Estimating Fully Parametric Hazard Rate Models with Time-Dependent Covariates

Abstract

This article shows (a) how a widely available algorithm for nonlinear least squares estimation can be accommodated in order to estimate fully parametric hazard rate models by the method of maximum likelihood; (b) how the algorithm allows for a flexible treatment of time-dependent covariates in fully parametric models. An empirical analysis of the duration of jobs illustrates the use of the algorithm. The data are taken from The Norwegian Life History Study for Men. The Appendices discuss and list the algorithm used.