractive to and suitable for these women may be created. Because of the inertia of these processes, the availability and acceptability of jobs presumably depend more on aggregate education some years before than on the current situation. Better-educated women who are forerunners, in the sense that few had such high education in the immediate past, may find it particularly difficult to make use of their schooling [Note 6]. Unfortunately, changes over time in aggregate education cannot be measured with the data available in this study.

In other words, if the temporal dimension is disregarded for simplicity, it can be concluded that a high average education may reduce the opportunity costs because of a competition effect and add to them because of increased supply and acceptance of well-paid jobs. To complicate this further, opportunity costs may be substituted by direct costs by purchasing child care, which is likely to be particularly cheap when only a few women work away from home.

Of course, if a general rise in women's education (with or without an accompanying rise for men, which is addressed below) contributes to undermine old ideas about women's rights and obligations compared to men, it is not only the opportunity costs of childbearing that will be enhanced. An improvement of women's status as a contextual phenomenon, and the changes in women's individual position that follow in the wake of this, may influence fertility for many other reasons, a few of which were reviewed above. This may influence also the fertility of women who themselves have little education.

Besides, broader economic transformations are likely to take place as a result of a better-educated work force, in the long run. In addition to the increase in the number of well-paid jobs in the modern sector, as referred to above, it is, for example, plausible that the agriculture will be generally modernized, including that among uneducated farmers. This will undermine children's relative importance in the fields, and thus the incentive for childbearing. Moreover, higher productivity in agriculture, combined with a general knowledge level that may facilitate the establishing of manufacturing industries that benefit particularly strongly from economies of scale, may lead to an increased concentration of people in urban areas. In other words, it can be argued that investments in education in a district may contribute to stimulate the urbanization of that district, although very slowly. Also the generally higher wealth that is likely to be a part of these transformations may have an effect on fertility, although not necessarily a positive one.

As yet, there is no clear empirical evidence about interactions between individual and aggregate education. Jejeebhoy [Jejeebhoy 1995] concluded that, in countries where women's literacy is high, primary education is more likely to push fertility down, and the negative effect of secondary education is particularly sharp. This is, of course, the same as claiming that the literacy rate has the clearest negative effect among the better-educated women. Her conclusion was based on a review of a number of studies with different, and more or less relevant, control variables included, and its conclusion did not have a very solid basis. She found, for example, that fertility desires responded most sharply to secondary education in the countries with a generally low educational level, and that education was least powerful in weakening the breastfeeding norm, which would contribute to a particularly strong negative effect on fertility, in these settings. No clear picture has emerged from the very few original multilevel studies either [Kravdal 2000, Lesthaeghe et al 1985, Tienda, Diaz and Smith 1985].

2.3 Determinants of Education

In order to find out how investments in a woman's education will influence her fertility, one must, of course, estimate models that include various individual and structural factors that both are determinants of education and, for other reasons, influence fertility. For example, the individual woman's education is likely to be determined by resources and attitudes in her family of origin as well as society's needs for and ability to finance education and the general attitudes towards women's schooling. These aggregate factors, which may well differ across the country, are of course also determinants of men's and other women's education.

More specifically, people's educational level probably depends strongly on whether they live in urban or rural areas. The more advanced economy in cities both requires and allows investments in education, and returns to investments may be high because of the higher population density. This calls for the inclusion of an urban/rural variable in the models, which the Zimbabwean data allowed. (According to the 1994 ZDHS, 29 per cent lived in urban areas). The fact that there may also be a reverse relationship of presumably less importance, with aggregate education fuelling urbanization in the long run, is returned to below.

The data include no relevant information about the family of origin, but the woman's own religion can be supposed to be a stable, often socially inherited, characteristic that has a bearing on decisions related to her education (although the opposite causality cannot be completely excluded). (55 per cent of the Zimbabwean population are Christian, according to the 1994 ZDHS).

Unfortunately, it was not possible to control for the general attitude towards women or the economic wealth in the districts. The implications of this are also discussed below.

Given the selection mechanisms briefly reviewed here, an additional argument for an interaction pattern seems plausible: If particularly sharp negative effects of a high education are seen in societies with very low average levels, it might reflect that the few who have broken the barriers under these conditions have parents who are far more wealthy than average or hold very different values.


Introduction Data and Methods

A Search for Aggregate-Level Effects of Education on Fertility, Using Data from Zimbabwe
Øystein Kravdal
© 2000 Max-Planck-Gesellschaft ISSN 1435-9871