Can Google's IPO Make Everyone Happy?
from the might-be-tough dept
Back when the stories of Google shopping for investment banks first appeared, there was a lot of talk about whether or not they would go with W.R. Hambrecht and use a dutch auction IPO method. Suddenly stories started appearing from all the Wall Street publications talking about how terrible the dutch auction method was – despite the fact it seems like a pretty good way to accurately price shares in your company without leaving money on the table. However, all the big investment banks were scared silly that a dutch auction would leave them without a reason to collect their hefty underwriting fees – and without the ability to do a little share flipping themselves (and for their friends). Then, the news came out that Google basically signed up every investment bank that ever existed to underwrite their IPO – including W.R. Hambrecht. So, now speculation is beginning again about whether or not Google will offer some form of a dutch auction IPO. The article linked here suggests that they’re trying to work out a hybrid method – where Hambrecht can offer shares in a dutch auction, and the other banks get to do their thing. While they point out that this could give Google more leverage with the traditional banks, they don’t explain exactly how this would work in reality. I don’t see how you can price shares using two different methods (one by auction, and one by picking it out of thin air) considering the price needs to be the same.
Comments on “Can Google's IPO Make Everyone Happy?”
No Subject Given
wow, a link to fortwayne.com?
Pricing IPO shares - how it works
The price doesn’t legally *need* to be the same at all. What happens in a typical underwritten offering of this type is that the i-banks promise to pay Google (the “issuer” of the new shares) some amount for all the shares and then the i-banks bear the burden of selling the shares. (Of course, the most important part for the banks is the part where they keep 7% of the gross as their fee.)
Google could certainly sell its shares to all the banks at the same price and let them sell the shares at whatever price the banks can get. Google’s part of the deal is mostly over once they get the huge check from the underwriting banks. The banks then own the shares and can go sell them. As long as they don’t violate any securities laws, they can sell them for more or less, or even keep them in their own accounts.
So, Hambrecht could pay the same $9.30 to Google for shares that every other bank will sell at $10. Hambrecht can then auction off its shares starting at $9.30 and see what happens. It would be a dutch auction if Hambrecht agrees to pay Google 7% of whatever those shares eventually sell at, which would presumably be higher than $10. Figuring out who would pay more than $10 is easy once you think about the limited supply of shares and the belief of some people that the shares will eventually increase in value. If I can buy shares at $11 but I can’t get them at $10, that’s the same as me going out onto the open market and buying at $11 on the day of the IPO, except that I’m probably more likely to get the shares at $11 in the dutch auction if the IPO is a stereotypical Valley IPO with a nice first-day pop.
Re: Pricing IPO shares - how it works
Ah, very good points, but that’s not truly a dutch auction. The point of the dutch auction is to set the real market price so that the company knows what price to sell their shares at as well. In your version, the i-banks still set the actual price that Google would get, and then it’s just the allocations to (some) shareholders that goes at dutch auction. This pretty much defeats the real point of a dutch auction, though, it will make it easier for some retail investors to buy shares – though, most likely at very inflated prices.
Re: Pricing IPO shares - how it works
of course there doesn’t “legally” need to be the same price. But in terms of economics, there needs to be one price, because otherwise, people will buy at the cheaper one, sell at the higher, and the instant arbitrage opportunity exists for someone other than the banks to make money off of.
Nothing you’ve described is a Dutch auction. In a Dutch auction, the bidding starts high, and lowers when the demand for shares at that price drops. If Hambrecht agrees to pay Google 7% of whatever those shares are selling at, then there’s no Dutch auction. If they start by dropping the price below that of the institutional investor, then that’s not a Dutch auction either.
Dutch auctions
First, investment banks in the U.S. cannot resell the shares for more than the offering price. They may have to sell them for less but they are not allowed to sell them for more. This is why they set the price at the last possible moment, after the book has been filled, and then place all of the shares as quickly as possible, since they face downside risk but no upside potential. It also gives them an incentive to price a bit on the low side, although that certainly doesn’t explain most of the initial returns.
Second, “Dutch” auctions (more accurately, uniform price auctions – these are not Dutch auctions) have been used around the world, and they have not led to accurate pricing. The idea of an auction always setting the “true market price” only works in an independent private values setting, where everyone already knows his or her own valuation of the item, and that valuation is independent of what anyone else thinks. To see if this fits Google, ask yourself how much the shares are worth to you. $27.50? $12.25? $49.99? If you can’t answer instantly, without any thought or effort, and if your value isn’t permanently fixed regardless of what you see other people willing to pay, then the “auctions lead to perfect pricing” theories don’t apply. It takes effort to price an IPO, which means that there is a free rider problem, which means that a uniform price auction open to large numbers of people is unlikely to produce an accurate estimate of the aftermarket price. In practice, uniform price IPO auctions have led to highly inaccurate offering prices precisely because the method attracts short term investors that plan to flip the shares quickly.
Third, an auction is not the only way to open up access to large numbers of people. Google could use the method that is used is Europe and Asia, a method that, unlike auctions, has worked well in practice in many, many countries. Opening up the offering to everyone has nothing to do with pricing the offering – they’re two separate things.
Some of this is explained in the Wall Street Journal article by Raymond Hennessey from Feb. 17, 2004. If you want to check it out for yourself, look up what happened with the IPO auctions of Singapore Telecom, Argentina Telecom, Japan Railway East, Japan Railway West, Japan Tobacco or Global Securities (in Turkey).
Or just think about how a “Dutch” auction works – the highest bidder is first in line but doesn’t pay what he/she bids. So, if Google auctioned its shares, how could you be absolutely sure of getting shares? Easy – just bid really high. If you’re the only one that does this, it will work great. If there are lots of people that bid high because they all really want Google, you’ll all overpay. Oops!
Re: Dutch auctions
no, if lots of you are bidding high, none of you are overpaying.
but your point about the theory of perfect pricing in Dutch auctions being impossible is correct.
The whole reason why this model exists for Google to look at is because the whole Google community exists inside an academe bubble. The whole of the Google nerd-surge is fascinated by computer science’s new foray into economics. Game theory! Auctions! they think they’ve discovered how the world works. It’s only because auctions are in vogue right now in academic research circles in CS that this is even on the table. It appeals to those who think that the academe knows something the rest of the world doesn’t.
No Subject Given
There’s not necessarily an inconsistency in banks saying that “Dutch” auctions are a bad idea but still being willing to try them, if demand is strong enough. I think that all three of the following are true:
1) investment banks genuinely think that this pricing method is a bad idea;
2) they also don’t want to do anything that will lead to lower fees;
and
3) they’ll use whatever method they have to use to get a deal.
Yes, investment banks are greedy, but they also have to compete. “Dutch” online auctions have not been popular for IPOs, for seasoned follow-on equity offerings or for corporate debt because issuers won’t choose them, not because investment banks refuse to offer them. WR Hambrecht so far has done a grand total of 11 OpenIPOs (plus Google), 1 OpenFollowon (for Overstock, which also did an OpenIPO) and 2 OpenBooks (for Dow and Ford Credit).
There were three online “Dutch” auctions for corporate bonds in less than a week in August, 2000 (by Deutsche Bank, Bear Stearns and WR Hambrecht), with three investment banks racing to be first and proclaiming that this was the wave of the future. Since then, there has only been one more, because issuers refuse to use this method.
Do you know of any country where “Dutch” IPO auctions open to everyone have worked well? I can name many countries where they’ve flopped but haven’t been able to find even one where they’ve been consistently popular among issuers. Why? They’re too risky, and they don’t price accurately.