On the reliability of the BDM mechanism

The Becker-DeGroot-Marschak (BDM) mechanism is frequently used in experimental research to measure willingness-to-pay (WTP). The mechanism is very appealing given its incentive compatibility property. However, a growing body of experimental evidence finds that WTP is sensitive to the underlying price distribution of the BDM mechanism. The literature suggests that these findings can be rationalized only if subjects have context-dependent preferences. Under context-dependent preferences, the use of the BDM mechanism momentarily changes subjects' preferences; subjects do not have incentives to report their intrinsic valuation, which is what the BDM attempts to measure, but instead they report the amount they are willing to pay after their preferences have momentarily changed. This lack of incentive compatibility would raise serious concerns about the validity of the BDM as an elicitation procedure; indeed, virtually all studies that use the BDM mechanism choose that elicitation procedure precisely for its incentive compatibility property. In this paper, we propose that an assumption of context dependence is not necessary to rationalize the aforementioned experimental findings. We provide an alternative plausible theory and experimental evidence to support it. Our theory relies on the premise that subjects need to exert effort to learn their preferences. We find that the price distribution of the BDM mechanism influences the subject's incentives to exert effort. The results presented here are particularly appealing to the experimental research: our theory is consistent with the experimental findings that WTP is sensitive to the price distribution of the BDM mechanism while preserving the incentive compatibility property.