Other Effects of Aggregate education Summary and Conclusion

7. The Importance of Husbands' and Men's Education

The estimated difference in fertility between women with low and women with high education partly reflects that the former (if married) tend to have a husband with low education and the latter a husband with higher education. Husband's education may have an independent effect on fertility, stemming from two inseparable components: Given a woman's education, a higher education for the husband means that the educational gap between the spouses is wider. Besides, husband's education may have an impact more in its own right. Anyway, the impact of an expansion of women's education from that in the bottom-10 districts to that in the top-10 districts that was calculated above will only be a good prediction of fertility changes if men's educational level rises correspondingly (i. e. so that the distribution of men's education, given that of the woman, is kept constant). In the very hypothetical situation where only women's educational level is enhanced, women will not on average get better-educated men, but (disregarding the possibility that the marriage intensity generally falls) the degree of homogami will change. For example, women with low education will found it more difficult to get a better-educated husband.

In order to establish a good platform for predicting fertility responses to educational expansion, one should estimate independent effects of men's and women's education (at the individual level, as well as the aggregate, as explained below). Lack of couple data is a major obstacle to this. DHS surveys and similar sources typically include information about the husband of currently married women, but not about the education of the husband at previous points of time.

In this study, an impression of the importance of controlling for husband's education is provided by simply restricting the higher-order birth model to a sample of women who were married at interview. It is assumed that they have had the same husband throughout the five-year period of analysis. Women in this sample may well be selected for high fertility, for example because divorce may be partly a result of low fertility, but the effects of education are not necessarily severely biased. That depends on whether the link between education and fertility depends markedly on (changes in) marital status.

Fortunately, when the sample was restricted in this manner, effects of women's individual and aggregate education were about the same as found for the larger sample (compare Model 1, Table 5 with Model 4, Table 1). A control for husband's education turned out to be quite important. When that variable was included in the model, the effect of women's individual education was substantially attenuated (Model 2, Table 5).

(Table 5 about here)

Also the effect of women's individual education on contraceptive use was reduced when husband's education was included (not shown). On the other hand, very little change was seen for fertility desires and post-partum susceptibility, where husband's education was found to play a minor role.

Similarly, the estimates of aggregate education effects reported in Table 1 capture a combined impact of men's and women's educational distribution, which in turn can be interpreted both as `absolute' effects and gap effects, as commented above for the individual level. In Zimbabwean districts where women's education is low, also that of men is low, although not quite as low. In districts where women's education is relatively high, men's education is even higher (but with a somewhat smaller gap between the sexes than in the other districts). The aggregate effect of expanding women's education exclusively will be even less negative than suggested by the estimates in Table 1, unless men's education at the aggregate level actually stimulates fertility or is without importance.

The impact of women's and men's educational distribution partly operates through husband's education as an individual-level effect. Given a woman's education, her chance of marrying a man with a high education will probably increase if many men have high education and decrease if many other women have a high education. According to the model estimated with the Zimbabwean data, the already small aggregate effect of women's education (which also incorporates an effect of men's education) is slightly reduced when an individual variable for husband's education is included.

Ideally, one should estimate independent effects of men's and women's education at the community level, but this is impossible with the available data because of the very strong correlation between men's and women's literacy rates and the lack of published sex-specific data for education at higher levels (which would surely also have been very strongly correlated anyway).

 

Other Effects of Aggregate education Summary and Conclusion

A Search for Aggregate-Level Effects of Education on Fertility, Using Data from Zimbabwe
Øystein Kravdal
© 2000 Max-Planck-Gesellschaft ISSN 1435-9871
http://www.demographic-research.org/Volumes/Vol3/3