Fertility and Female Employment: Problems of Causal Direction

Seemingly contradictory results have been reported recently on the causal relationship between fertility and female employment. Data from the Panel Study of Income Dynamics are used to investigate four possible explanations of the contradictions: (1) multicollinearity in nonrecursive models; (2) misspecification of models; (3) discrepancies between attitudes or intentions and behavior; and (4) differences between static and dynamic models. All four explanations are pertinent to the controversy. It is found that the dominant effects are from fertility to employment in the short run and from employment to fertility in the long run. Multicollinearity often is a serious problem in nonrecursive models. Dynamic models are much more informative than static models. Plans generally seem to be consequences more than determinants of behavior. Results (sometimes) depend upon how variables are measured, upon what time intervals are used, and upon aggregate trends and cycles and historical circumstances. Other substantive findings and general methodological issues are discussed briefly.