The third birth in Sweden: Fid-Bau Portal

The third birth in Sweden

Heckman, James J. / Walker, James R.

Abstract This paper considers the formulation, estimation and interpretation of microdynamic models of fertility. Our model explains parity choices, sterility, childlessness, interbirth intervals and initiation of pregnancy within a unified framework. We develop a general methodology for estimating the determinants of transition times to births of different orders. Our procedure incorporates time-varying explanatory variables and unobservables. We present conditions that justify conventional formulae relating hazards to survivor functions when time-varying variables enter hazards. We also consider the validity of widely-used piecemeal estimation strategies that focus on one birth at a time. We consider methods for selecting a best model among a class of non-nested models. Two criteria are set forth and used to evaluate the detminants of third births in Sweden. We find that two models fit Swedish microdata equally well. One model is consistent with neoclassical economic theory. It assigns a central role to the wages of men and women in explaining the timing and spacing of births. The other model is purely demographic and excludes wages. Purely statistical criteria cannot distinguish these models although in other work we show that the economic models are more parsimonious in terms of the number of parameters that must be estimated and are better able to forecast aggregate time series. We demonstrate how to interpret the output of multistate fertility models. Wage effects on third births are decomposed into two components: (a) an in direct effect that determines whether a woman is at risk to have a third birth, and (b) a direct effect on the transition rate to the third birth given that a woman has had two births. We find that female wages play an important role in postponing first births but play only a minor role in explaining childlessness. Female wages substantially affect third births. Male wage effects are weaker. We find that female wage effects weaken for more recent cohorts of women. This evidence is consistent with the introduction of progressively more pronatal Swedish policies.