Shaded Truth under Consequential Risk: Theory and Evidence from Discrete Choice Experiments
Abstract:
Stated-preference surveys are widely used in policy analysis, yet practitioners frequently observe respondents who state low or zero willingness to pay and cite distrust of policy delivery as their reason. Standard practice classifies such responses as protests and either drops them or applies ad hoc corrections. This paper proposes consequential risk, a framework in which such responses are rational: respondents who believe the survey is consequential but perceive implementation failure strategically shade their WTP downward. Building on the consequentiality framework of Vossler, Doyon, and Rondeau (2012), we separate policy selection from benefit delivery and derive an attenuation identity: under risk neutrality, the stated valuation object is a product of the full-delivery WTP and the perceived delivery probability. This multiplicative structure motivates a design that pairs a controlled experiment with a field survey. We characterize a spectrum of controlled experiments, from low-cost survey variations to lab experiments with induced values to field delivery of real goods, each offering a different tradeoff between cost and identification strength. At the strongest tier, the distribution of failure beliefs is nonparametrically identified via multiplicative deconvolution of the experiment-identified and survey-identified WTP distributions. When the key conditional independence assumption is relaxed, partial identification results obtain and the aggregate welfare gap remains point-identified. The framework yields an empirically implementable decomposition of stated valuations into full-delivery value and perceived failure beliefs, and provides a principled basis for interpreting, rather than discarding, protest responses in policy valuation.