摘要:I present an alternative approach for estimating hazard models that include flexible baseline hazards and nonparametric unobserved heterogeneity, which avoids the computational difficulties encountered by other researchers who have used this sort of model specification. My method places restrictions on the differences of the parameters in a flexible baseline hazard specification, which permits information from other time periods to be used to estimate the parameters of a specific time period. This method is illustrated with an empirical example.