Tackling the endogeneity of fertility in the study of women's employment in developing countries: Alternative estimation strategies using data from urban Brazil

Opinions differ about whether family structure, especially fertility, should be considered endogenous in models of behavior in developing countries. Faced with a dearth of good instruments, mainstream researchers often urge working in reduced form and, therefore, losing variables of policy interest or limiting the type of questions they ask to those where good instruments are available. Rather than treating endogeneity as a yes or no characteristic, we suggest instead that researchers consider the likely magnitude of endogeneity bias before moving to reduced form. Facing a situation where endogeneity bias is often presented as a concern but where we expect little endogeneity bias, we tackle endogeneity using multiple econometric techniques not available to the average researcher. We find support for our hypothesis that little bias arises due to the assumption of exogeneity of recent fertility in a model of women's employment.