Interpreting demographic effects in duration analyses of first birth intervals

Estimated demographic effects in proportional hazard models of first birth intervals could reflect time-invariant differences in the risk of a birth, or differences in the timing of a shift in the risk, or both. This paper attempts to distinguish between these possibilities. The procedure is to estimate a more general model than the proportional hazard specification, in which the evolution of the risk of a birth can differ with demographic characteristics. The proportional hazard specification is nested within this more general model. Consequently, the consistency of the data with the risk or the timing interpretation of demographic effects can be tested. The data studied do not lead to a rejection of the proportional hazard specification.Department of Economics, University of Pennsylvania. The initial stage of this research was supported by a National Science Foundation Graduate Fellowship. I have benefited from insightful comments from David Bloom, Andrew Foster, Zvi Griliches, V. Joseph Hotz, Duncan Thomas, anonymous referees, and participants in seminars at Harvard University and the University of Pennsylvania, and in the Economic Demography sessions of the 1988 Population Association of America annual meetings. Any remaining errors are my own.