When a poor index becomes a good proxy: on the predictive value of individual fertility preferences at the cohort macro-level

Eva Beaujouan, Wittgenstein Centre (IIASA, VID/ÖAW, WU)
Laurent Toulemon, Institut National d'Études Démographiques (INED)

Establishing a link between fertility prospects and aggregate fertility has been a concern for many years, rarely explored in Europe to date. In particular, there appears to be a gap between ideal and actual family size, but little is known about whether this has varied over time. Likewise, whether ideals are more closely related to cohort fertility or to period fertility is a widely discussed question. Finally, highly educated women are generally less likely to reach within-cohort fertility expectations. We checked first whether the same holds true for ideals, for both men and women, and second, whether ideals have the same predictive power across cohorts in the three groups of low, medium and high educated. Using an innovative approach, the correlation between aggregate preferences and actual number of children is explored in a period and in a cohort perspective. Long time-series on fertility preferences are scarce, and we use a very consistent French annual time-series of ideal family size (CREDOC, 1979-2012) in order to precisely model the correlation with cohort and period total fertility rates. There is a persistent gap between ideal family size and fertility indicators. However, in terms of both trends and of year-on-year changes, ideals are not related to the period total fertility rate, but completed fertility and reported ideal family size are strongly linked at the population level, and especially so for men. The gap between ideal family size and cohort fertility is also growing across educational groups for men and women. However, the correlation across cohorts is weaker among lower educated, suggesting that their fertility behaviour is less well predicted by initial ideals than in the other educational groups.