Ch. 7: IPO Underpricing 395
of bookbuilding. Positive price revisions presumably follow when informed investors
reveal positive information, and this is precisely when underwriters need to reward co-
operative investors with favorable allocations. At the same time, however, institutions
are given similar allocations in overpriced as in underpriced deals, which is consis-
tent with the prediction that underwriters ‘bundle’ IPOs over time and regular investors
sometimes are expected to buy ‘cold’ IPOs.
Aggarwal, Prabhala, and Puri (2002)analyze a more recent dataset covering 164
IPOs managed by nine different banks in 1997 and 1998. As inHanley and Wilhelm
(1995), institutional investors are allocated the lion’s share of IPO stock and institu-
tional allocations increase in the price revision relative to the filing range. Underpricing,
in turn, is larger the more stock institutions were allocated. This makes sense within
the Benveniste–Spindt framework, since underwriters likely use both price (i.e., under-
pricing) and quantity (i.e., allocation size) to ensure truthful revelation of particularly
positive information.
Ljungqvist and Wilhelm (2002)depart from the previous two studies by estimating
thestructurallinks between IPO allocations, price revisions, and initial returns. They
argue that these three variables are jointly determined, in the sense that the degree of
price revision depends on how much (positive) information investors reveal, which in
turn depends on their expected economic reward in the form of allocations of under-
priced stock. Using aggregate allocation data from France, Germany, the U.K., and the
U.S., they find that price revisions increase in institutional allocations and vice versa,
and initial returns increase in price revisions but decrease in institutional allocations.
The latter result suggests that constraints on the size of institutional allocations—which
are widespread in France and (during the early 1990s) in the U.K.—result in underwrit-
ers relying more on price than on quantity to reward truthful revelation. This is costly
to issuers, since blanket underpricing rewards both informed and uninformed bidders.
There is one key prediction of theBenveniste and Spindt (1989)framework that can
be tested without proprietary bid or allocation data. Revisions in the offer price and the
number of shares offered during bookbuilding likely reflect investors’ level of interest
and the aggregate nature of their information. An IPO for which positive information
is revealed should be priced towards the upper end of the indicative price range (or if
the information is particularly positive, above the range) whereas a less well received
offering should be priced towards the lower end. Benveniste and Spindt’s model sug-
gests that underpricing should be concentrated among the IPOs drawing the highest
level of pre-market interest. In other words, even though the underwriter adjusts the
price upwards, he does so only partially, in order to leave enough money on the table
to compensate informed investors for their truthful revelation.Hanley (1993)was the
first to provide empirical evidence of this ‘partial adjustment’ phenomenon. Numerous
subsequent studies have corroborated this finding, both in the U.S. and internationally.
Loughran and Ritter (2002)criticizeHanley’s (1993)interpretation of the partial ad-
justment phenomenon, by showing that underwriters, when setting the offer price, do
not fully incorporatepublicinformation in the form of pre-pricing returns on the market
index. (See alsoBradley and Jordan, 2002.) This appears to contradict theBenveniste–