Identification and Estimation of Additive competing Risks Models with Unknown Transformation of Latent Failure Times

Abstract

This paper presents the methods of identification and estimation of additive competing risks models with unknown transformation of latent failure times. In our set up, we assume that the latent failure times are generated by nonparametric additive separable transformation regression model. The model in this paper includes a competing risk version of log-linear model, mixed proportional hazard model, accelerated failure times model, and linear transformation model. Identification of unknown additive function is accomplished using marginal integration method. Our identification strategy does not depend on identification near zero, and it does not require exclusion restriction. Given our identification results, we developed uniform consistent sample analogue estimator.