Mitigating hypothetical bias

Abstract

The overestimation of willingness-to-pay (WTP) in hypothetical responses is a well-known finding in the literature. Various techniques have been proposed to remove or, at least, reduce this bias. Using about 30,000 responses on WTP for a variety of power mixes from a panel of 6,500 German households and the fixed-effects estimator to control for unobserved heterogeneity, this article simultaneously explores the effects of two common ex-ante approaches – cheap talk and consequential script – and the ex-post certainty approach to calibrating hypothetical WTP responses. Based on a switching regression model that accounts for the potential endogeneity of respondent certainty, we find evidence for a lower WTP among those respondents who classify themselves as definitely certain about their answers. Although neither cheap talk nor the consequential-script corrective reduce WTP estimates, receiving either of these scripts increases the probability that respondents indicate definite certainty about their WTP bids.