132 S. Dasgupta and R.G. Hansen
Other work has explored signaling aspects of Dutch auction repurchases relative to
fixed price tender offers (Persons, 1994) and relative to paying dividends (Hausch and
Seward, 1993). Persons (1994)considers a situation in which shareholders demand a
premium (perhaps due to capital gains tax frictions) to tender their shares, but this
premium varies across shareholders, resulting in an upward sloping supply curve. Re-
purchases are costly to existing shareholders because the tendering shareholders must
be offered a premium. Importantly, the slope of the supply curve is random. In a fixed-
price tender offer, the price is fixed, while the quantity of shares tendered adjusts to the
random slope of the supply curve; in a Dutch auction, exactly the opposite is the case.
If the manager intends to signal the true value by maximizing a weighted average of the
intrinsic value and the market value of the shares (as in the dividend signaling model of
Miller and Rock, 1985), fixed-price offers are more effective signals of the manager’s
private information; on the other hand, if the manager needs to buy back a specific num-
ber of shares to prevent a takeover threat, a Dutch auction is better as it guarantees that
the required number of shares will be tendered.
4.9. Auction aspects of initial public offerings (IPOs)
In the summer of 2004, the internet search firm Google completed the world’s largest
initial public offering to be conducted via an auction procedure. Google sold 19.6 mil-
lion shares at an offering price of $85 each, for a total of $1.67 billion raised. The
auction method used was a variant of the Wall Street Dutch auction, covered imme-
diately above. Initial public offerings of equity shares would seem to be excellent
candidates for an auction procedure: multiple units of the same item for sale, with un-
certainty over value and ability of a seller to commit to a sales method.
Interestingly, however, the evidence suggests that formal auctions are not favored as
a sales mechanism. Instead, the IPO procedure known as “bookbuilding” attracts most
of the market in regions where multiple sales methods can legally exist (Sherman, 2005;
Jagannathan and Sherman, 2006; Degeorge, Derrien and Womack, 2004). A fair amount
of theoretical work has been done to explore differences in sales mechanisms for IPOs as
well as issues within any one sales method. There is also a literature examining relative
performance of auctions versus other sales methods, for in some countries we do have
different sales methods co-existing.
In applying auction theory to IPOs, the place to start is the literature on uniform price,
multiple unit auctions. The main initial contributions here areWilson (1979), andBack
and Zender (1993). A recent contribution is byKremer and Nyborg (2004a, 2004b).
The reason this literature is so important is that it shows how simple auction analysis
yields the main underpricing result from IPO studies (that is, that the initial stock market
returns immediately after setting the IPO price are overwhelmingly positive).^42 The
auction models show in fact that uniform price, multiple unit auctions have a multitude
(^42) For a detailed account of various theories of IPO underpricing, seeChapter 7of Ljungqvist in this volume.