A general semiparametric approach to inference with marker-dependent hazard rate models

We examine a new general class of hazard rate models for duration data, containing parametric and nonparametric component. Both can be mix time effect possibly time-dependent covariate effects. A number well-known are special cases. In counting process framework, profile likelihood estimator is developed the component model shown to asymptotically normal efficient. Finite sample properties investigated in simulations. The applied investigate long-run relationship between birth weight later-life mortality.