Kathryn Smith

In an open-ended contingent valuation (CV) survey, a respondent can state any non-negative valuation for a given good. As the set of possible responses is infinite, the GibbardSatterthwaite theorem implies that such a survey cannot be incentive-compatible. That is, participants in an open-ended CV survey have an incentive to increase their utility by misrepresenting their preferences. This naturally raises the question of whether the misrepresentation predicted by theory occurs in practice. In order to answer this question we must be able to test CV surveys for the presence of strategic bias. There are two possible tests. Both are crucially dependent on correctly predicting how strategic behaviour alters agents’ responses. If our descriptions of strategic behaviour are incorrect we may fail to notice its presence. Although producing correct descriptions of strategic behaviour may seem trivial, existing models of strategic behaviour seem not to have incorporated important features of the CV process. In particular, existing models omit the ‘screening’ of responses to eliminate those which deemed too low or too high. They also neglect the fact that participation in a survey cannot be compelled. We present two models of the CV process, one which includes the screening and another that includes the participation decision. We find that incorporating screening alters the nature of strategic behaviour. In particular, it is no longer the case that strategic behaviour consists of responses that are simply ‘too low’ or ‘too high’. Rather, agents ‘cluster’ according to their type, and strategic bias affects both the mean and the variance of responses. We also find that there is reason to doubt the claim that CV surveys will induce universal participation.