On Cohort Forecasts of First Marriage For U.S. Women

P. C. Roger Cheng, National Central University, Taiwan

This paper develops a simple age-period-cohort framework in completing cohort incomplete first marriage schedules and makes full use of the 1935-1993 U.S. data to obtain robust outcomes. From a theoretical perspective, we show that the period effect represents not only transitory macro shocks that impinge on all cohorts, but also captures the interaction between the age and cohort effects. Empirically, we indicate that the period effect is the key to transforming a marriage level into a marriage schedule. Accompanied by the smoothed version of the adjusted period measure proposed in Kohler and Philipov (2001), we approximate the cohort marriage schedules fairly well and the estimates of all distributional parameters can be thereby obtained. Our approach is easy to implement and the data requirement is relatively light, indicating that the proposed method is readily applicable to countries whose data lengths are not long enough.