Going, Going, Google
By Barry Nalebuff and Ian Ayres
Originally published in The Wall Street Journal, August 20, 2004
As with first impressions, you never get a second chance to make an initial public offering. Google just squeaked by with its IPO and it wasn't pretty. The Playboy interview that attracted the SEC's scrutiny and the unregistered options didn't help generate confidence in the process, especially when Google couldn't speak to defend itself. But now it's a public company and the folks there can turn their attention to making the business really worth $23 billion -- make that $27 billion after yesterday's pop.
In spite of the difficulties, we very much admire Google's "why not?" approach to employing an auction to set the share price. There are plenty of vested interests -- most investment banks, for starters -- who were eager to decree the auction approach a failure and revert back to their business-as-usual approach. Mutual funds were happy to trash-talk the IPO in order to discourage other bidders. If Google could just barely pull off an IPO auction, what hope is there for the mere mortal company?
The fault lies not in the auction, but in how Google chose to run it. For someone who dares to follow in Google's tracks, here are some suggestions for how to do it better:
For starters, you have to make the bidders' job easier. We don't just mean simplifying the bidder-registration process. Bidders have a very hard time when they don't know how long the bidding will last or how many shares will be sold. Remember that the auction is supposed to find the "market" price -- as opposed to following the investment banker's Soviet-style IPO approach of setting the price via fiat. But if a firm gets to extend an IPO auction until the number of bids allows it to reach its target price, then it would be hard to conclude that supply and demand really determined the price.
When bidders know that the number of shares can be cut, that's another cause for alarm. Imagine that you offered $85, a price you thought was a safe bid. Then you were told that Google had reduced its IPO to one share and that you were the one lucky high bidder. How lucky would you feel? You'd know that everyone else had bid below you. Do you think you've overpaid? This problem is known as "winner's curse" and you have to bid extra low to protect yourself.
In this case, winner's curse was exacerbated by the fact that the Google auction invited all of the potential buyers, but not all of the potential sellers. Those who thought even $85 was too high a price had to wait until the stock started trading before they could take this position. If Google wanted to find the true market-clearing price, they could have allowed bidders to sell as well as buy shares in the auction, just as they do in the regular market. Or they could have allowed trading on a when-issued basis (as occurs with Treasury auctions) so there is a market price even before any shares are sold.
A solution to these issues is greater transparency. Google could have pre-announced a closing time and made a commitment to a number of shares. It could also have provided information along the way about where the bidding stood. And it could even have allowed short sellers to take a when-issued short position and thus compete with Google to be sellers in the stock.
There's yet another feature that made this auction more like a tall skinny half-caf latte than a cup of joe. Google had the ability to purposefully underprice the stock and then ration the shares. And that's what they did. They took two steps back to $85 so they could take one step forward. They knew there was excess demand at $85 and so they left money on the table to engineer a pop in the initial price. The prospectus said Google might do that and so bidders had to guess the likelihood that Google would underprice when figuring out how aggressively to bid.
As to why Google might want to leave money on the table, recent work in behavioral economics and finance offers an explanation. People value something more when they can't have it. Thus if something is rationed, its perceived value goes up. Moreover, when an IPO is rationed, there is a stock of unsatiated buyers who can offset the short-sellers when the regular market trading commences. Thus underpricing combined with some rationing can lead to an increase in demand. Even if the company is giving some money away, they know they are about to come back to the well -- and in Google's case, come back especially soon -- and thus want to have some pent-up demand.
Perhaps the biggest paradox associated with running an auction is that no one should participate in it. If the point of the auction is to arrive at the "true" market price, then there will be no excess return on Google's stock. Hence there is no special reason to go through all the effort to read the prospectus, sign up for an account and then devise a bidding strategy. You won't get rewarded for these efforts. Of course -- and here's the paradox -- if no one did any of this work, then there wouldn't be any information behind the stock price and thus we couldn't conclude that the price that resulted from the auction was efficient.
The need to reward people for showing up was made painfully clear in the Netherlands, where the government lost several billion guilders in July 2000 from the way they set up their "Dutch" auction to sell spectrum for next-generation cell phones. They offered up only five licenses to a market with five incumbents. Potential entrants to the market decided that the game would be locked up by the incumbents and so it wasn't worth playing. There was only one other bidder besides the incumbents and the auction was almost over when it started.
* * *
All this may seem to suggest a rationale for the status-quo IPO: Investment bankers and their favored clients get rewarded for doing the research to learn about newly public companies. Well, not quite. Their friends just get rewarded for being good clients or good friends.
A better solution to incentivizing bidders is to democratize access to the underpriced securities the way Google did. You can think of this as Google saying to potential investors: We're starting off with a going-into-business sale to give you an extra reason to make the trip out to our IPO. And we won't let professional buyers take all the great deals before you arrive. The problem was that Google didn't promise to do this and so bidders couldn't count on it.
At the end of the day, we think auctions are still underutilized. And not just for IPOs. Houses in Australia are commonly sold via auction, and the result is more liquidity in the market and lower real-estate commissions. But it is not as if one can simply wave the magic auction wand and count on getting an efficient result. One needs to get the rules right.
Google teaches us that going once, going twice, still going twice, is not so nice when it comes to running an auction.