This paper considers a panel duration model that has a proportional hazards specificationwith fixed effects. The paper shows how to estimate the baseline and integratedbaseline hazard functions without assuming that they belong to known, finitedimensionalfamilies of functions. Existing estimators assume that the baseline hazardfunction belongs to a known parametric family. Therefore, the estimators presented hereare more general than existing ones. This paper also presents a method for estimatingthe parametric part of the proportional hazards model with dependent right censoring,under which the partial likelihood estimator is inconsistent. The paper presents someMonte Carlo evidence on the small sample performance of the new estimators.
Centre for microdata methods and practice