Incentive Compatibility and the Consequences When It Is Missing: Experiments with Water Quality Credits Purchase.

This article implemented four treatments to elicit preferences for a nonmarket good, including (1) a hypothetical referendum, (2) a real referendum lacking incentive compatibility, (3) a real choice with incentive compatibility, and (4) a hybrid approach based on (2) and (3). We develop a method to estimate the probability that observed choices do not identify the highest-utility alternative in a choice question. We find that in the hypothetical referendum, the estimated percentage of individuals choosing the alternative that gives the highest utility is the lowest among the treatments. Adding policy consequentiality or payment consequentiality increases the percentage of truthful responses.