134 S. Dasgupta and R.G. Hansen
in the number of bidders caused by a mixed strategy equilibrium in the game of entry
into an IPO auction. If bidders are free to enter the IPO auction, then if there is some
cost to entering and some classes of bidders are ex-ante identical, the equilibrium in
the entry game has a probability of entry for at least some bidders; the result is un-
certainty over the actual number of bidders. Sherman claims that this uncertainty over
the actual number of bidders causes the IPO price to vary and in particular to vary in
its relation to a “true” underlying value. Sherman observes that this additional uncer-
tainty further worsens the “winners’ curse” and considerably complicates the optimal
“bid-shaving” calculation that is required when there is winners’ curse. She also shows
that each investor optimally collects less information in a uniform price rather than a
discriminatory auction, because of the free rider (moral hazard) problem in the uniform
price auction.^43
Sherman assumes that in the bookbuilding process, the underwriter can select the
number of investors to invite into an information-acquisition process; this makes the
bookbuilding process more like the first-price auction in terms of the variance of its
outcomes.Jagannathan and Sherman (2006)rely on this model to explain their findings
of a worldwide abandonment of IPO auctions in favor of bookbuilding; they also sup-
port the theoretical model with evidence on the variance in number of participants for
IPO auctions. The issues of number of bidders and information collection would seem
to be key in an optimal IPO pricing/allocation mechanism. One wonders, however, if
a slight twist on assumptions for the auction models—let the auctioneer control some-
how the selection of bidders, à laHansen (2001)would bring equivalence back to the
two mechanisms.Sherman (2005, p. 619)does note that “If the term “auction” is in-
terpreted in a broad sense, it is almost a tautology that an appropriate auction could be
designed for IPOs”. This exemplifies a general difficulty in building theoretical models
of two different institutions to explain their empirical performances: one can capture
the sense of institutional differences by making clear assumptions (e.g., the underwriter
can select the number of potential investors for bookbuilding but not for auctions) but
one is left wondering if the assumptions really do justice to what actually happens in
practice.
In anther recent attempt at comparing bookbuilding to auctions,Degeorge, Derrien
and Womack (2004)show that bookbuilding seems to dominate empirically (they look
at France, where for a time auctions and bookbuilding had roughly equal market shares,
but now auctions are virtually extinct) and they offer a justification for issuers’ pref-
erence for the bookbuilding method that is based not on the price performance of
bookbuilding but on the investment bankers’ preference for the method. While one
might understand why investment bankers prefer a method that creates more demand
for their services, the link to issuers’ interests is less clear. Degeorge et al. hypothe-
(^43) In the uniform price auction, since the auction price is set by the actions of bidders who have already paid
the information gathering and processing costs, there is an incentive for uninformed bidders to free ride and
jump in with a high bid.