Abstract
We analyze choices of sellers, each setting a reserve price in a laboratory first price auction with automated equilibrium bidding. Subjects are allowed to gain experience for a fixed period of time prior to making a single payoff-relevant choice. Behavior of more experienced sellers was consistent with benchmark theory: average reserve price for these sellers was independent of the number of bidders and equaled the predicted level. Less experienced sellers however deviated from the theoretical benchmark: on average, they tended to shade reserve price below the predicted level and positively relate it to the number of bidders.
Highlights
For the independent private values model under benchmark conditions, auction theory predicts that the reserve price set by sellers should be independent of the number of bidders
Before presenting results from the tests of these hypotheses, we considered whether the expected revenue maximizing reserve price identified above might be a reasonable prediction when decisions were subject to sampling noise
We found that reserve price chosen by more experienced sellers did not depend on the number of bidders (Column 4), as predicted
Summary
For the independent private values model under benchmark conditions, auction theory predicts that the reserve price set by sellers should be independent of the number of bidders (see, e.g., Krishna 2010 [1]). We ask, for first-price auctions, whether there may exist alternate learning protocols that can enable learning and conformity of decision with prediction This is firstly because even with the issue of conjecturing bidder behavior eliminated by informing sellers about the automated bidding rule, the problem of deciding the reserve price remains complex. Introduced a protocol with backloaded incentives and a single payoff-relevant decision coming at the end of a learning or experience-gathering or practice phase They argued that particular advantages of this ‘learn-before-you-earn’ (LBE) protocol, which separates the learning from the payoff-enhancement process, over the standard ‘learn-while-you-earn’ (LWE) protocol were that the former prevents a focus on the myopic stimulus-response aspect of the problem and does not penalize low-payoff outcomes during the experience phase, providing subjects who have not completed learning some insurance against the downside risk of experimentation, thereby permitting them to choose riskier learning strategies with the potential to yield greater eventual payoff.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
