Combining Averting Behavior and Contingent Valuation Data: An Application to Drinking Water Treatment

This paper is an empirical application that combines averting behavior with contingent valuation data. Consistency tests are performed incorporating alternative heteroscedastic structures in the bivariate probit models by taking advantage of the different information content that characterizes each data source. We look at three covariates not yet examined in the literature when combining stated and revealed preferred data to explain the variance in the models: income, the bid in the contingent valuation questionnaire, and the distance between the bid and the averting expenditures with drinking water. The models estimated include between and within data sources heteroscedasticity. The results obtained allow the combination of the two data sources under a common preference structure.