Most of hedge funds databases are now keeping history of dead funds in order to control biases in empirical analysis. It is then possible to use these data for the analysis of hedge funds lifetimes and survivorship. This paper proposes two nonparametric specifications of duration models. First, the single risk model is an alternative to parametric duration models used in the literature. Second, the competing risks model consider the two reasons why hedge funds stop reporting. We apply the two models to hedge funds data and compare our results to the literature. In particular, we show that a cohort effect must be considered. Moreover, the reason of the exit is a crucial information for the analysis of funds' survival as for a large part of disappearing funds, exit cannot be explained by low performance or low level of assets.
TSE Working Paper, n. 10-174, March 2010