Estimating the transition probabilities from censored Markov renewal processes

The problem of estimating the transition probabilities for the Markov chain associated to a Markov renewal process is considered. The estimators are to be based on censored observations of the Markov renewal process. For a class of Markov renewal processes whose transition distributions factor, nonparametric estimators are defined. They are shown to be consistent and to converge weakly go Gaussian random variables. The result builds upon those in Gill (1980) for nonparametric estimation in this setting.