When Google successfully launched its initial public offering (IPO) by auction in 2004, there was much media and industry fanfare. In reality, using an IPO auction to determine who gets to buy how many shares at what price is an old approach that either never took off in some countries, or has gone out of fashion in others. IPO auctions were attempted in more than twenty countries in the 1980s and early 1990s, but are rare today.

Google’s high-profile IPO did not alter the trend: only three of more than 250 U.S. IPOs in the twelve months that followed were conducted using auctions. Why is this so?

imageIn a recent paper, Professors Ravi Jagannathan (Finance Department, Kellogg School of Management) and Ann E. Sherman (Finance Department, University of Notre Dame) set out to find the answer. They scrutinized various explanations as to why auctions are losing out to the seemingly inferior bookbuilding method and fixed-price public offers as the preferred ways to conduct IPOs. They conclude that unpredictable variations in the number of bidders and the presence of unsophisticated investors often make standard, uniform-price auction outcomes unstable and less than desirable.

Economic theories show that when certain conditions are met auctions can be very efficient in allocating a scarce good. They allow price discovery and may even maximize a seller’s profit. Hence, it is somewhat surprising that virtually all of the many countries that once used IPO auctions have abandoned them.

Two common explanations given for the unpopularity of auctions are reluctance on the part of issuers to try a new experimental method, and pressure from investment banks to use bookbuilding.

The first claim does not stand up to scrutiny, say Jagannathan and Sherman. IPO auctions were used as an alternative to traditional, fixed-price public offers in Italy, Portugal, Sweden, Switzerland, and the United Kingdom in the 1980s, and in Argentina, Malaysia, Singapore, and Turkey in the 1990s. However, they were abandoned in all of these countries long before bookbuilding was introduced from the United States. Hence, the rarity of IPO auctions cannot be due to unfamiliarity.

No more convincing is the suggestion that investment banks have pressured issuers to use bookbuilding because they garner higher fees. Jagannathan and Sherman point out that in most countries where IPO auctions were tested and later abandoned, they were usually replaced by fixed-price public offers, and public offer fees were typically as low as, or even lower than, auction fees.

So why do we still see the replacement of auctions by bookbuilding, which is potentially prone to abuse, and by fixed-price offers, which can lead to substantial underpricing?

Jagannathan and Sherman argue that there are two serious shortcomings with standard, uniform-price auctions under the IPO setting: the winner’s curse and the free rider problem.

The winner’s curse, as its name suggests, is the tendency for the winner of an auction to overbid. Theory predicts that this will occur when the extent to which a bidder values the auction item depends on other bidders’ valuation. The winner’s curse applies to the case of tradable shares, since one’s valuation of the share depends on everybody else’s valuation.

In principle, the winner’s curse can be easily overcome: a bidder could simply revise her bid downwards to make allowance for her optimism. In reality, however, bidders often find it difficult to adequately adjust for the winner’s curse, especially when the number of bidders is uncertain and difficult to anticipate. Jagannathan and Sherman provide a simple, stylized example to illustrate what happens when the actual number of bidders turns out to be far bigger than what each bidder had assumed when they computed their bids. The winner’s curse will be severe, and opting out of the auction may make more sense.

The free rider problem, on the other hand, arises because some investors have the incentive to rely on others to collect information on the IPO stock. These uninformed investors will bid high, hoping that the auction clearing price will be set by those who have done their homework. However, high bidding by uninformed bidders reduces the incentives of sophisticated investors to devote time and resources to correctly value the shares—since they are less likely to win due to the aggressiveness of the uninformed investors. Hence the standard, uniform-price auction mechanism becomes unstable.

The available data on IPO auctions is sparse and not easily amenable to precise qualitative analysis. Nonetheless, Jagannathan and Sherman examined a small sample of IPO auctions in Singapore to find empirical evidence on the winner’s curse and the free rider problem. They found that IPO clearing prices were indeed an increasing function of the number of bidders, and at the same time, a higher clearing price tended to be associated with lower IPO returns later on.

Reforming the Building Process for IPOs

Given that standard, uniform-price auctions are often touted as an alternative to the abuse-prone bookbuilding method, can IPO processes still be improved if auctions have generally proved disappointing?

In a related paper written in 2005, Jagannathan and Sherman point out that much of the widespread concern over IPOs’ underpricing may be misdirected, as underpricing is necessary in order to compensate larger, more influential investors for their crucial role in building the demand curve for a market that did not previously exist. The real issue after all may not be the presence of underpricing, but the extent of underpricing.

The solution, argued Jagannathan and Sherman, is to adopt a “hybrid” mechanism—one that preserves the main advantages of bookbuilding while addressing concerns about its exclusivity and lack of transparency.

Specifically, this involves improving the well-established bookbuilding process by applying transparency features of standard, uniform-price auctions.

Standard, uniform-price auctions weigh bids on only one single factor: price. Jagannathan and Sherman propose that bids be weighted on their quality and timing as well. For example, limit bids (orders to buy at no more than a specified price) could be favored over market bids, since limit bids demonstrate the willingness of investors to purchase shares at the bid price. As such, limit bids convey more information on investors’ valuations than market bids (orders to buy at the current price), which simply mean that the investors want shares at the offering price.

Jagannathan and Sherman discussed several variations of the “hybrid” solution in their paper, but the basic principle behind these variations is the same: add more transparency into the bookbuilding approach.

The debate over the reform of the IPO process has focused mainly on replacing bookbuilding with standard auctions. But large, multi-unit IPO auctions will work only if essentially all participants are highly knowledgeable, disciplined, and sophisticated.

Jagannathan and Sherman meticulously demonstrate the impossibility of this task. The very nature of IPOs—occurring sporadically, with each issuer different from the previous one—makes it impossible for millions of potential investors to obtain such a high level of skill and sophistication.

However, the authors suggest that not all is lost. Their analysis of the strengths and weaknesses of IPO auctions provides useful insights on how certain features of standard auctions can be incorporated into the well-established bookbuilding process (or vice versa) to address concerns regarding exclusivity and lack of transparency. Whether you call this method a nonstandard auction or modified bookbuilding, it may be the way forward.