An Information-Theoretic Approach to Estimating Willingness To Pay for River Recreation Site Attributes

This study applies an information theoretic econometric approach in the form of a new maximum likelihood-minimum power divergence (ML-MPD) semi-parametric binary response estimator to analyze dichotomous contingent valuation data. The ML-MPD method estimates the underlying behavioral decision process leading to a person’s willingness to pay for river recreation site attributes. Empirical choice probabilities, willingness to pay measures for recreation site attributes, and marginal effects of changes in some explanatory variables are estimated. For comparison purposes, a Logit model is also implemented. A Wald test of the symmetric logistic distribution underlying the Logit model is rejected at the 0.01 level in favor of the ML-MPD distribution model. Moreover, based on several goodness-of-fit measures we find that the ML-MPD is superior to the Logit model. Our results also demonstrate the potential for substantially overstating the precision of the estimates and associated inferences when the imposition of unknown structural information is not accounted explicitly for in the model. The ML-MPD model provides more intuitively reasonable and defensible results regarding the valuation of river recreation than the Logit model.