How did you get interested in Dutch auctions?
Some of the work in my dissertation looked at dividends and stock repurchases. Conventional wisdom said that repurchases were done because of the higher marginal tax rate on dividends than on capital gains. The 1986 Tax Reform Act was passed while I was a graduate student at Stanford, and people were saying that it spelled the end of the repurchase. Well, it didn’t. There were actually more repurchases in 1987 than in 1986, leading me to think that we really didn’t understand repurchases. When I looked at repurchases, they struck me as such an interesting potential substitute and potential complement for a dividend payout.
As I was documenting the increased use of repurchases, I became aware of this new innovation, the Dutch auction. Firms didn’t start doing it until 1981, and they didn’t really start doing it until the late 1980s. So it was something the markets and firms were still learning about. I contacted the firms that had done Dutch auction repurchases and asked them if they would give me their confidential shareholder tendering responses. If the firms want to buy back shares in a Dutch auction, they solicit what ends up being a supply curve for their shares, and they don’t have to ever disclose the asks they observe. All they do at the end is to pick the price where supply equals demand.
I was able to get the majority of firms that had done the Dutch auction to give me that confidential information. It was important for a number of reasons. One, the heterogeneity of the shareholder asks was quite striking. Second, it allowed me to predict what the tendering responses would look like. The paper that I published in 1992 in the Journal of Finance was really the first study of Dutch auction repurchases and remains the seminal work on Dutch auctions.
My subsequent paper, published in 1999 in the Journal of Financial Economics, takes the analysis further. It recognizes that firms did not default to Dutch auctions and stop doing fixed-price tender offers. Instead, I saw firms choosing which auction is the better one in which to transact their shares, and I was able to predict whether a firm would use one repurchase method or the other. It’s like with the Google IPO: all of a sudden are we going to stop seeing firm commitment underwriting and see only Dutch auction IPOs? This paper suggests the answer is no. The repurchase research gave me a lot of insights about which firms might continue to do firm commitment underwriting for IPOs and which firms might instead lean toward the Dutch auction.
As an economist, I am very intrigued by the efficiency aspects of the Dutch auction. Early in my career I did some work demonstrating that the Dutch auction gives shareholders an incentive to tell the truth. That alone is a really interesting feature — the fact that you’re getting people to tell you what they really think the stock is worth. If you’re a company that wants to buy or sell shares, that’s incredibly valuable information.
If a firm asks shareholders to sell back their shares at a premium price, the ones who do are the ones who really don’t like the firm that much. So the firm has systematically gotten rid of the ones who like it the least and kept the ones who are more sanguine. In my 1991 RAND paper, I argue that a firm can deter hostile takeovers by skewing its population of shareholders toward a group more positive about the future valuation of the company. When you compare the fixed-price tender offer to the Dutch auction, the Dutch auction will provide the deterrence more efficiently because of how it sets the closing price. So if your primary goal is to spend a certain amount of money and get as much deterrence as you can, the Dutch auction will do it better.
But there may be other goals for the repurchase: it may be about signaling. The fixed-price tender offer provides a better signal that the firm’s really confident, willing to commit to a price — it doesn’t have to wait and ask shareholders what they think. That’s a stronger signal. The 1999 paper takes these ideas and then looks at firms’ behavior and asks, Can we predict which choice they’ll make? The answer is yes, and the research identifies why a firm would choose one versus the other.
It was in that context that I looked at the Google IPO. People have said this auction stinks and that it’s never going to happen again. Well, that’s not what my research suggests. In the early 1980s, the Dutch auction repurchases didn’t all go smoothly either. But over time the bankers learned about them, the markets and shareholders learned about them, the firms learned about them, and now they work beautifully for the right firms — not for every firm, but for the right firms. That’s what I’m looking for with regard to IPOs as well. I think that, like with the repurchases, we’re going to see some of both methods going forward.
What should firms be looking at when they’re deciding which auction to use to transact their shares?
A Dutch auction IPO lacks the intermediary role of the underwriter. So you want to ask yourself, In what circumstances is that intermediary role incredibly important? If a firm is not well known to potential buyers, then it needs the intermediary. It needs him to rummage up demand. It needs his certification and insurance. At the other extreme is a company like Google. People know what it is. It’s a brand name, something that people will think about putting in their portfolio. Maybe such a company doesn’t need the additional costs and benefits of the underwriter. Maybe instead it would rather have a more efficient mechanism, one where supply and demand meet more naturally.
Other considerations include things like, What do the shareholders look like? Who do we want them to be? How liquid do we think this block of shares is? How broad a market do we need to go to, to be able to absorb the block we’re talking about? Those kinds of considerations have proven to be very important in the repurchase decision and, analogously, should prove important in the IPO decision as well.
In the Hermes article, you mention that some people criticized the Google Dutch auction for excluding small investors.
The firm commitment underwriting commonly excludes small investors, because the underwriters determine how the shares can be allocated. Some people hoped that the Dutch auction would democratize the process. One of the problems with that idea is the fact that the SEC limits who can participate in an IPO, even in a Dutch auction, because there are certain requirements that preclude many small investors. But that has nothing to do with the Dutch auction — those are SEC requirements.
Similarly, Google chose to limit how much it disclosed in terms of forward-looking information reasons related to liability. Had it been more forthcoming, that might have made the stock more accessible to a smaller investor. Having said that, it looks like some small guys did get in. In that sense I think it was a step toward the democratization of the process.
The question remains whether democratization really is the goal of the IPO. Google stated it as one of its explicit goals. For many companies, I’m not sure that it is the primary goal. You pay a service cost per shareholder for proxies and annual reports and so forth, whether they hold one share or 100,000. And there are other reasons why maybe you want a sophisticated clientele holding large positions. In choosing the transaction method, a firm affects its clientele.
When you were trying to predict which firms would use the Dutch auction for repurchases, what was your hypothesis?
My basic hypothesis was based on the observation that when you look at the repurchase tendering responses, for some firms there wasn’t that much disparity and for other firms there was an incredible amount of disparity. If you think about the efficiency argument, if there’s a lot of disparity and I set a price in advance, I may really set it wrong. Instead I can let my shareholders’ actions tell me where supply equals demand. That efficiency gain is more valuable if there’s a lot of dispersion and uncertainty than if I knew for sure, because if I knew for sure, I could set the fixed-price tender offer at exactly the right price. Heterogeneity might be an argument for the Dutch auction, whereas signaling is an argument for the fixed-price tender offer.
Based on your research, did the Google Dutch auction play out more or less the way you expected it to?
More or less, yes. There were some unfortunate mistakes that obfuscated what was going on — the Playboy interview, some SEC issues, the timing of the IPO when the market happened to be soft — but these things would have occurred even if Google had done a firm commitment underwriting. Unfortunately, many people don’t understand the Dutch auction well enough to realize that those mistakes have nothing to do with the Dutch auction. For example, I have read a huge amount of press that says the fact that Google lowered the price is a bad thing. It’s actually quite the opposite. The fact that they can assess what their demand is and then set a price is the strength of the Dutch auction, not the limitation. But it requires a level of understanding of the auction that an unsophisticated investor wouldn’t have. But over time, just as investors now understand dividends and repurchases, people will understand this too.
Laurie Hodrick is professor of finance and economics at Columbia Business School. In recognition of her research on Dutch auctions, she won the Presidential Young Investigator Award from the National Science Foundation, the Smith Breeden Prize for Distinguished Paper in the Journal of Finance and the Roger F. Murray Award for Excellence from the Institute for Quantitative Research in Finance.