Optimal Designs for Discrete Choice Contingent Valuation Surveys: Single-Bound, Double-Bound, and Bivariate Models

This paper finds the designs for discrete-choice contingent valuation surveys which maximize the precision of a statistic of interest about the public′s willingness to pay (WTP) for a change in the environmental quality, such as median WTP. An optimization problem is set up, in which the efficiency of the estimate of median WTP (or a related concept) is the objective function to optimize with respect to the survey design (i.e., the set of bids offered to the respondents). In deriving the optimal designs, we consider alternative econometric models, such as the basic "single-bound" model that uses the responses to the first payment question, the traditional interval-data (or "double-bound") model for surveys with a follow-up payment question, and the recently proposed bivariate binary response model. Simulations are carried out to compare the alternative design principles and assess their performances with respect to the statistic of primary interest, median WTP, and other welfare measures, such as mean WTP. Other designs that are constructed ad hoc are also examined.