Fixed or moving targets? Consistency of desired number of children within cohorts across surveys in DHS and predicting fertility changes

Bruno D. Schoumaker, Université Catholique de Louvain

Data on fertility preferences (e.g. desired number of children) were first collected in the 1940s in the US, and have since then been collected routinely in most demographic surveys. A primary objective for collecting data on fertility preferences was to help forecasting fertility changes. However, there is no consensus on the utility of data on fertility preferences for projecting fertility changes. In this paper, our objectives are threefold. The first objective is to evaluate, at the aggregate level, whether fertility preferences of cohorts are fixed targets or moving targets. If fertility preferences are a fixed target within cohorts (stable with age), they provide a potential basis for projecting fertility changes 10-15 years ahead. While strong arguments exist for the moving target model, to our knowledge no empirical test of the fixed vs. moving target model in a wide range of countries has been performed at the aggregate cohort level . Growing evidence for a moving target model has been found over the past years at the micro level; Yet, consistency of preferences at the individual level is not the main issue for the analysis of demographic changes and projections that rely on aggregate measures. The second objective is to document the relationships between changes in fertility and changes aggregate fertility preferences. In other words, we evaluate empirically if fertility changes mirror changes in fertility preferences. Most existing research on the links between aggregate fertility and fertility preferences uses cross-sectional data; we use date on changes in a wide range of countries to document the relationships over time. The third and more speculative objective of this paper is to discuss how aggregate changes in preferences can be incorporated in projections of fertility changes.