The effect of experience on Internet auction bidding dynamics

Abstract

On the basis of the bidding history of a panel of new eBay bidders, we examine the impact of different types of experiences on bidding behavior evolution. Accounting for unobserved bidder heterogeneity, the results indicate that losing experiences make the bidders' bidding behavior evolve toward the normative predictions of auction theory, in that they submit fewer bids and bid later. Winning experiences, however, do not have such an effect. Moreover, the experience effect pertains to the bidder's entire previous bidding experience regardless of product categories. We also assess the potential bias introduced by using feedback ratings (compared with actual participation) as experience measures.

The Impact of Buy-Now Features in Pay-per-Bid Auctions

“…As the dependent variable, we use the absolute deviation of bids compared to the optimal number of bids (i.e., the upper threshold (BNP i -p i,r /(b i · α i ))) each time a bidder uses the BNF. As an independent variable, we use the logarithm of the cumulative number of auctions in which each bidder participates as a measure of experience (e.g., [29]). We also include the logarithm of the cumulative number of BNF usages to control for BNF experience and the logarithm of the cumulative number of auctions won.…”

Rejoinder—Causes and Implications of Some Bidders Not Conforming to the Sealed-Bid Abstraction

“…However, this caution does not apply to our estimates of demand in the second stage because our first-stage estimates of the probability of bidding style conditional on feedback (interpreted as a mere endogenous proxy for unobserved heterogeneity) and Years on eBay (proxy for cumulative experience) are not necessarily biased. Wang and Hu's (2009) results suggest that within a given entry cohort (i.e., within a generation of bidders all entering at the same time), bidders' behavior asymptotically approaches sealed bidding over time, and so the violation of the sealed-bid abstraction we document should be weaker within each cohort as time passes. We agree that a study interested in individual behavior can therefore still invoke the sealed-bid abstraction on a properly selected subset of experienced enough bidders.…”

“…We interpret feedback as a measure of bidder experience. Srinivasan and Wang clarify the interpretation of this correlation and suggest other measures of bidder experience following Wang and Hu (2009). Tracking a cohort of bidders for the first six months of being registered on eBay, Wang and Hu find a negative correlation between total experience (the total number of auctions in which the bidder submitted a bid) and multiple bidding.…”

Estimating functional agent-based models

“…The spectacular growth of internet auctions and the availability of massive amounts of auction data has lead to new insights into bidder-seller behavior [4,11,8,25], the impact of the auction format [20], the auction process [27,24,1,5,13,22] and its dynamics [18,6]. However, while most of the extant literature focuses on behavior that is directly measurable from observed data (such as the timing and the magnitude of individual bids), we are interested in behavior that is unobservable.…”