Testing Welfare Measurement Gains of Combining Stated and Revealed Preferences Methods Using Count Data Models

This research updates the joint estimation of revealed and stated preference data of Cameron (1992) to allow for joint estimation of the Travel Cost Method (TCM) portion using count data models. Further these count data models reflect correction for truncation and endogenous stratification associated with commonly used on-site recreation sampling. Our updated modeling framework also allows for testing of consistency of behavior between revealed and stated preference data rather than imposing it. Our empirical example is river recreation visitors to the Caribbean National Forest in Puerto Rico. For this data set we find consistency between revealed preference and stated preference data. We also find little gain in estimation efficiency in our data. This may be due to our contingent valuation question eliciting willingness to pay for existing site conditions, a benefit measure conceptually very similar to what is estimated with TCM. However, our updated joint estimation may make a significant improvement in estimation efficiency when the contingent valuation scenarios involve major changes in site quality not reflected in the TCM data.