The role of incentives on preference revelations in auctions versus rankings

The literature carries an extensive account of discrepancies in preference parameter estimates across different value elicitation methods. This study provides a detailed analysis of the effect of incentive compatibility on discrepancies in reported preferences. Specifically, we construct several partial ranking models (for best and worst alternatives) and conduct various statistical tests to determine where, how, and why discrepancies occur between an incentive compatible auction and a non-incentive compatible ranking task. While most previous comparisons were done using consumer valuation, this study uses preference orderings to provide a more robust comparison. For this purpose, the bids reported under the auction exercise were recoded as implied ranks. Our results highlight the importance of different incentive structures on the way people report their preferences in the presence and absence of incentive compatibility. In general, individuals were following different behavioral rules to evaluate the alternatives under the two elicitation mechanisms. However, preference parameter estimates for the best and worst alternatives seem to be more immune, although for different reasons, to the effect of different incentive structures than for middle ranked alternatives. This suggests that incentive compatibility and incentive structure should not be ignored when the analysis of preferences over the middle ranked alternatives is desired.