394 A. Ljungqvist
of future deal flow increases the investor’s incentive to co-operate with the bank today.
Since the authors have revealed the identity of their respective bank to me, I am able to
confirm that Cornelli and Goldreich’s bank is associated with substantially larger deal
flow.
Second,Benveniste and Spindt’s (1989)argument assumes that the bank has access to
a set ofinformedinvestors whose information it seeks to elicit with the help of favorable
allocations of underpriced stocks. The quality of the information it acquires is clearly
related to the quality of the investors it has access to. And it is not unreasonable to
assume that banks differ in the quality of their investor networks. Indeed, bids by U.S.
investors comprise only 1% of the sample in Jenkinson and Jones versus 13% in Cornelli
and Goldreich. In sum, it appears likely that Cornelli and Goldreich’s bank is both
more active and better connected and thus in a better position to extract pricing-relevant
information from investors.
No corresponding bookbuilding data are available for U.S. banks. Thus, whether
these European results can be generalized to the U.S. depends on how similar book-
building techniques are in Europe and the U.S.Ljungqvist, Jenkinson, and Wilhelm
(2003)provide evidence from 65 countries showing that the quality of bookbuilding—as
measured by the underpricing cost of inducing truthful information reporting—heavily
depends on whether a U.S. bank lead-manages the issue and on whether U.S.-based in-
vestors are targeted. Indeed, bookbuilding by non-U.S. banks targeted at their domestic
(non-U.S.) clients appears to provide no pricing advantage over fixed-price offerings
completed without bookbuilding.
Controlling for the fact that issuerschoosewhether to hire U.S. banks and have their
IPOs marketed to U.S. investors,Ljungqvist, Jenkinson, and Wilhelm (2003)show that
underpricing is reduced by 41.6% on average when U.S. banks and U.S. investors are
involved. This benefit doesn’t come free: U.S. banks charge higher underwriting fees
than do domestic banks. But on net, 73% of issuers would have been worse off had they
chosen local banks and local investors instead, in the sense that the resulting increase
in underpricing cost would have exceeded the savings on the underwriting fees. The
median firm switching to the ‘cheaper’ strategy would have suffered a reduction in net
proceeds of US$11.7 million. These findings are consistent with the prediction that
access to informed (U.S.) investors favors certain U.S. investment banks.
While no other datasets have yet matched the level of detail ofCornelli and Goldre-
ich’s (2001, 2003)andJenkinson and Jones’ (2004), several studies have usedaggregate
allocation data on the fractions of an IPO allocated to institutional and retail investors,
respectively. If institutions are more likely to be informed than retail investors, this
allocation split can be thought of as a crude approximation of the extent to which un-
derwriters favor informed investors in their allocation decisions.
Hanley and Wilhelm (1995), for instance, use a sample of 38 U.S. IPOs conducted
by a leading (unnamed) investment bank over the period 1983–1988. IPO allocations
clearly favor institutions over retail investors: institutions are allocated 66.8% of the
average IPO. Cross-sectionally, institutional allocations are larger the more the offer
price exceeds the midpoint of the indicative filing range established at the beginning