Are choice experiments incentive compatible? a test with quality differentiated beef steaks

This study compares hypothetical and nonhypothetical responses to choice experiment questions. We test for hypothetical bias in a choice experiment involving beef ribeye steaks with differing quality attributes. In general, hypothetical responses predicted higher probabilities of purchasing beef steaks than nonhypothetical responses. Thus, hypothetical choices overestimate total willingness-to-pay for beef steaks. However, marginal willingness-to-pay for a change in steak quality is, in general, not statistically different across hypothetical and actual payment settings. Key words: CVM, conjoint analysis, heteroskedastic extreme value model, hypothetical bias, multino-mial probit, random parameters logit. Economists have used choice experiments (CE) and choice-based conjoint analysis to estimate the value of nonmarket goods, the trade-off between environmental or food qual-ity attributes, and market share of novel prod-ucts (e.g., Adamowicz et al.; Jayne et al.; Layton and Brown; Lusk, Roosen, and Fox; Unterschultz et al.). The frequent use of CEs might be explained by a number of factors. First, CEs are flexible because numerous at-tributes can be simultaneously valued. Con-ventional contingent valuation (CV) methods require cumbersome repeated questions to achieve the same level of complexity. Second, CEs are consistent with random utility the-ory (Ben-Akiva and Lerman) and Lancaster’s theory of consumer demand, which posits that consumers derive utility from consumption of attributes embodied in a good. Third, indi-vidual CE questions are typically framed in a manner that closely resembles consumer pur-chasing decisions (e.g., making a choice among alternative products). Because CE questions closely mirror actual consumer purchasing situations, it has bee