Google made its long awaited debut on the stock market Thursday August 19th. The outcome of the IPO was not what many would have predicted when Google filed to go public in April. Battered by a string of negative publicity, Google was forced to scale back both the size and suggested price of its offering in one of the few public gaffes the company has made. Instead of the IPO being the phenomenal success that appeared to be its destiny, Google's executives were probably just relieved to have the process completed at all. Why did this happen and what can future issuers learn from this seminal IPO?
At a very minimum, Google demonstrated how not to conduct an IPO, particularly one that involves an unproven auction method. The company displayed a certain amount of hubris by suggesting a price range of $108 to $135 per share, which translated into a market value of $29 to $36 billion. This priced the company for perfection and offered no discount relative to Google's more experienced and diversified competitor Yahoo, especially after tech stocks fell in the two months prior to the offer date. The expected triple digit price also left many investors cold. On top of that were the frequent revisions Google had to make to its S-1 filing after the revelation of some new stock grant or sale provision, not to mention notorious interviews. Compounding matters was the uncertainty investors faced with the unproven auction process.
The final result for the IPO was not surprising, given the events leading up to the offer date, but it also contradicted Google's initial intent (see Trying to Make Sense of the Google IPO: A Preview for a pre-IPO analysis of what could happen and why). The offer price of $85 was well below initial expectations, but understandable in light of the adverse market conditions that forced most issuers in July and August to either revise downward their price range or withdraw their IPO entirely. The lukewarm investor interest in the offering, coupled with Google's own missteps along the way, put further downward pressure on the price. However, the price jump to just over $100 by the end of the first trading day, giving IPO investors an initial return of 18 percent, is at odds with the auction supposedly determining a fair market price and leaving little money on the table. Google's underwriters almost certainly underpriced the offering below the auction-clearing price to ensure that the priced 'popped'. The need to make a strong market debut after all that negative publicity was clearly paramount. But by doing so, Google moved away from the egalitarian spirit of the auction, which raises the questions: did the auction experiment fail and if so why?
Was all this predictable?
Identifying what Google did wrong is fairly easy; the more difficult challenge is to determine what role the auction method had in contributing to the IPO's troubles. It is easy to claim, as many have done, that the auction process was confusing to investors. This suggestion is both hard to accept and rather insulting to investors' collective intelligence. The process was fairly straightforward: investors first had to obtain a registration number and, if they so chose, could then submit a bid to one of the syndicate members specifying the number of shares they wanted to buy and the maximum price they were willing to pay. Considering that millions of people participate in the auctions on eBay, Google's auction doesn't appear to be too taxing for investors. This is not to say that investors weren't faced with considerable uncertainty about Google's share valuation and how the auction would turn out. But those are very different concerns than how the auction actually works.
The broader question is whether there is something more fundamental about auction mechanisms that would make the somewhat disappointing outcome for Google's IPO, measured relative to initial expectations, not only explainable but also predictable for other issuers. To analyze this issue it is necessary to quickly review what transpired during the lead-up to the offering. The IPO was initially greeted with great enthusiasm, especially by retail investors who believed that the auction would give them equal access to the shares. The auction was anticipated to draw possibly millions of bidders. As the process moved along, trepidation amongst investors set it because of valuation concerns and fears of an adverse outcome (as Google suggested was a possibility because of the winner's curse). By the end it was clear that far fewer investors chose to participate than what was expected. The reduction of the number of shares offered at the last minute is a clear acknowledgement that the demand just wasn't there. Consequently, the IPO's fate underwent a reversal of fortune that ultimately precipitated the underpricing.
In this diagnostic examination of the Google IPO, it is instructive to compare auctions with the standard bookbuilding method - so-called because underwriters canvass institutional investors in the primary market trying to build the book of orders for the offering. While it is impossible to know with certainty the outcome of the IPO had bookbuilding been used, the discussion of how things could have been different also reveals why auctions can come up short. Obviously, it would be incorrect to lay all of the blame for the IPO's problems on the auction mechanism, but many can be traced back to properties of auctions that are not unique to Google's specific failings.
The dichotomous investor interest in Google's offering, tremendous early enthusiasm followed by apprehension as the offer date neared, is typical of the extreme outcomes that characterize IPO auctions, in particular large-scale and highly public offerings. Countries as diverse as the U.K., Japan, Singapore and Argentina have experienced large IPO auctions, usually involving the privatization of a well-known state-owned enterprise. The auctions were often either wildly oversubscribed or bordered on being under-subscribed. Both extreme events are associated with their own problems. Oversubscribed auctions can have an upwardly biased offer price because some investors bid excessively high, either to guarantee an allocation or out of naivetÇ¸. The price inevitably falls in the aftermarket, leading to the winner's curse. Under-subscribed auctions have the opposite problem: generating too few proceeds.
Two factors contribute to the volatile nature of auction outcomes. The first is that investors often succumb to a herd mentality, moving on mass either into or out of an IPO. Positive word of mouth about an issuer will attract more investors to the IPO, especially if underpricing is expected. Investors will instead stay away in droves when doubts begin to surface about the success of the offering. The early buzz can, in both cases, lead to self-fulfilling expectations. The latter scenario seems an apt description of Google's IPO.
The second critical factor is that with an auction the underwriter has little control over who participates in the IPO. Investors cannot be denied the right to bid, although Google's syndicate members seemed to be pretty selective on this matter. Even more important, though, is that they can't be forced to participate. Since allocations in auctions are determined by the bids and there should be little underpricing, there is not much an underwriter can do to entice reluctant investors to bid. Herd behavior can afflict any IPO mechanism, but bookbuilding grants the underwriter discretion over the share allocation that can alleviate this problem. The reason why will be discussed shortly.
A defining attribute of the Google IPO was the pervasive uncertainty. Investor beliefs about what Google was worth were widely dispersed, making it very difficult for investors to get a read on what the likely market price would be. The use of the auction, with its corresponding volatile outcome, only exacerbated this problem. There was simply too much risk associated with bidding, without a commensurate reward, for many potential investors in the IPO. When the amount of risk involved in an IPO, for both investors and the issuer, surpasses a critical level, the entire process can come undone.
Relying on an auction to set a market-clearing offer price entails risk for both the issuer and investors. The offer price could vary high above or far below the issuer's initial expectation of the share's value. Given the time and effort put into going public, most issuers want to minimize the uncertainty associated with the final outcome. Trading-off a lower price for less risk, an attractive option for some, is not feasible in an auction. The principal risk for investors, particularly when the issuer is well known, is the winner's curse. The winning bidders can suffer from buyer's remorse if they realize they paid too much and the price falls in the aftermarket. When this risk seems too great, as it did initially in Google's auction, more and more investors will abstain.
A fundamental difference between bookbuilding and auctions is that underwriters act as an intermediary between the issuer and investors in the former, but have only a limited role in the latter. One job of the intermediary is to re-allocate the risk of the offering efficiently among the issuer, investors and the bank. Bookbuilding allows the underwriter to eliminate the winner's curse risk by disregarding extreme bids and setting an offer price not likely to exceed the clearing price in the aftermarket. The underwriter can also transfer the issuer's risk of an uncertain market reception to investors by intentionally underpricing the IPO below the expected market price and guaranteeing the offer price. IPO investors are compensated for taking on this risk with the underpricing. This appears to be what Google's bankers did, even though it was officially an auction. If so, they wouldn't be the first. The CFOs of recent issuers cited the need to reward investors for taking on the risk of the IPO as the number one reason for underpricing. 1 Thus, even if bookbuilding results in a lower expected offer price compared to an auction, and that's debatable, the reduced risk can make it the more attractive method for issuers.
Favoritism or coalitions?
Auctions democratize the IPO process by (theoretically) giving investors equal access to the shares. This type of democracy can be great for retail investors, although the initial returns should be negligible, but it can also be detrimental to the issuing companies. It is important to remember that a primary objective of the IPO process is to enable issuers to sell shares for the maximum obtainable price in the most efficient manner possible. The optimal selling mechanism may require favoring some investors, namely institutions, over others. As long as retail investors are not being harmed in the process, it is not clear why their interests of fairness should take precedence over the goals of the issuer.
In the aftermath of the Internet bubble it was revealed that many of the allocations in those hot offerings were going to the favored clients of the investment banks. Institutional investors gained these coveted positions in return for their trading commissions, some of which were illegally tied to the allotted shares, and occasionally from a commitment to buy more shares in the aftermarket at higher prices, also a no-no. The combination of enforcing existing laws, proposed new reforms and public scrutiny should, hopefully, prevent the banks from abusing their discretion like this again. However, some critics argue that bookbuilding is still flawed because the allocations are contingent on an investor being part of Wall Street's exclusive fraternity, not their merits as a shareholder.
There is another way to interpret the allocation patterns that appear to be the result of favoritism. Underwriters form coalitions of institutional investors who regularly participate in their IPOs, thereby serving a valuable economic purpose. The formation of regular investor pools allows underwriters to bundle together multiple IPOs. Participation in the pool involves a quid pro quo for the investors. They receive the largest allocations in the hottest IPOs, but in return they must accept allocations in the less desirable offerings. Passing on one too many lemons results in banishment from the pool. The benefit to issuers of investor coalitions is that it smoothes out demand across offerings, mitigating the negative consequences of the herd moving against their offering. Since underwriters have no say in the allocation with auctions, they're hard pressed to prevent a poor offering. Institutional investors probably would not have felt so free to sit out the Google IPO if they thought they would face consequences from the bankers.
Auction pricing fallacy
Auction proponents like to point to the fact that the market determines the offer price, which is therefore more informative than the bookbuilt price. Furthermore, an auction avoids the intentional underpricing, thereby increasing the issuer's proceeds. These claims rely critically on the assumption that the market-clearing price, whether it is set in the secondary market or the auction, is determined independently of the IPO method used. For example, it assumes that the initial market price of an issuer will be $20 per share no matter what. The auction-determined offer price would then be about $20 as well, whereas the bookbuilt price would be, say, $17. The problem with this assumption is that there is good reason to believe that both the accuracy and level of the prices are influenced by the IPO method.
Market prices are a function of the information possessed by the investors who trade the stock. The number of investors who become informed about an issuer around the IPO and the amount of information they produce is contingent on their incentives. Uniform price auctions like Google's can suffer from a free-rider problem. Since the offer price is determined by the lowest winning bid, many investors may not conduct costly analysis and instead submit high uninformed bids to guarantee an allocation, with the belief that their small order will not affect the price. But if too many investors do this, the price will not be very informative. Even worse is that the price can fall in the aftermarket when investors realize what happened.
Insufficient information production can also occur if too few investors decide to participate in the auction. This could happen if a herd mentality takes hold and most investors choose to be observers rather than participants. Another possible reason reflects the paradox of auctions. If the auction is successful in producing a market-clearing price that will be relatively unchanged in the aftermarket, then there really isn't much incentive for investors to bid in the auction as opposed to buying later. As more investors bypass the auction the price becomes less informative. At an extreme, the price could be completely uninformative, but then there is an incentive to bid again. The tautology that arises means that the auction outcome is indeterminate; the price could be good, bad or somewhere in between.
Bookbuilding has an advantage over auctions that can actually make its offer price more informative. By forming a coalition, the underwriter can tap into the investors' information, which they have an incentive to produce if they are going to be buying shares, to set the offer price. The underwriter and institutional investors discuss the issuer's value during the marketing phase of bookbuilt IPOs. The investors may not submit price sensitive bids, but they make it known what will be an acceptable price. Because coalitions cannot be formed when auctions are used, the price is not guaranteed to be as informative. Investment banks often get criticized for their poor job of pricing IPOs. Yet in over 15,000 IPOs starting from 1960 to the present day, excluding the few hundred IPOs in 1999 and 2000 whose share prices almost or more than doubled on the first day, the average initial return is approximately 15 percent. The median is even lower at about 11 percent. These numbers are within an acceptable range of underpricing and demonstrate that bookbuilt prices are generally pretty good.
These arguments suggest why bookbuilt prices can be as if not more informative than the auction price. As for the claim that auctions generate more proceeds for the issuer because it minimizes underpricing, there are also valid reasons why this is not true either. For the reasons stated above, an auctioned IPO can draw weaker investor interest than the equivalent bookbuilt offering. The price that will clear the auction at the desired number of shares offered could then be less than the bookbuilt price, even after accounting for underpricing.
We can only speculate what would have happened to Google's IPO if they had gone the conventional route. Chances are they would have started with a much lower price range, perhaps similar to the final one (and they probably would have done a share split to get the price down to a normal level). The lower range would have generated far more positive excitement, especially with the certainty of underpricing. With greater investor interest it seems very plausible that the offer price would have been above the actual $85.
What's a CEO to do?
What are the key decision makers who will be taking their companies public in the future left to conclude from Google's offering? For starters, it is important not to infer too much from the IPO on how well auctions work because it truly was a one of a kind event. The IPO was one of the largest in U.S. history, even at the reduced offer terms, and no other issuer should expect to go public with as much brand awareness and hype as Google did. So the people who are pointing to this IPO as proof that auctions can work well and the others who are claiming the opposite are not being particularly objective in their assessment. Rather it is likely that people are staking out these positions because it is in their own self-interest to have their preferred method triumph.
At the end of the day the onus is still on auction proponents to make a convincing case. The investment bank W.R. Hambrecht has led 10 IPOs through its Open IPO auction process, but with only mixed success. The most vocal critics of bookbuilding routinely fall back on the experience of 1999 and the first half of 2000 as all the evidence they need to condemn this method. We all know by now the corrupt behavior of that time, but all the willing participants "banks, investors and issuers" must accept some responsibility. The problem with selectively pointing to the Internet bubble is that it ignores the effectiveness of bookbuilding prior to and even since this period. It also ignores the international evidence of IPO auctions failing and the fact that bookbuilding is now the dominant method world-wide because other countries believed it was superior to the other options they were using, which included auctions.
Just as some people are promoting auctions out of their own self-interest, the participation of investment banks and institutional investors in the IPO market is also contingent on it being in their best interest. The apparent highly profitable collusion that goes on between these two groups in bookbuilt IPOs at the expense of issuers and retail investors obviously implies that they have an interest in maintaining the status quo. Even if we take this facetious statement at face value, it is necessary to consider the alternative. If auctions were to become the dominant method, either because of regulatory changes or issuer insistence, why should institutional investors continue to devote resources to evaluating issuers without a guaranteed allocation and compensation in the IPO? The same reasoning applies to investment banks. With reduced fees and an inability to profit from the IPO through other means, the incentive to be in the market is diminished. The overall effect, as demonstrated by the experience in other countries, is that it would be harder for companies to go public. Venture capitalists also have to look out for their own self-interest and pushing an auction method that could ultimately hurt the vibrancy of the IPO market, which is vital to the industry's success, isn't the best course of action.
At this time the best advice for CEOs and VCs is that they continue to use the bookbuilding method. Going public is time consuming, expensive and rife with uncertainty about the final outcome. Auctions are still sufficiently unproven and have enough strikes against them to make them simply too risky for most issuers. Companies that are relatively well known, like Google or Overstock.com, should find it easier to attract investors to their auction, which would make this option more viable. But for a typical issuer, bookbuilding still offers the best guarantee of a successful IPO. Of course, it is important to keep in mind that a company's long-term success is far more dependent on its ability to execute a strong business plan than it is on its IPO method.