The Importance of Husbands' and Men's Education Acknowledgements

8. Summary and Conclusion

The fact that a woman's education influences her fertility, and usually negatively, is firmly established, although we still know little about the exact size of the effect in different settings. An additional fertility-depressing effect of the general educational level in the community, net of its aggregate determinants, is certainly not intuitively implausible either. However, such a spill-over effect from educational investments has yet to be well demonstrated empirically.

In this study, a significant negative effect of aggregate education was found for contraceptive use, just as reported by Amin et al. [Amin, Diamond and Steel 1996] in a study from Bangladesh. This reflects in particular the weaker fertility desires among women in districts with many literate women and a high average education, than among women at the same educational level in other districts. Only a negligible counteracting effect was estimated for post-partum susceptibility.

Nevertheless, actual birth rates the five years before interview were found to be unaffected by other persons' education (with the same control variables included in the models). Significant negative effects appeared only in birth rate models where the urban/rural character of the current place of residence was not included. (With the model specifications chosen in this study, it was easy to get an impression of the relative importance of aggregate education, and it was indeed large according to these simplest birth rate models.) This illustrates the weakness of some previous studies, where negative effects of aggregate education were found, but without controls for urban/rural differences. In fact, the present study has also demonstrated that the choice of urban/rural control variable can be quite critical.

The educational distribution is, of course, closely linked with a wide range of structural factors other than urbanization. When the intention is to assess the total effect of educational expansion, one should include in the models measurements of those of these structural factors that are determinants rather than consequences of education. However, the direction of causality is inherently blurred. Had it, for example, been possible to control for women's general position compared to men, which may well differ across the country, the effect of aggregate education on fertility desires might have been attenuated (or become completely negligible, implying that, in districts with more egalitarian ideas about gender roles, people would have wanted few children regardless of education). On the other hand, the educational distribution is likely to feed back on women's position, in the long run, so the total impact might be more clearly negative than suggested by estimates from models with such a variable included.

Given the lack of access to structural control variables, the 0-effect estimate from the birth rate models is a comfortable result. The true effect of aggregate education, net of urbanization, cannot be negative unless there are important determinants of education that have a clear fertility-stimulating effect, which is quite unlikely. A negative effect of aggregate education (at a given individual educational level) only seems reasonable to the extent that investments in education in a region contribute to stimulate the urbanization of that region, which must be a very slow process.

The fact that a significantly negative aggregate effect failed to show up in these Zimbabwean fertility models does not mean that such an effect is generally non-existent. One cannot, of course, reject that possibility that richer regional data for Zimbabwe might have revealed effects. It is also possible that stronger effects would have been found in countries with much lower literacy rates. According to Caldwell, it is the impact of introducing compulsory education in a society where almost everyone is illiterate that is most pronounced. However, most countries are far beyond this level now, so a similar analysis of such a change would be difficult to make. Besides, this study has cast some doubt on the idea that breadth of education is more important than depth.

No significant interaction effects between individual and aggregate education appeared in the Zimbabwean data. The point estimates did not even display a systematic pattern, but weakly suggested that the negative effect of aggregate education that was found for fertility desires may be confined to women at a low educational level. Changing the perspective slightly, this lack of a clear interaction pattern for birth rates implies, of course, that individual effects of education are independent of the aggregate level. This is a noteworthy conclusion in light of the currently strong scholarly interest in context dependence.

It is hard to believe that aggregate education effects are of any substantial importance, and thus need to be taken into account when predicting the fertility response to investments in education, before they are seen in at least one country in a statistically solid study, and with a focus on births rather than proximate determinants. Such studies should be based on aggregate data for a large number of regional units at a level no higher than the districts used here, and preferably for different points in time. These data should, of course, include some contextual variables that have an influence on aggregate education and are potentially important for fertility, although the strong correlations and ambiguous causality will always make it difficult to identify a pure total education effect. With richer data, it might also be possible to check whether there are thresholds, saturation points, other non-linear effects, or even non-monotonic effects of aggregate education. Besides, they might allow a search for more complex interaction patterns (guided by theoretical arguments) and the separation of effects of men's and women's education at the aggregate level.


The Importance of Husbands' and Men's Education Acknowledgements

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