Revisiting Opt-Out Responses in Contingent Valuation

This paper revisits the topic of opt-out responses in contingent valuation from a fresh perspective. We first acknowledge the probabilistic nature of referendum-style contingent valuation and set up the expected utility framework. Within this framework, we theoretically find that respondents who are indifferent between the possible outcomes and/or who perceive a survey as inconsequential choose to opt out. Using survey data, we empirically find that respondents who perceive the survey as inconsequential are more likely to opt out than to choose yes or no. Also, we empirically test whether opt-out responses are more similar to yes or no votes and examine the consequences of discarding opt-out responses in terms of parameter estimates, sample means, and welfare estimates. We present empirical tests that can be used as criteria to decide what to do with opt-out responses