SAMPLE SELECTION BIAS IN THE ESTIMATION OF RECREATION DEMAND-FUNCTIONS - AN APPLICATION TO SPORTFISHING

Including the participation decision and performing the appropriate statistical analysis require the use of sample selection models. Most of the discussion of sample selection models focuses on the statistical properties of the estimates. The choice of an appropriate model may be dictated as much by the model's behavioral implications as it is by its statistical properties. The choice is critical when demand functions estimated via sample selection methods are ultimately used to calculate welfare measures. In this paper we explore three methods for correcting for sample selection bias and apply each to a sportfishing demand problem. We choose this example since valuing access to sportfishing and changes in the quality of trips are common non-market valuation questions addressed by policy makers. For each of the sample selection methods, we explore the behavioral implications, estimate the demand model, and compute the resulting benefits of access. We also include the benefit results which would occur from the application of ordinary least squares to the unadjusted individual observations. -from Authors