28 Feb 2005  Research & Ideas

Amazon, eBay and the Bidding Wars

"Sniping" is a popular way of winning a bid in the world of online auctions. But how far can it change the playing field? HBS professor Alvin Roth takes a look at how bidding rules change the way the game is played.


Beware of competitors lying in the grass, says Harvard Business School professor Alvin E. Roth. His study of bidding practices on eBay suggest that those who wait until the last minute to bid—a practice called sniping—is an effective way to not only get what you want, but to keep the final price lower.

But sniping isn't a universal strategy for success, says Roth, who teaches in the School's Negotiation, Organizations & Markets unit. It's all in the rules of the game. On Amazon.com, for example, where end-of-auction deadlines can be extended, snipers are out of luck.

One lesson for managers: What happens at the start of a bidding process is not always indicative of how it will end up.

Sara Grant: What attracted you to study late-bidding behavior on eBay and Amazon? Were there any surprises from what you originally anticipated might be the results?

Alvin Roth: My colleagues and I who work in market design are interested in how the rules by which markets are organized influence the behavior of participants. So, when I noticed, back in the early days of eBay, that a lot of auctions received bids right near the end of the auction (eBay participants call this "sniping"), I wanted to know why.

When colleague Axel Ockenfels and I started to look into this, we noticed that there was much less late bidding in Amazon auctions than in eBay auctions. And, at the time we compared them, those auctions had only one important difference in their rules: They had different rules about how an auction ends. EBay auctions have a firm deadline: When the scheduled end time of the auction is reached, the auction is over. Amazon auctions also have a scheduled end time, but the auction is extended if there are bids near the scheduled end; the rule is that the auction can't end until at least 10 minutes have passed without a bid.

Q: In your study of eBay, how much sniping was going on, and what categories or products were most likely to be sniped?

A: In the eBay data we collected for our 2002 American Economics Review paper, "Last Minute Bidding," 55 percent of the auctions have bids in the last 10 minutes, 37 percent in the last minute, and 12 percent in the last 10 seconds. Antiques were more heavily sniped than computers, but there was lots of sniping for computers too.

There was much less late bidding on Amazon. Only 11 percent of the auctions in our data had to be extended because a bid was made in the last 10 minutes of regularly scheduled time.

Q: When is sniping a winning strategy?

A: On eBay, people snipe to avoid price wars. By bidding at the last minute, they don't give others who might want to change their bids in response the time to do so. On Amazon, sniping is much less effective, since the automatic extension rule means that other bidders always have 10 minutes to respond.

And sniping is not without cost: Planning to make a late bid may mean that you fail to make a bid, either because the auction closes before your bid gets through (when you're trying to make a very late bid), or because something comes up that prevents you from getting around to making a bid. (There is sniping software to address this latter problem, but when we surveyed eBay users we didn't find that it was being widely used.)

So, the question is, when is sniping worth the risk of not getting your bid in? There are lots of reasons it might be. The easiest case to think about is of a bidder who has some expert information about the item being sold, information that might affect how much other buyers are willing to pay. For example, suppose that a dealer in Persian carpets sees a desirable one on eBay. Many bidders may not be able to tell the difference between valuable carpets and cheap imitations, and so if they can identify that a dealer is bidding on a particular carpet, that will tell them that it is a good one. So, the dealer can try to conceal who he is, to avoid starting a price war, but for a dealer who bids a lot, it is hard to have lots of different usernames. A simpler strategy is for him to snipe, so that others can't piggyback on his expertise. But even without special expertise, a bidder may feel that an early bid will inspire others to raise their bids, and in such a case, sniping may be a winning strategy.

Q: How pervasive is sniping and will it eventually wipe out incremental bidding? Can or should eBay and other online auction houses try to level the playing field?

A: EBay is still evolving, and there's a growing part of its business that is hardly an auction market at all, as retail sellers use it as simply another channel to move goods that are in abundant supply. Sniping won't play a role in that part of the business, but it will no doubt continue to be a feature of the auction markets. If sniping were to entirely replace sequential bidding, eBay auctions would become a lot like sealed bid auctions, in which bidders submit bids that are all opened simultaneously at the end. In fact, there are third-party providers of bidding software and services (such as esnipe.com) that, for a fee, allow eBay bidders to submit their snipes early, which is a lot more convenient way to snipe. Of course, if eSnipe ever started to mediate a lot of eBay's auctions in this way, eBay could take all their business by offering a sealed bid option to bidders. Presumably the reason that eBay is reluctant to do this is that they think it is good for the market for bidders to be able to see each other's bids, precisely because that may increase the total amount of bidding and raise auction prices.

Q: What might managers take away from your research, with regard to strategy and pricing?

A: What happens early in an auction may not be a good predictor of the final outcome, because much of the bidding activity takes place just before the end. One good reason for this delay is that early bidding reveals intentions and information that may be exploited by competitors if there is still time to react. So, be aware of competitors hiding in the grass. It's not over 'til it's over.

Q: What are you working on now?

A: Several of my current projects have to do with market design, with trying to get the rules right. With Tayfun Sonmez and Parag Pathak here at HBS (and with Atila Abdulkadiroglu at Columbia), I've been looking at how students are matched with schools. We helped New York City design a new matching system for assigning students to high schools. (Under their old system, 30,000 students were assigned to high schools they didn't choose, while in the first year of the new mechanism, only 3,000 were, out of the almost 100,000 students who entered NYC high schools this year.) We're at an earlier stage of a similar project with Boston Public Schools. Tayfun and Utku Unver and I have been working on kidney exchange, to facilitate exchange of live-donor kidneys when there are kidney patients who have a willing transplant donor who is incompatible with them. In September 2004, the Renal Transplant Oversight Committee of New England gave the go-ahead to a kidney exchange program we proposed together with Drs. Francis Delmonico and Susan Saidman at the Massachusetts General Hospital. And Muriel Niederle [at Stanford University] and I are trying to help the American Gastroenterology Association make reforms in their labor market.


From Roth, Alvin E. and Axel Ockenfels "Last-Minute Bidding and the Rules for Ending Second-Price Auctions: Evidence from eBay and Amazon Auctions on the Internet," American Economic Review, 92 (4), September 2002, 1093-1103.

About the author

Sara Grant is publications coordinator of HBS Working Knowledge.