Winner's curse is often a problem for auctions.

Economists have long viewed auctions as a highly efficient pricing mechanism for buying or selling products and services. But it isn't widely understood that the rules and terms of a given auction can profoundly affect its outcome. Therein lies danger for governments which want to capture the benefits of auctions but don't fully understand the process.

For instance, Europe plans to auction licenses for the so-called Universal Mobile Telecommunication System services. This third-generation mobile telephony promises to do just about everything short of picking up laundry and preparing meals.

If the reality matches the hype even halfway, it will be nothing short of a revolution in personal convenience. Consumers will be able to download videos, surf the internet, participate in conferences and check their e-mail—all on the go.

The European Commission envisages these technologies becoming available to consumers in 2002 and has called on member states to develop licensing processes for these services by January 1 of next year. So far, the United Kingdom has announced that it plans to auction five licenses early in 2000 and raise about $4 billion. Germany is also reported to favor an auction by next spring, and other states, including Sweden and France, may well follow suit.

The question of how many licenses will be issued, what their geographic coverage will be, and how the provision of "roaming" rights will work have been left to individual member states.

But unfortunately, the details of the licensing schemes are fuzzy. As experiences in the U.S. and elsewhere have shown, public authorities neglect such details at great risk.

Billed as the "greatest auction in history," the U.S. Federal Communications Commission held 16 licensing auctions from July 1984 to July 1998, raising $22.9 billion in revenues, far above pre-auction estimates.

The bad news was that some companies bid beyond their means. Dallas-based General Wireless, which had bid $1.1 billion in one of the auctions, subsequently sought protection under Chapter 11 of the U.S. Bankruptcy Code. Similarly, the Indian government, which recently auctioned its telecom licenses, is in the process of finalizing a bailout package that might cost $11.6 billion for licensees who later realized they had overbid.

Game theorists point to such predicaments as examples of "winner's curse": This is where an object of uncertain and "futuristic" value—of which a broadband telecom services license is a good example—is won by a bidder which regrets its high bid after it wins.

To avoid fear of such regret, authorities must adhere to some basic principles in setting up such auctions.

Encourage transparency. Auction participants typically have a nagging anxiety about whether someone else has more information than they. If a bidder is thoroughly involved in the bidding process, he gets better knowledge about his competitors, reducing the potential for overestimation and winner's curse.

To that end, the open bid ascending price procedure applied in U.S. auctions is preferable to sealed or descending-price auctions. Under the former, participants place bids, and try to top the highest one on record.

Moreover, information is more valuable if transmitted among truly interested participants. In the U.K.'s radio spectrum auction, interest among players is high.

All of the wireless incumbents, Vodafone Group Plc, Cellnet, Orange Plc, and One2One, as well as British Telecom have expressed an interest; others such as COLT and Energis would find such licenses to be entry vehicles into the broadband access market for smaller business and residential customers; foreign players like MCI Worldcom would view the award of licenses as an end to the local-loop monopoly of incumbents such as BT. This interest level, which was encouraged by the government, will help ensure more open and level-headed bidding.

Look out for signs of collusion. Of course, the more open the process, the more susceptible it is to collusion, including manipulation by bidding rings. Big players or collusive rings could enforce low bids, with threats to punish anyone who defects from such implicit agreements by aggressively going after a market the defector really covets.

In the U.S. auctions, for example, GTE ended its bids with the digits "483," which spells GTE on the telephone keypad, apparently so that other bidders would know of GTE's bid and interest level in that license. The president of PrimeCo, George Schmitt, less subtly but credibly, declared to competitors: "You mess with me in Chicago, you pay."

The ultimate impact of this collusion was small, most likely because the auctions were sufficiently complex (with approximately 2,000 licenses being auctioned simultaneously in one case) and had a sufficiently large number of bidders so that it was difficult to effectively send signals and follow through with threats.

However, in a smaller auction, such as Europe's, this may be more of an issue. Strict anti-collusion rules, monitoring bid patterns, ruling out initial, inconsequential but signal-rich bidding and similar measures can help weed out would-be colluders.

Don't hold back privileged knowledge. Sellers frequently have a better perspective on the true valuation of what is being auctioned than bidders. By virtue of having owned it, they have had more time to benchmark what might have been paid for similar assets, or they might have a better perspective on forthcoming regulatory or technological changes.

This is particularly true if the seller is the public sector, and the buyer is a new entrepreneur—as would frequently be the case in spectrum auctions.

While it might be tempting to hold back such information, we have found that it is foolhardy to do so since reducing uncertainties around valuations can help ensure that a bidder doesn't bite off more than he can chew.

Share risk. If the risks of failure are high, the process should accommodate risk-sharing arrangements. For example, the payment for a license could have two components: a fixed fee, and a fee contingent on the revenues generated by the service. Such arrangements reduce the fear of overestimation and decrease the risks of defaults.

In both the U.S. and India, preemptive risk-sharing arrangements helped avoid costly renegotiations when the winner was unable to make license payments. These agreements also mean that the government retains the potential to share in any revenue upside. Attending to these basic principles is crucial not only in the upcoming European telecom auctions, but in other venues as well.

In 193 A.D., Didius Julianus won the auction to become emperor of Rome. Two months later, he was beheaded. Ever since, the auction and its accompanying "winner's curse" have continued to penetrate our day-to-day lives. It is important, then, that Europe's current players not lose their heads—or shirts. As with most things, the devil lies in the details.

— from The Wall Street Journal Europe