Modeling Demand for Recreational Trips of Differing Lengths : A Competing Risks Duration Model with Censoring and Time-Varying Covariates

In this paper we characterize seasonal recreational participation with a model of between-trip durations, a variation on seasonal participation models that provides certain advantages relative to modeling total trip counts directly. Employing a competing risk framework, we are able to model anglers? demands for trips of different lengths. The duration framework also allows us to model the effect on total trips of intra-seasonal variations in explanatory variables such as fish catch rates. Further we develop the model for a context where the available data are limited to detailed information on the most recent trip, supplemented by the return date of the survey. The proposed resolution to the challenges of left- and right-censored duration data employs the ?memoryless? exponential distribution for the duration model, which implies a Poisson count model. With another dataset, we provide an external validation test of the predicted number of angler trips from the model