Econometrica: Nov, 1990, Volume 58, Issue 6
The Relationship Between Wages and Income and the Timing and Spacing of Births: Evidence from Swedish Longitudinal Data
James J. Heckman, James R. Walker
This paper estimates semiparametric reduced-form neoclassical models of life-cycle fertility in Sweden. Rising female wages delay times to all conceptions and reduce total conceptions. This result is robust across a variety of empirical specifications. In the best fitting models in which marital status is excluded, male incomes--defined to be zero for unmarried women without a cohabiting male partner--reduce times to conceptions and increase total conceptions. The results on female wages are robust across a variety of empirical specifications, but those on male incomes are not robust to the introduction of marital status and cohabitational status variables, even though models with such variables can be rejected by our model specification tests. We find a particular neoclassical model that predicts fertility attained at different ages as well as the aggregate time series of birth rates. A model that excludes wages and incomes predicts fertility attained at different ages but fails to predict the aggregate time series and is dominated by the neoclassical model in terms of non-nested test criteria. Cohort drift found in estimated parameters is consistent with the expansion of pronatal social programs. The estimated neoclassical model produces strong short-run responses of birth rates to wages and incomes of the sort that have been found in the time-series literature on fertility while generating the relatively weak long-run responses to economic variables found in the cross-sectional literature on completed fertility.