|
February 2001, Volume 69, Number 2 |
|
Part 1 |
Part 2 |
|
For consumers and many businesses, the more direct evidence of auction fever comes from the Internet. Since 1995, when eBay opened the inventories of millions of garages and basements to geographically dispersed bidders, Web auction sites have skyrocketed. By 1999, market analysts were predicting 14 million people would participate in online auctions by 2003 and that a wide range of business-to-consumer and business-to-business auctions would force drastic changes in business plans. Business School alumni/ae, whether they took game theory courses or not, found themselves having to analyze the impact of new auction formats. Sites like AuctionWatch.com, cofounded by Rodrigo Sales, MBA '99, sprang up to help bidders and sellers keep track of multiple auction sites and manage post-auction payment, shipping, and insurance. The Internet sites provide natural experiments for researchers trying to understand bidding behavior. Among those who have studied consumer-level Web auctions are Patrick Bajari, an assistant professor in Stanford's economics department who also looks at how sellers might combat bidder collusion in construction contract bidding, and Alvin Roth, a former student of Wilson's and a Stanford PhD graduate in engineering, who is now a professor of economics at Harvard and at the Harvard Business School. Roth has used game theory to design labor markets such as the National Resident Matching Program, which matches most U.S. medical residents to hospital and clinic openings. He also studies the cognitive limitations on players and how they adapt as they learn how a market "game" works. Bajari and Roth, like many eBay visitors, noticed that the bidding gets fast and furious near the end of eBay auctions, which have precise deadlines. The deadlines have led to a practice called "sniping," where bidders try to place their bids at the last possible second, some using software that automates the practice. At Amazon.com's auction site, sniping is less common, the researchers say, because there is no firm deadline. An Amazon auction is declared closed only after a 10-minute period without bids. The auction closing rules, according to Bajari and coauthor Ali Hortacsu, Stanford doctoral student in economics, would make eBay less seller-friendly than Amazon. com, were it not for the fact that eBay draws many more people, and the more competition, the better the price sellers can expect. Roth says eBay's deadline allows bidders to avoid bidding wars and helps experts avoid tipping off novices to the value they place on a particular item. Some analysts have speculated that Internet auctions will end fixed prices, putting consumers in the driver's seat and reducing profit margins to the absolute minimum. The largest dollar impact, many have said, will be in business-to-business dealings, because a General Motors, say, could force suppliers of tires or batteries to bid against each other in "reverse" auctions. But Stanford's Milgrom and McMillan say the case has been overstated. "One of the things people are very confused about is where the value added comes from with the Internet," says Milgrom. "The Internet in business is about facilitating machine-to-machine communication, or automating processes that used to be very costly." The recent controversy over Firestone tires illustrates another limitation of simple auction institutions: Price is not the only dimension of a product or service that buyers care about. Thanks to the Internet, McMillan says, "it's much easier now to find people offering something you want and to comparison shop, but if you are not sure of the quality of what you are getting, then you will buy the old-fashioned way, which means trading with somebody whose reputation you know." Milgrom is chief economist for a company called Perfect.com that offers market-makers software to automate business negotiations on multiple attributes of products and services. Buyers generate requests for quotes on as many as 100 weighted variables, linking aspects of quality, delivery time, and cost. These preferences are matched with customized seller offers. Both sides can make "indicative" requests, rather than solid bids, which human negotiators may modify later. "The real savings will not come from beating up suppliers in price-only auctions," Milgrom says, "but rather from reducing the costs of sourcing for both buyer and supplier." Many markets will be confidential, he adds. "If you are Long John Silver planning a seafood special, you do not want Red Lobster to know about your orders." Research Directions Security and the prevention of collusion are big areas for research, and Stanford's Business School and economics department are in the thick of it, along with some electrical engineers who work on "shopbots," or robots for automated Web buying. Professors and graduates meet regularly in reading groups and seminars, and many say they were attracted by the University's leadership in this area. Andrzej Skrzypacz, for example, came to the GSB as an assistant professor last fall because of its concentration of game theorists. Currently studying collusion, he is analyzing sealed bid auctions of the type used by public agencies in contracting. His theoretical work indicates that when the same bidders compete repeatedly, as in school milk contract auctions, collusion can still be fruitful, although much less so, if bidders forgo illegal meetings and monetary exchanges beforehand but agree to take turns winning, depending upon the history of past winners. Designing markets for products or services that are linked or complementary is another active research area. Electricity is a good example because "it doesn't do any good to generate electricity if you don't have the transmission capacity to deliver it," Milgrom says. "The unlinking of those markets leads inherently to inefficiencies, according to theory, and that has been a problem in the design of electricity markets." Nevertheless, governments are hopeful that power markets will be more efficient than regulated monopolies, and Wilson has spent much of his time recently working on electricity auctions. "No two designs among liberalized electricity markets are the same," he tells his class in game theory, "so in effect, an enormous experiment is under way." Designers must cope with the fact that the market is "inherently incomplete and imperfectly competitive," he says, partly because electricity is a flow, not a stock that can be easily stored, and cannot be metered perfectly. Someone has to operate a transmission system, which is a natural monopoly, and some generators may have enough capacity that they can withhold power or sell it in such a way as to affect the overall price. The procedural rules used in these markets so far are crude, Wilson believes, compared to innovations that theorists and practitioners are likely to develop as they gain experience. Auctions, like bridges, date back centuries, but until recently they were "designed" by trial and error, not engineered by economists. Dutch flower merchants, for example, came up with a descending bid auction so those with the farthest to travel could buy the perishable product quickly at the highest price. It wasn't until 1961 that the late William Vickrey demonstrated that many auction formats were equivalent—the so-called revenue equivalence theorem that was stressed by the Royal Swedish Academy in awarding him the 1996 Nobel Prize. Vickrey also devised an auction (in which the winning bidder pays the second highest bid) that has the same efficiency but which reduces the transaction costs associated with trying to outwit other bidders' strategies. Wilson then came along and made game theory as essential to applied economics as to theory, says Harvard's Roth. "His early papers on modeling auctions as games of incomplete information are landmarks." It's been years since I bought the Hayride bulb at auction, and it has yet to produce more than a few strappy leaves. Meanwhile, imported Dutch irises, which I bought outright for $3 a dozen, pop their purple heads each March. At the auction, I overheard one bidder telling another that irises should not bring prices as high as orchids because their blooms were short-lived. "But each one is so unique," her friend said of the hybrids. The game theorists at Stanford take into account both types of bidders in their work. After my experience with Hayride, I tend to agree with McMillan: When in doubt, I plan to buy the "old-fashioned way." Return to "The Auction Architects: Part 1"
|
This is an official Stanford Graduate School of Business Web page
Copyright © 2001 Stanford University - Graduate School of Business