Estimating Recreation Demand with On-site Data: An Application of Truncated and Endogenously Stratified Count Data Models

Abstract

In travel cost models of recreation demand the dependent variable is typically the count of trips taken over the year, and data based on on-site surveys are often used. The appropriate estimator must take into account the fact that the dependent variable is a nonnegative integer from a truncated, endogenously stratified sample and that real data frequently exhibit overdispersion. In this paper truncated count data models are employed to estimate recreation demand and benefits per trip using on-site data from three adjacent forest recreation sites near Helsinki, Finland. As the Poisson model was rejected due to overdispersion in the data, the paperfocuses on truncated negative binomial models with special emphasis on endogenous stratification. Resulting in somewhat better fit and smaller standard errors, the truncated and endogenously stratified Negbin model slightly outperformed the respective non-stratified model. However, adjusting for endogenous stratification had little effect on the estimated coefficients and related benefit estimates. In addition to the basic model, a specification with site specific price slopes is presented with average as well as site specific estimates for consumer surplus per predicted trip.