Short- and Long-Term Unemployment. A Parametric Model with Time-Varying Effects.

In this paper we present the first application to unemployment duration analysis of a mixture distribution model, initially proposed in the biosciences literature by Blackstone, Naftel, and Turner (1986). The model is characterized by the decomposition of the aggregate hazard function into a number of distinct hazard functions. The approach allows us to attribute to each function a different set of covariates as well as coefficients Using U.S. data on displaced workers, we are able successfully to decompose the time varying hazard into two distinct phases - corresponding to short-term and long-term unemployment - and in the process evaluate (and reject) the proportionality assumption. We also compare the results from the model with those obtained from the Cox proportional hazards model and with a parametric hazards model in which a Burr specification is employed for the baseline hazard. Copyright 1995 by Blackwell Publishing Ltd