Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
Competing risks models recognize that there may be more than one exit from a given state, but they make the strong assumption that there is no correlation between unobserved heterogeneity components in each state. Here, a competing risks model that uses a semiparametric method of estimation and controls for the correlation between unobserved heterogeneity components in each state is compared with a traditional competing risks model of exit from retirement. The unobservable heterogeneity components of the competing risks are insignificantly positively correlated and the effects of policy-relevant variables are changed somewhat. Copyright 1989 by MIT Press.