The Effect of Protest Votes on the Estimates of Willingness to Pay for Use Values of Recreational Sites

Contingent Valuation studies are often characterized by a considerable amount of protest responses, which may have an important effect on the final estimates if the protest responses are not randomly distributed across the sample. If the standard procedure of censoring protest responses is adopted, the estimates may be biased. Sample selection models can detect and –if necessary– correct selectivity bias. We apply a sample selection model to data on valuation of forest resources for recreational use, where WTP responses are obtained through a mixed dichotomous choice-open ended elicitation method. Dealing with continuous data for WTP allows us to apply the Heckman 2-steps method, and compare it to the full ML estimator. Either method has its own drawback: computational complexity for the ML method, susceptibility to collinearity problems for the 2-steps method. The latter is observed in our model. The results show that censoring protest responses in this study would lead to overestimates of the willingness to pay.