Going, Going, Gone...Sucker!
Business Week Online
20 March 2000
How ''winner's curse'' could undermine this e-commerce channel
In this age of e-commerce overdo, it's often proclaimed that online auctions are the ideal way to match buyers and sellers--fast, frictionless, and perfectly fair. The public has certainly bought that message: EBay Inc.
(EBAY) alone has 10 million customers and a market value of $21 billion. Now, business is picking up the electronic gavel. The auto-parts auctioneer planned by General Motors
(GM), Ford (F), and DaimlerChrysler (DCX) is expected to handle $250 billion in transactions a year.
But auctions aren't a perfect form of commerce--far from it. They're prone to two fundamental flaws: winner's curse, which hurts buyers; and collusion, which usually hurts sellers. While it's possible to minimize both flaws through clever design of auction mechanisms, it's not clear that Internet auctioneers are paying the subject much attention. For now, online auctions are so popular that almost any site can lure business. But that could change abruptly. ''If this isn't done right, it will depress the size of the auction economy,'' says Bhaskar Chakravorti, an auctions expert at the Monitor Group, a Boston consultancy.
Winner's curse is the bane of clueless newbies on auction sites like eBay's. It's what people suffer when they win an auction by overestimating how much something is worth and therefore bidding too much. It's a sucker's game, yet it's part of daily business in online consumer auctions. Collecting money from newbies who overbid because they're ill-informed is ''part of the business plan,'' says one message poster on AuctionWatch.com's site. ''Started an item at $9 & it closed at $366+!!!'' brags another.
Paul Klemperer, an Oxford University economist who specializes in the economics of auctions, illustrates winner's curse to his students by auctioning off a jar with an undisclosed number of pennies. The students bid a little below their estimate of the jar's contents to leave a profit. Every time, though, the hapless winner of the jar is the student who overestimates the number of pennies by the greatest amount, and therefore overpays by the most.
FEEDBACK. Economists say there is only one situation in which winner's curse is not a problem: when there's no objective measure of a thing's value. Each bidder has his own concept of its value and isn't influenced by what others think. That's sometimes the case with tchotchkes on eBay. But most auctions--especially business auctions--are more like the one of pennies, in which the value is concrete but unknown. In those cases, the variation in bids comes about because of differences in information, not taste.
The soundest way to dodge winner's curse is to gather more information about the true value of what's being sold. That's one of eBay's strengths: Bidders can study past auctions of like items. And they can find out more about sellers by reading feedback about them. But if there's still uncertainty, bidders often watch the behavior of fellow bidders. That's where things get really complicated. If few others are bidding, does that mean the object isn't worth much--or are you in line for a bargain? Is your rival in a bidding war knowledgeable, or just crazy?
Second-guessing what other bidders are thinking is an especially serious problem in an English auction, such as eBay's, in which the price keeps rising until all bidders but one drop out. The trouble starts when one bidder is seen by others as being able to afford a higher bid--possibly because of a special ability to use the thing being auctioned. In an off-line auction conducted by the federal government in the mid-1990s, for instance, GTE Corp. wanted wireless licenses in certain markets where it had potential synergies. To signal its presence, it ended the dollar amounts of its bids in 483, which spells GTE on the telephone keypad. When other bidders realized that they could beat GTE in those markets only by overpaying, they dropped out early, handing GTE a bargain and reducing the government's auction revenue.
MANCHESTER MANEUVER. As that case shows, auctions don't work well when bidders differ in their ability to pay and can see what each other is doing. Last year, the awareness of that was enough to persuade British regulators to stop Rupert Murdoch's British SKY Broadcasting Group
(BSY) from buying the Manchester United soccer club. They feared owning the club would give BSkyB an extra incentive to acquire rights to televise games of the league to which Manchester United belongs, thus frightening off other bidders and depressing the value of the rights auction.
Collusion is the other inherent flaw of auctions. Because so much price information is on public display, it's easier for either buyers or sellers to collude in an auction than when transactions are done one on one. On eBay, for instance, one illegal trick is for two bidders to team up. One enters a low bid and the other then enters a high bid that no one can top. Afterward, the high bidder withdraws his bid and the sale automatically goes to his low-bidding buddy. Sellers can cheat, too, by using shills to enter fake bids and drive up the price.
Collusion is especially easy when there are few buyers or sellers, as in many business-to-business auctions. Businesses don't even need to break the law. If a few companies are competing for a customer's business, they could tacitly agree that it's in their mutual interest to charge the same high price and share the business. If one supplier defected from the group by dropping its price, its rivals could punish it by dropping their own prices.
What to do? Auction experts are divided over which mechanism works best. Monitor's Chakravorti favors ascending-price auctions, a la eBay, because they provide the most information about bidders' feelings. People tend to underbid when they're afraid of winner's curse. They will bid more freely if they can see that other bidders are also in the game, Chakravorti says.
Oxford's Klemperer argues the opposite. He says that sealed-bid auctions--or the functionally equivalent descending-price auctions--are more likely to get people to bid what they truly think something is worth. If bids are sealed, a heavyweight couldn't scare off other bidders, so it would have to bid high or risk losing, he says. Collusion is also harder to carry off in a sealed-bid auction, Klemperer says, because a defector from the low-bidding group can't be detected until the auction is over.
True, Chakravorti says. But there are other ways to minimize the drawbacks of an ascending-price auction. In business-to-business auctions, he argues, you could lessen the problem of heavyweight bidders' scaring off weaker rivals by creating an auction room open only to bidders that are similar to each other.
To auctioneers, this is no sterile debate. While most Net sites use ascending prices, a few--including Basement.com Inc. and OutletZoo.com--have dared to start prices high and drop them in increments until they sell out. It turns out that this induces the same bidding behavior as a sealed-bid auction. OutletZoo CEO Edward J. Samp says customers have told him his falling-price site is less vulnerable to shills.
But don't look for a wholesale change of Net auction strategies. EBay's popularity, says spokesman Kevin Pursglove, proves it's doing something right. He says that accusations of fraud of any kind--including use of shills--arise in only one of 25,000 transactions. As for critiques of ascending-bid auctions: ''I apologize for laughing,'' Pursglove says.
It's easy to laugh when you're No. 1. But auctioneers who don't continue to work on minimizing the flaws inherent in their auction mechanisms could wind up on the losing end of winner's curse.