Sniping “Impossible” Auctions
Cao, W., Sha, Q., Yao, Z., Gu, D., & Shao, X. (2019). Sniping in soft-close online auctions: Empirical evidence from overstock. Marketing Letters, 30(2), 179–191. https://doi.org/10.1007/s11002-019-09487-7
Online auctions are vastly more complicated than in-person auctions. Beyond just the complicated strategy arising around advertising on search that we have been learning about in class, even ascending price auctions become more complicated when done on the web. As several other articles on this blog have documented, sniping is a market-distorting phenomenon that occurs on internet auctions, where bidders bid at the last second of auction close. Sniping leads to an inefficient outcome where the buyer with the highest valuation does not get the item, and the item sells for less than it should. To prevent sniping, several online auction sites use a “soft-close” which delays the auction close if a price increase occurs close to the end of the auction. This extra time is supposed to allow the buyer with the highest valuation time to increase their bid as many times as necessary until the true market clearing price is reached. In fact, the existing literature largely treats sniping in a soft-close auction as impossible!
Sniping in soft-close online auctions: Empirical evidence from overstock presents empirical evidence from Neiman jewelry auctions on Ebay, which does hard-close auctions, and on Overstock, which did soft-close auctions. While the authors do confirm that soft closes do result in greater prices, they also show that sniping behavior is amazingly still just as prevalent in soft-close auctions. Sniping activity simply shifts to right before soft-close delays start to occur. The authors do present data that shows that this sniping behavior is less likely to be successful with a soft close, but fascinatingly, they show that when successful, snipers pay significantly less than they would on hard-close auctions. Thus, in spite of its difficulty, sniping is still a viable strategy on soft-close auction sites because of the increased payoff. The authors actually speculate that the soft close and presence of multiple items could be what makes sniping give higher payoffs on soft-close auctions, but this speculation is not at all the focus of the paper. I assume that more theoretically minded researchers will eventually develop models that explain the quirks of sniping behavior in soft close auctions more fully.
In class and in the textbook, we discussed the idea that all auctions lead to the same result, although it may not always appear that way at first. This paper provides evidence for this theme. Sniping is prevalent on all types of internet auctions, and it will be difficult for an auction format to change that. Furthermore, we have been working through how bidder behavior when auctioning on multiple items is much more complex and less understood (as with GSP equilibria) than the simple auctions we studied earlier. The puzzling sniping behavior demonstrated on crowded online marketplaces is a perfect example of how we understand multi-item auctions much less well.