JOUFWAL.

OF
Fhancd
ECONOMICS
ELSEVIER Journal of Financial Economics 41 (1996) 291-310

Ownership dispersion, costly information, and

IPO underpricing
James R. Booth*+‘, I,ena Chuab,c
“College of Business, Arizona State University, Tempe, AZ 85287, USA
bCollege of Business, University of Hawaii, Manoa, Honolulu, HI 96822, USA
‘World Business Department Thunderbird - The American Graduate School of International
Management, Glendale, AZ 85306, USA

(Received April 1992;final version received August 1995)

Abstract

We develop an explanation for IPO underpricing in which the issuer’s demand for
ownership dispersion creates an incentive to underprice. Promoting oversubscription
allows broad initial ownership, which in turn increases secondary-market liquidity.
Increased liquidity reduces the required return to investors. Broad initial ownership,
however, requires an increase in investor-borne information costs. These information
costs are offset through initial underpricing. Empirical results are consistent with initial
underpricing reflecting the level of ownership dispersion.

Key words: Initial public offerings; Underpricing; Ownership dispersion; Information

cost
JEL classi&ation: G24; G32

*Corresponding author.
The authors benefited from discussions with Cliff Smith (the editor), comments of Jay Ritter (the
referee),and presentations at the Securities and Exchange Commission, Tulane University, National
University of Singapore, Southern Methodist LJniversity, University of Missouri, liniversity of
Arizona, University of Alabama, University of Mississippi, and Hong Kong University of Science
and Technology. Discussions with Rex Thompson, Tim Opler, Clifford Smith, Eric Amel, Francis
Koh, Richard Smith, Hank Bessembinder, Kalok Chan, Mike Hertzel, Susan Woodward, Rob Neal,
Fred Furlong, Joe Williams, Paul Spindt, Jeff Coles, and Dan Deli were beneficial. Thanks to Jay
Ritter for providing part of the data used in this study. Part of the work done in this study was
completed while James Booth was a visiting scholar at the Federal Reserve Bank of San Francisco.

0304-405X/96/$15.00 0 1996 Elsevier Science S.A. All rights reserved

SSDI 0304-405x(95joo862-9
292 JR. Booth, L. Chua/Jouunal of Financial Economics 41 (1996) 291-310

1. Introduction

An extensive empirical literature documents large underpricing for initial

public offerings (1POs)l. Explanations for this phenomenon frequently rely on
the adverse-selection consequencesof information asymmetry as a motivation
for underpricing. The information asymmetry is assumed to exist among differ-
ent investors (e.g., Rock, 1986) or between issuer and investors (e.g.,Allen and
Faulhaber, 1989; Chemmanur, 1993). Many of these models argue that issuers
underprice to encourage investors to produce information. Studies of IPO
markets in which issuers observe information about the price elasticity of
demand for sharessuggestthat issuers focus on maximizing total proceeds at the
preferred ownership structure (e.g.,Levis, 1990).In these markets, issuers often
choose between a lottery and partially filling orders to achieve the target
ownership structure. [For example, Levis (1990) and McDonald and Jacquillat
(1974) note that for certain issues in Great Britain and France, respectively,
investors bid both price and quantity. They note that even with this additional
demand-curve information, issuers price the issue and distribute shares to
achieve the preferred ownership structure.] Previous explanations, however,
have not explicitly considered the impact of target ownership structure on initial
underpricing.
In this paper, we develop an explanation for IPO underpricing in which the
issuer’s demand for a broad ownership dispersion and a liquid secondary
market for the shares determines the equilibrium level of underpricing.’ Since
we assume investors produce costly information-about the firm, the issuer’s
demand for a liquid secondary market increases equilibrium initial under-
pricing. Since we also assume that investors value secondary-market liquidity,
total proceeds are maximized when the rate of increase in total information
costs (and equilibrium underpricing) equals the rate of increase in the total
market value of the firm.
To illustrate the role of ownership dispersion in initial underpricing, we start
with a simple information framework in which everyone has equal quality, but
noisy estimatesof value. Information is then produced in two ways: First, issuers
produce common-value information through the use of mechanisms such as
reputable underwriters and firm-commitment contracts. Next, investment
bankers promote the issue, and encourage an adequate number of investors to
investigate the issue and produce private estimates of value. Considering
the issuer’s preferred ownership dispersion and secondary-market liquidity,

‘See Ibbotson and Ritter (1994) and Smith (1986) for a summary of empirical findings.
‘In a related paper, Brennan and Franks (1995) assume the objective for dispersed ownership is to
retain control of the firm. While this objective is not inconsistent with our framework, it is not
considered here.
 J.R. Booth, L. ChuafJouvnal of Financial Economics 41 (1996) 291-310 293

investment bankers then set a final offer price based on the estimate of total
information costs and the expected demand for shares.The final offer price is set
together with expected initial underpricing so that investors recoup total in-
formation costs (if not on each issue, at least across multiple issues in equilib-
rium) and continue to participate in the IPO market.
We examine a sample of IPOs and find that, consistent with earlier studies,
underpricing is a positive function of proxies for information costs. Specifically,
we find that underpricing is negatively related to investment-banker prestige for
firm-commitment offerings, but unrelated for best-efforts issues. We find evi-
dence consistent with clustering issues resulting in information spillover, thus
lowering information costs. In turn, these lower information costs reduce the
cost of ownership dispersion and IPO underpricing. The benefits of information
spillover are generally higher for less well certified best-efforts issues. Overall,
these findings suggest the demand for ownership dispersion affects IPO under-
pricing.
The remainder of the paper is organized as follows. In Section 2, we present
a model of equilibrium underpricing when all investors initially possessequal
but incomplete information. In Section 3, we offer the empirical implications of
the model. We present the empirical tests and results in Section 4. The related
evidence is presented in Section 5. In Section 6, we provide a brief summary and
conclusions.

2. Ownership dispersion and underpricing

To provide a framework for our explanation of how issuers’ demand for

dispersed ownership affects IPO underpricing, we first focus on the information
structure in the going-public process. We establish the importance of informa-
tion production, and we illustrate how potential investors are compensated for
incurring information costs. Using information costs as the foundation, we then
demonstrate how ownership dispersion and corresponding secondary-market
liquidity determine the level of underpricing.
We start with the assumption that primary shares are issued through a firm-
commitment contract to finance a growth opportunity. We assumeinitially that
no ex ante information asymmetry exists between issuer, investment banker, and
investors, and that all initial estimates of value are noisy. Through the use of
mechanisms such as the use of a prestigious underwriter and a firm-commitment
contract, the issuer first produces common-value information about the shares.
The investment banker then carries out the due diligence processto improve the
estimate of value, and sets the offer price filing range in the preliminary
prospectus. With this improved estimate of value, the investment banker con-
ducts ‘road shows’ to market the issue, and encourages potential investors to
incur investigation costs. During the marketing phase, the investment banker
294 JR. Booth, L. ChuajJournal of Financial Economics41 (1996) 291-310

usesindications of interest to obtain an estimate of investors’ demand for shares.

This is then used to set the final offer price.
For an issuej, assume that an investor, i, can further improve his estimate of
the market price per share, E(V), for a cost, Ci. Investors choosing to incur this
cost compare their improved estimates of value with the final offer price, OP, to
determine whether to bid for shares3 All potential investors who incur informa-
tion costs (informed investors) become part of the potential-investor base. [This
is similar to the argument by Merton (1987) on the choice of rights versus
underwritten securities offerings.] Compared to those who choose not to incur
information costs (uninformed investors), informed investors are more likely to
participate in secondary-market trading and future offerings. The size of the
potential-investor base is important in promoting secondary-market liquidity,
as reflected in IPO listing requirements for major U.S. stock exchanges. Unin-
formed investors are assumed not to bid for shares for fear of the adverse-
selection consequencesof uninformed bidding.
In our framework, the number of investors encouraged to incur information
costs, n, is set by the issuer in conjunction with the investment banker. This
involves promoting information production sufficient to induce oversubscrip-
tion. Thus, the issuer achieves the initial ownership dispersion and secures
a broad potential-investor base for the firm’s shares in the secondary market.
We assume that investment bankers first solicit those potential investors with
lower information costs to investigate the issue. Subsequent investors who
investigate the issue incur a cost that increasesat an increasing rate. This means
that total information costs, C(n), represent a function with positive first and
second derivatives, i.e., K(n)/% > 0 and aC2(n)/an2 > 0.
To illustrate the benefits of promoting oversubscription, we first assume an
ownership baseof one shareholder. To achieve a preferred level of oversubscrip-
tion, the investment banker must first encourage an adequate number of
investors to purchase information. We start with a framework in which only one
bidder is successful,and is allocated one share. We assumeall potential investors
purchasing information choose to bid for shares. For convenience, we assume
also that all bidders have equal probabilities of winning. In this framework, issue
proceeds, R (here, equal to final offer price, OP), are maximized, with investors
recovering information costs, when
R = El/@*)- C(n*), (1)

3As in Wilson (1977) and French and McCormick (1954), we assume private estimates of value are
drawn independently from the same distribution, which is conditional on the actual value of the
shares. Since the improved estimates of value still contain uncertainty, investors also incorporate the
potential for a winner’s curse.
 J.R. Booth, L. ChualJournal of Financial Economics 41 (1996) 291-310 295

where
EV(n*) = investment banker’s estimate of value at the optimal level of oversub-
scription; this function is assumed to increase at a decreasing rate;
C(n*) = total information costs for all potential investors at the optimal level
of oversubscription; the function is assumed to increase at an increas-
ing rate;
n* = optimum number of potential investors purchasing information.
In equilibrium, maximizing proceeds involves calculating the estimated value
and setting the final offer price (OP), taking into account the number of potential
investors purchasing information, such that
[EV(n*) - OP] - C(n*) = 0. (2)
This occurs when the initial returns to the winning bidder equal the total
information costs for all potential investors.
The model developed here is similar to the results of Wilson (1977) and
French and McCormick (1984) in which, with a finite number of bidders, the
expected value of the winning bid is lower than the expected value of the asset. In
models of this type, potential investors enter the auction process until the
winner’s expected profit equals the sum of all bidders’ information costs.
Fig. 1 illustrates the optimum number of potential investors purchasing
information, n*, that maximizes issue proceeds while investors recover their
pre-bid information costs. In this framework, El/(n) increases with the level of
oversubscription, n. Two reasons are given for this assumption: First, value
increases when the issue is widely promoted, since the most optimistic investors
are more likely to be bidding for shares. This is similar to the argument
presented by Merton (1987). Second, by observing that the issue is being widely
marketed, investors estimate a certain level of liquidity for the shares in the
secondary market, and set their estimate of value accordingly. Incorporating
secondary-market liquidity into required return, and therefore price, is similar
to the arguments presented by Amihud and Mendelson (1986).
Total information costs, C(n), also increase with n, since more investors
purchasing information are expected to be compensated in equilibrium. Since an
increase in n leads to an increase in both EV(n) and C(n), offer price (OP) can
either increase or decrease,
OP = El/(n) - C(n), (3)
aOP/an = aEV(n)/an - K(n)/&. (4)
When aEV(n)/an > K(n)/&i, then aOPli3n > 0. OP increases with oversubscrip-
tion because the rate of increase in EV(n) is greater than the rate of increase
in C(n), i.e., the cost associated with informing the marginal investor.
When aEV(n)/an < aC(n)/i%, then aOP/& < 0, which means OP decreases
296 J.R. Booth, L. Chua/Journal of Financial Economics 41 (1996) 291-310

Information costs per share C(n)

Market value per share W(n)
Offer Price (Oq)

n* n’ Level of oversubscription
Fig. 1. The optimal level of oversubscription per one-share allocation in an IPO.
The figure illustrates that with increased competition for shares, estimates of value increase, but at
a decreasing rate. Assuming investment bankers first approach investors with the lowest costs for
investigating the issue, then total information cost C(n) will increase at an increasing rate. As a result,
the offer price (OPj) is maximized when the optimal level of oversubscription (n*) is achieved.

with oversubscription. OP is maximized at n* when aOP/an = 0, i.e., when

aEV(n*)/ih = X(n)/&. At n*, investment bankers have promoted oversub-
scription to the point where the benefit in terms of the increase in expected value
is equal to the information cost of a marginal investor. At n*, the issuer
maximizes total proceeds (here, equal to OP), subject to investors recovering
information costs through initial underpricing.
The importance of initial ownership dispersion on secondary-market liquidity
can be evidenced by firms listing their shares on one or more exchanges. To
promote secondary-market liquidity, stock exchanges impose listing require-
ments related to the minimum number of publicly held shares and shareholders,
as shown in Table 1.
To illustrate the impact of secondary-market liquidity on underpricing, we
now consider the initial ownership dispersion. For convenience, we assumethat
each successful bidder is allocated one share. We assume also that the invest-
ment banker has achieved the optimal level of oversubscription per share, y1*,as
described above, and that the benefits of broad dispersion and corresponding
 J.R. Booth, L. ChuaiJournal ofFinancial Economics 41 (1996) 291-310 291

Table 1
Exchange listing requirements related to the number of public shares and the width of distribution

Minimum number of Minimum number of

Exchange shares in public floata public shareholders

NYSE 1,100,000 2000

AMEX and NASDAQ NMS 1,000,000 400
or 500,000 800b
NASDAQ (small cap market) 100,000 300

Source: NYSE Fact Book 1988, AMEXFuct Book 1993, and NASDAQ Fact Book 1989.
“Exclusive of the holdings of officers, directors, controlling shareholders, and other concentrated or
family holdings.
bIf a firm maintains 500,000 shares but maintains a constant daily trading volume of 2,000 shares,
the minimum number of public shareholders is 400.

secondary-market liquidity increase at a decreasing rate. Potential investors

estimate the number of allocations and factor this into their estimate of second-
ary-market liquidity. Total information costs increase at an increasing rate with
the number of allocations, since the issue is assumed to be promoted to that
segment of the market with the lowest costs of acquiring information first.
Fig. 2 illustrates the optimal number of shares (ownership dispersion), S*,
consistent with issuers maximizing proceeds and potential investors recouping
pre-bid information costs across multiple issues. At S*,
EV(n*)l S - C(n*)I S (5)
is maximized. This occurs when
a[EV(n*)j S - C(n*)j S]/&S = 0
or
[aEv(n*)I syas = a[c(n*)I syas . (7)
When secondary-market liquidity is incorporated into potential investors’ esti-
mate of market value, issuers’ proceeds are maximized when the rate of increase
in total market value equals the rate of increase in total information costs
(assuming the optimal level of oversubscription per allocation). At distributions
greater than S*, the increase in total information costs exceeds the benefits of
increased secondary-market liquidity, resulting in lower total proceeds. At
distributions less than S*, the savings in information costs are offset by the
decrease in value due to lower secondary-market liquidity.
Since information is costly to produce, underpricing is used to compensate
investors for incurring pre-bid information costs in IPOs. Our model illustrates
that an optimal level of underpricing is reached when the issuer maximizes
298 J.R. Booth, L. ChuaJJournal of Financial Economics 41 (1996) 291-310

Total information costs C(n*,S)

Total market value EV(n:S) Cfn : S)

Total net proceeds

Tatal Initial

S’ S Number of shares

Fig. 2. The impact of ownership dispersion on IPO proceeds and total underpricing.
Assuming the optimal value of oversubscription per share (n*) and one-share allocations, the
optimal ownership dispersion to maximize total proceeds occurs at S*. If the issuer desires
a ownership dispersion greater or less than S*, then total proceeds are not maximized.

proceeds. Note that S” is partly determined by the value of the secondary-

market liquidity. Maximizing proceeds involves promoting secondary-market
liquidity through initial ownership dispersion and oversubscription, to the point
that the rate of increase in total market value equals the rate of increase in total
information costs.
Our discussion thus far has focused on firm-commitment issues. Next, we
consider the information flow in best-efforts issues. In best-efforts IPOs, the
investment bankers, together with the issuers,estimate the value of the assetand
set the final offer price. After setting the offer price, underwriters encourage
investors to purchase information, and accept nonbinding orders for shares
during the marketing period.
In best-efforts issues,investment bankers cannot adjust the offer price, which
is set early in the IPO process. Therefore, it is more difficult to enforce an
underpricing equilibrium that maximizes issue proceeds and compensates
potential investors for purchasing information. These issuesface a higher prob-
ability of excess(oversubscription) or insufficient demand (failure) during the
marketing period. With less common-value information associated with
investment-banker certification (seeSmith, 1986;Booth and Smith, 1986;Beatty
and Ritter, 1986), private information costs are likely to be higher. Higher
 JR. Booth, L. ChualJoumal of Financial Economics 41 (1996) 291-310 299

private-value information costs and a lower probability of success with the issue
contribute to higher equilibrium underpricing for these issues.

3. Empirical implications of the model

In our framework, an underpricing equilibrium is achieved so that issuers

maximize proceeds at the preferred ownership dispersion. Since potential inves-
tors recover their information costs through underpricing, our model predicts
that underpricing is negatively related to the probability of receiving an alloca-
tion.
We expect the cost of achieving ownership dispersion and corresponding
secondary-market liquidity to be larger for best-efforts issues, since these issues
are much smaller on average (see Ritter, 1987). Thus, as a percentage of
proceeds, investor-borne information costs are expected to be higher for these
issues. Additionally, in best efforts issues, investment bankers cannot adjust the
final offer price after obtaining indications of interest, as they can in firm
commitment issues. As a result, these issues fail and are withdrawn at a much
higher rate than the firm-commitment issues. All these reasons suggest why
best-efforts issues are more underpriced, even if total information costs are equal
in both contract types. Since investment bankers do not underwrite best-efforts
issues, their reputation is less important in these relative to firm-commitment
issues,
In our framework, the underpricing equilibrium relates to the issuer’s demand
for a dispersed ownership and a large potential-investor base. To gauge the
importance of dispersion, we examine the number of shares and offer price
relative to total proceeds. We find that the number of shares does not depend
systematically on the size of proceeds, but that offer price is a positive function of
total proceeds. If investor information production contains fixed costs and the
number of shares is fixed, then the offer price provides a reasonable proxy for the
cost of secondary-market liquidity.4 If so, equilibrium underpricing should be
negatively related to offer price. This proxy can have greater explanatory power
for best-efforts issues, for which information costs are higher.
To examine whether information spillovers lower the costs of estimating value
for individual IPOs, we examine the clustering of new issues in the market. We
explore this argument (put forward by Merton, 1987) empirically in this paper
by examining new-issue intensity by industry, and new issue intensity in the IPO
market as a whole. The idea is that similar issues are often clustered in time so as
to lower the total information costs of individual issues. We expect these smaller

4This is consistent with Christie and Schultz (1994)‘sargument that NASDAQ market makers tend
to have a $0.25 bid-ask spread independent of price.
300 JR. Booth, L. ChualJournal of Financial Economics 41 (1996) 291-310

Table 2
Descriptive statistics of issues and issuing firm characteristics for all IPOs from 1977 to 1988

Average Median Range

Gross proceeds/(issue size) $26,079,638 $11,982,517 $1.61m-1.89 bil

Total assets 42,221,840 8,622,OOO 0.86-7.67 bil
Age of firms (years) 9.38 5 O-85
Offer price $8.44 $7.75 $1-$35
Number of shares 1,705,795 1,150,000 O.lm-65m
Underpricir$ 13.1% 3.13% - 87.69%362.5%

Data for 1977784 are provided by Ritter (1987) and data for 1985-88 are from the Registered
Offerings Statistics tape provided by the SEC. Issues with offer price below $1 are deleted resulting in
2,151 total offerings.
“Number of years the firm is in business before going public.
‘Underpricing is measured as [(MP - OP)/OP] x 100, where iWP is the aftermarket closing price of
the first trading day recorded by CRSP and OP is the offer price.

information costs to be translated into lower underpricing. Since the informa-

tion costs and underpricing of best-efforts issues are larger, we expect these
issues to benefit the most from clustering.

4. Empirical results

4. I. Sample selection

Most studies of IPOs examine only firm-commitment issues.5 To increase the

power of the empirical tests of our model, we use both firm-commitment and
best-efforts issues. [However, we exclude issues with an offer price below $1 to
be consistent with earlier studies (e.g., Ritter, 1991).] Our sample is comprised of
Jay Ritter’s database for 1977-84 and data for 1985-88 from the Security and
Exchange Commission’s Registered Offering Statistics (ROS) tape. Where pos-
sible, data are checked for errors and verified against other public sources. The
following types of issues are excluded:

a) ‘direct’ offerings in which no investment bank was employed,

b) ‘stand-by’ underwritings associated with rights issues,
c) offerings of closed-end funds, investment companies, banks, and other
financial institutions,
d) unit offerings and offerings priced below $1.

‘See Carter and Manaster (1990) and Barry, Muscarella, and Vetsuypens (1991). A notable exception
is Ritter (1987).
 J.R. Booth, L. Chua/Jounzal of Financial Economics 41 (1996) 291-310 301

The sample is further limited to firms listed on the Center for Research in
Security Prices (CRSP) tapes for the NYSE, AMEX, and the NASDAQ stock
market. For issues from the ROS tape, we confirm that these issues are IPOs by
matching the offer dates with the first trading prices available on CRSP. The
final sample consists of 2,151 issues, of which 1,930 (89.7% are firm-commitment
issues and 221 (10.3%) are best-efforts issues.
Summary statistics for this sample of IPOs are presented in Table 2. The
average initial return for the sample is 13.1%. Firm-commitment issues average
11.9%, and best-efforts issues average 52.5%. These levels of underpricing are
comparable to those in earlier studies described by Smith (1986) and also found
by Ritter (1987) and Carter and Manaster (1990).
Mean and median issue sizes for the whole sample are approximately $26
million and $12 million, respectively, when measured in terms of 1994 purchas-
ing power. The mean size of firm-commitment issues is five times as large as
best-efforts issues, and the median size is three times as large. The data on the
total number of shares per issue suggest that the constraints for listing on
NASDAQ, AMEX, and NYSE, which specify the minimum number of publicly
held shares, are not binding for the vast majority of issues. The mean number of
shares is approximately 1.7 million for firm-commitment issues and 2.1 million
for best-efforts issues, respectively. Regressing issue size independently on both
offer price and the number of shares reveals that in a first approximation, as
issue size increases, firms raise offer prices but do not increase the number of
shares. This result is consistent with the notion that to promote liquidity, even
small issues are broadly distributed.

4.2. Empirical model

From our model, we expect cross-sectional differences in underpricing to be

related to measures of the pre-bid information costs adjusted for the expected
level of oversubscription, Following earlier studies, underpricing is measured as
the initial offer-day return. The dependent variable is:

Underpricing = (market close price on offer day - offer price)/offer price.

Independent variables used to control for the level of pre-bid information costs
are:

Size = natural log of total issue proceeds (expressed in terms of

1994 purchasing power) from the IPO,
Best = binary variable that takes a value of one for best-efforts
IPOs, zero if firm-commitment IPOs,
Ofprice = natural log of offer price of the issue,
Firm x Urank = underwriter rank for firm-commitment IPOs, ranges
from 1.12 to 10.37, zero otherwise,
302 JR. Booth, L. ChualJournal of Financial Economics 41 (1996) 291-310

Best x Urank = underwriter rank for best-efforts IPOs, ranges from 0.78
to 6.79, zero otherwise,
Best x OfSpprice = offer price for best-efforts issues, zero otherwise,
Best x Size = natural log of total proceeds for best-efforts IPOs, zero
otherwise,
Intensity = total number of IPOs in the three calendar months
immediately preceding the issue,
Best x Intensity = total number of IPOs in the three calendar months
immediately preceding best-efforts IPOs, zero otherwise,
IndLlntensity = total number of IPOs in the same industry in the 12
calendar months preceding the issue,
Best x IndLZntensity = total number of IPOs in the same industry in the 12
calendar months preceding best effort IPOs, zero other-
wise.

We develop Urank as a measure of underwriter reputation for the 481 lead

underwriters in our sample. We base Urank on the assumption that underwriter
characteristics are correlated with the prestige rankings provided in Carter and
Manaster (1990). We obtain a predicted ranking for each of the 117 investment
bankers analyzed by Carter and Manaster by regressing their ranking on the
natural log of total dollar proceeds raised and average offer price of issues
underwritten by each investment bank during the sample period.6 The correla-
tion between the predicted rankings and actual Carter and Manaster rankings is
0.81. The model is then used to rank the 365 investment bankers in our sample
that are not included in the Carter and Manaster study. The resulting ranking is
a continuous variable ranged from 0.78 to 10.37, with 10.37 as the highest
reputation ranking.
We expect investment-banker prestige to signal the quality of the issue and
reduce underpricing. From the previous IPO literature, investment-banker
prestige is expected to be more important for firm-commitment than for best-
efforts issues (seeBooth and Smith, 1986).To control for possible differences in
the role of investment-banker prestige across contract types, we interact Urank

‘jThe following model is fitted to the Carter and Manaster ranking (t-values in parentheses):
Urank = 0.9504 + l.O114E-09 (Total) + 0.3735 (Augop),
(2.96) (3.91) (10.315)
Adjusted R2 = 0.65,
where
Urank = Carter and Manaster’s ranking of 0 to 9,
Total = natural log of total gross proceeds of issues underwritten by the investment banker in
this sample,
Augop = average offer price of issues underwritten by the investment banker in this sample.
 J.R. Booth, L. ChuaJJouunal ofFinancin1 Economics 41 (1996) 291-310 303

with Best (a binary variable taking a value of one if the issue is marketed on
a best-efforts basis) and Firm (a binary variable taking a value of one if the issue
is marketed on a firm-commitment basis).
Size is the natural log of total proceeds of the issue, and serves as a control
variable. Since larger IPOs are generally easier to value, we expect this coeffi-
cient to be negative. Best is expected to have a positive coefficient, since
best-efforts issues have higher information costs and a lower probability of
receiving an allocation.
Previous studies suggest offer price (OfSprice) is a proxy for uncertainty about
value. Thus, as O&$-ice increases, the expected level of underpricing should
decrease. In our framework, Offprice proxies for the total information costs
incurred to achieve secondary-market liquidity. For issues with a low offer price,
these costs tend to be high, and so will the expected level of underpricing.
To control for the effects of offer price across contract types, we include
Best x Ofprice.
Intensity is a proxy for the rate at which new issues come to market
three calendar months prior to a particular issue. It is expected to reflect
the clustering of issues to lower information costs, and thus underpricing.
Therefore, we expect the coefficient for Intensity to be negative. To control for
differences in this measure for best-efforts IPOs, we interact Best with Intensity.
We expect the coefficient for this variable to have a negative sign, since the
benefit of clustering best-efforts issues to lower information costs should be
larger.
To account for potential industry effects associated with informational ex-
ternalities, we construct an industry (two-digit SIC code) issue-intensity
measure. To capture the within-industry potential information spillover,
we included new-issue intensity in the industry, IndLlntensity, for the
12 calendar months prior to each issue. If, as Merton (1987) suggests, there
are positive information spillover effects to clustering issues in the same
industry, then we expect the coefficient for Indlntensity to be negative. To
account for any differences in the importance of intensity for best-efforts versus
firm-commitment issues, we interact IlzdLintensity with Best. Since best-efforts
issues are argued to be less well certified, and thus contain higher residual
uncertainty than firm-commitment issues, industry information might be more
valuable for these issues. Thus, we expect the sign for Best x Inddntensity to be
negative.

4.3. Empirical jindings

We present regression results of initial-day underpricing on issue and invest-

ment-banker characteristics in Table 3. All regressions are corrected for hetero-
skedasticity using White’s (1980) procedure. The explanatory power of the
model, as measured by R2, is between 0.08 and 0.10. These figures are higher
304 J.R. Booth, L. ChualJournal of Financial Economics 41 (1996) 291-310

Table 3
Estimates of the relation between initial returns on IPOs and information cost proxies for oversub-
scription and new-issue intensity for a sample of 2,151 IPOs in 1977-1988 (r-statistics are in
parentheses)

Eq. 1 Eq. 2 Eq. 3 Eq. 4

Constant 21.42 26.43 22.73 26.55

(1.74) (2.15)** (1.79) (2.16)**
Size - 0.19 - 0.22 - 0.22 - 0.22
( - 0.21) ( - 0.26) ( - 0.26) ( - 0.26)
Best 60.75 46.26 45.56 40.25
(0.5 1) (0.40) (0.38) (0.35)
Offprice 0.52 0.52 0.50 0.52
(2.99)* (3.02)* (2.82)* (2.95)*
Firm x Urank - 2.62 - 2.59 - 2.59 - 2.59
( - 8.31)* ( - 8.29)* (- s.14)* (8.36)*
Best x Urank 5.63 5.06 5.61 5.13
(1.19) (1.10) (1.23) (1.15)
Best x Offpprice - 8.89 - 7.89 - 9.23 - 8.21
( - 3.43)* ( - 3.23)* (- 3.64)* ( - 3.38)*
Best x Size - 2.93 - 1.17 - 1.42 - 0.68
( - 0.36) ( ~ 0.15) (- 0.18) (- 0.09)
Intensity - 0.05 - 0.05
( - 4.67)* (- 4.71)k
Best x Intensity - 0.17 - 0.14
( - 2.75)* (- 2.16)**
Inddntensity - 0.04 - 0.01
(- 1.43) (- 0.19)
Best x Intintensity - 0.34 - 0.18
( - 2.43)** ( - 1.26)
Adjusted R2 0.08 0.10 0.09 0.10

The dependent variable is the initial return defined as [(MP - OP)/OP] x 100, where MP is the
closing price of the first trading day and OP is the offer price. All standard errors are corrected for
heteroskedasticity using the procedure described by White (1980).
Definitions: Size is the natural log of total proceeds (measured in terms of 1994 dollar). Best takes on
a value of 1 for best-efforts issues, 0 for firm-commitment issues. Oflprice is the natural log of offer
price of the issue. Rank is the rank of the underwriter, which is a continuous variable that ranges
from 0.78 to 10.37, where 10.37 is the highest prestige underwriter. Intensity measures the total number
of IPOs in the three calendar months immediately preceding the issue. Indlntensity measures the total
number of IPOs in the same industry during the 12 calendar months preceding the issue.
*Statistically significant at the 0.01 level. ** Statistically significant at the 0.05 level.

than recent studies that use only firm-commitment offers (see Carter and
Manaster, 1990).
The coefficients for Best and Size are positive and negative, respectively, but
are not statistically different from zero. The coefficient for Firm x Umnk, the
 J.R. Booth, L. ChuaJJournal ofFinancial Economics 41 (1996) 291-310 305

measure of underwriter prestige for firm-commitment IPOs, is significantly

negative. This is consistent with investment-banker prestige playing an
important role in reducing information costs for firm-commitment issues.
The positive but insignificant coefficient for Best x Urank indicates no relation
between investment-banker prestige and underpricing for best-efforts
issues. Previous studies note that prestigious investment bankers often
only underwrite on a firm-commitment basis (Ritter, 1987). To control for this
possibility, we repeat the analysis on a subsample of issues by investment
bankers involved in both types of contracts. The results for these unreported
regressions indicate the same relation between Urank and underpricing based
on contract type.
The coefficient for O&price in Table 3 is positive and statistically significant.
The coefficient for Best x Offppriceis significantly negative, indicating that as the
offer price increases, underpricing decreases at a faster rate for best-efforts IPOs
than for firm-commitment IPOs. This result is consistent with investors placing
more weight on the offer price as a signal of value for best-efforts IPOs. If initial
ownership dispersion is relatively fixed, Best x O&$-ice can also be proxying for
the size of potential allocations for these issues.
In regressions 2 through 4 in Table 3, we present the results related to the
information-spillover benefits of clustering IPOs. The coefficient for Intensity
(total IPOs in the last three months) is significantly negative, consistent with the
notion that information-spillover effects lower information costs when issues are
clustered in time. The coefficient on Best x Intensity is negative and statistically
significant at the 0.01 level. This suggests larger benefits to clustering issues
when information costs are high. These results are consistent with the behavior
of nonnatural resource issues in Ritter (1984). The findings are also consistent
with model predictions based on the level of pre-bid information costs, and
investment bankers not being able to adjust offer prices for updates on expected
over-or undersubscription.
In eqs. 3 and 4 of Table 3, we present results for industry information-spillover
effects. In eq. 3, the coefficient for IndLlntensity is negative but insignificant. The
coefficient for Best x IndLlntensity is negative and statistically significant at the
0.01 level. This indicates that a high industry intensity reduces information costs
and hence, underpricing for best-efforts issues. These results indicate that a high
industry intensity is consistent with benefits from clustering issues in the same
industry. However, the results change when we include both market-wide
intensity and industry issue intensity in eq. 4. After controlling for market-wide
intensity, we find that the industry issue intensity is not significant in explaining
underpricing of IPOs.
The results presented are consistent with pre-bid information costs playing an
important role in explaining IPO underpricing. When issues are clustered
in the IPO market, underpricing is reduced, consistent with lower information
costs.
306 JR. Booth, L. ChuaJJournal ofFinancial Economics 41 (1996) 291-310

5. Related evidence

A number of studies have attempted to test various models of IPO underpric-

ing. Most of these models, as shown in Table 4, focus on adverse-selection,
information-acquisition, or signaling explanations. Studies of ex ante uncer-
tainty and oversubscription find evidence consistent with the adverse-selection
explanation. Studies of partial-price adjustment find evidence consistent with
the information-acquisition explanation. Little evidence is found to support the
signaling explanation.
Models based on different information quality among investor groups ration-
alize that issuers use underpricing to encourage less informed investors to bid
against more informed investors (seeRock, 1982, 1986; Beatty and Ritter, 1986).
Thus, underpricing reduces the adverse-selection consequences for the less-
informed investors. The most direct test of this model is a study by Koh and
Walter (1989), who find evidence that underpricing is a positive function of
oversubscription. More specifically, they find large orders (i.e., a proxy for
informed demand) are more sensitive to underpricing than small orders. They
interpret this evidence as consistent with Rock (1986). Rock noted, however: ‘It
is not immediately clear what advantage accrues to the issuer from uninformed
participation.’ The Koh and Walter (1989) sample had an average oversubscrip-
tion of about 30 times the number of shares available, thus it appears that
informed demand would easily exhaust the supply of shares.
Results from Levis (1990) confirm those of Koh and Walter for IPOs in
Great Britain. These results provide insights into issuers’ demand for

Table 4
Empirical evidence related to IPO underpricing models

Dependent variable: Underpricing

Independent variables tested as potentially explaining the level of underpricing:

Ex ante uncertainty
Purpose and amount of proceeds (Ritter, 1984; Beatty and Ritter, 1986)
Investment banker prestige (Carter and Manaster, 1990)
Venture capital backing (Barry et al., 1990; Megginson and Weiss, 1991)
Use of warrants to underwriters (Barry, Muscarella, and Vetsuypens, 1991)
Oversubscription
Total value of orders relative to issue size (Koh and Walter, 1989; Levis, 1990; Cheung, Cheung,
and Ho, 1993)
Partial price adjustment
Difference between final offer price and filing range (Hanley, 1993; Hanley and Wilhelm, 1995)
Signalling quality
Probability of future issues (Jegadeesh, Weinstein, and Welch, 1993) Insider selling (Garfinkel,
1993)
 JR. Booth, L. ChualJournal of Financial Economics 41 (1996) 291-310 307

secondary-market liquidity through initial ownership structure. He reports that

oversubscription can be handled with a lottery within order size classes, as in
Singapore, or through partial allocations. He states that if issuers do not achieve
the preferred breadth of ownership through the lottery, they choose partial
allocations to increase the number of shareholders.
Brennan and Franks (1995) suggest that dispersed ownership may be used as
a corporate control mechanisms. They examine change in ownership from the
IPO to seven years later and conclude that insiders reduce ownership slowly
over time. They interpret this behavior as consistent with an attempt to retain
control after the IPO.
Our results also confirm that underpricing is a positive function of the level of
oversubscription. In our model, however, the targeted ownership dispersion
influences the number of potential investors to which the issue is marketed and
the level of oversubscription. Unlike previous models, our framework assumes
that potential investors value secondary-market liquidity and incorporate the
expected level of liquidity into their estimate of value. Secondary-market liquid-
ity increases with the level of oversubscription and ownership dispersion, which
in turn increases total information costs and underpricing. Greater oversub-
scription and broader dispersion effectively raises the potential-investor base,
and thus lowers current and potential future financing costs.
Koh and Walter’s evidence, drawn from the more-regulated Singapore IPO
market, suggests that allocation restrictions are designed to allow free riding by
small investors on potentially better information of larger bidders.7 Our model
suggests that this leads to greater underpricing, but can be viewed as an
alternative mechanism for assuring a sufficient number of bids to achieve the
desired ownership dispersion. In our model, free-riding by uninformed investors
is discouraged by adverse selection. For those ‘hot issues’ in which a free ride on
information spillover might be possible, investment bankers can use discretion-
ary allocation to ration shares. Prohibiting discretionary allocation can increase
total underpricing, due to the larger total information costs resulting from
higher-than-optimal oversubscription.
Based on the argument about the possibility of adverse selection declining
with less uncertainty about value, Beatty and Ritter (1986) show that under-
pricing is directly related to uncertainty about value. In our framework, under-
pricing is positively related to total information costs, which are a function of the
preferred ownership dispersion and potential-investor base. Since total informa-
tion costs increase with uncertainty about value, issuers who bear these costs
have the incentive to produce credible (but noisy) common-value information
about the issue. Faced with the possibility of a winner’s curse, potential investors

‘Starting in 1992, according to Loughran, Ritter, and Rydqvist (1994), this is easier to do in the
Singapore market. In a related paper for IPOs in Hong Kong, Cheung, Cheung, and Ho (1993) find
that allocation restrictions in that market results in small investors earning positive excess returns.
308 J.R. Booth, L. ChuajJournal of Financial Economics 41 (1996) 291-310

choose to investigate issue value. Additionally, we do not rely on information

cascading, as suggested in Welch (1992), to motivate underpricing although his
model is not inconsistent with ours.
In our model, similar to Benveniste and Spindt (1989), underwriters conduct
a pre-market survey in which they produce information and gather indications
of interest. Underwriters use the pre-market to lower private pre-bid informa-
tion costs incurred by investors. We assume reputable investors truthfully reveal
their demand. Even when they do not, disguising demand is unlikely to influence
the final offer price if there is no collusion among investors. [Related evidence by
Beatty and Ritter (1986) suggests mispricing by investment bankers damages
their reputation in this market. We effectively assume reputation extends to
potential investors, and that misstatin, 0 their demand would result in being
rationed in-future issues. Thus, we assume this prospect binds them to truthfully
reveal their demand. Again, this prospect becomes relevant only if by doing this,
the potential investor can affect the final offer price.]
Hanley (1993)‘s empirical evidence on partial adjustments of the final offer
price is proof that issuers update offer prices based on information on investors’
demand. She finds that a strong pre-market demand, as indicated by a final offer
price above the mid-point of the filing range, results in greater underpricing.*
This is consistent with our argument that greater underpricing is related to the
larger total information costs, which are associated with a higher level of
oversubscription and associated secondary-market liquidity. Consistent with
a lower level of oversubscription and lower total information costs, issues for
which pre-market demand is weak are adjusted downward in price and are less
underpriced.
McDonald and Jacquillat (1974) provide perhaps the most direct evidence on
the roles played by information costs, oversubscription, and preferred owner-
ship structure in IPO underpricing. They examine pricing and issuing proced-
ures for IPOs on the Paris Bourse, a market structure closer to the traditional
auction framework. In that market, after consulting with an investment banker
and the stock exchange, the issuer determines the minimum price for shares.
Bids for share quantities and prices are accepted for several weeks. After all bids
are collected, the issuer, investment banker, and exchange meet and estimate the
market-clearing price subject to the issuer’s preferred ownership structure,
which usually includes partial allocations. All orders are then filled at one final
offer price. Officials attempt to underprice issues by 3% to 5% to encourage
potential investors to continue to participate in the IPO market. This is
consistent with preferred ownership dispersion, secondary-market liquidity, and
information cost recovery all being factors in determining the offer price and the
level of underpricing.

‘This result is confirmed for a different sample by Hanley and Wilhelm (1995).
 J.R. Booth, L. Chua/Jowmal of Financial Economics 41 (1996) 291-310 309

6. Summary and conclusions

We develop an explanation for IPO underpricing in which the issuer’s

demand for ownership dispersion motivates underpricing and oversub-
scription. In this framework, promoting oversubscription allows broad
initial ownership dispersion, which in turn achieves a liquid secondary
market for the shares. Since both increase investor-borne information costs, they
promote higher equilibrium underpricing. However, as investors also value
secondary-market liquidity, promoting oversubscription can also maximize
issue proceeds.
Using a sample of firm-commitment and best-efforts IPOs from 1977 through
1988, we find support for the importance of information costs in underpricing.
Best-efforts IPOs are; on average, more underpriced than firm-commitment
issues. Underwriter reputation plays a more important role in firm-commitment
IPOs than in best-efforts IPOs. Information spillover from clustering issues
results in lower IPO underpricing.
Our empirical results suggest that underpricing is a positive function of
ownership dispersion in the presence of costly information. Wide dispersion of
ownership suggests that average initial returns may be more reasonable than
previously thought. Our explanation, however, is not mutually exclusive of
previous explanations of underpricing based on adverse selection, information
collection, or signalling due to information asymmetry. Conclusive evidence
about the relative importance of alternative explanations for underpricing
remains to be developed.

References

Allen, Franklin and Gerald R. Faulhaber, 1989, Signaling by underpricing in the IPO market,
Journal of Financial Economics 23, 303-323.
Amihud, Yakou and Haim Mendelson, 1986, Asset pricing and the bid-ask spread, Journal of
Financial Economics 17, 223-249.
Barry, Christopher B., Christopher J. Muscarella, and Michael R. Vetsuypens, 1991, Underwriter
warrants, underwriter compensation, and the costs of going public, Journal of Financial
Economics 29, 113-135.
Barry, Christopher B., Christopher J. Muscarella, John W. Peavy III, and Michael R. Vetsuypens,
1990, The role of venture capital in the creation of public companies: Evidence from the
going-public process, Journal of Financial Economics 27, 447-471.
Beatty, Randolph P. and Jay R. Ritter, 1986, Investment banking, reputation, and the underpricing
of initial public offerings, Journal of Financial Economics 15, 213-232.
Benveniste, Lawrence M. and Paul A. Spindt, 1989, How investment bankers determine the offer
price and allocation of new issues, Journal of Financial Economics 24, 343-362.
Booth, James R. and Richard L. Smith III, 1986, Capital raising, underwriting and the certification
hypothesis, Journal of Financial Economics 15, 261-281.
Brennan, Michael J. and Julian Franks, 1995, Underpricing, ownership and control in initial public
equity securities in the U.K., Working paper (London Business School, London).
310 JR. Booth, L. ChualJouvnal @Financial Economics 41 (1996) 291-310

Carter, Richard and Steven Manaster, 1990, Initial public offerings and underwriter reputation,
Journal of Finance 45, 1045-1068.
Chemmanur, Thomas J., 1993, The pricing of initial public offers: A dynamic model with informa-
tion production, Journal of Finance 48, 2855304.
Cheung, Yan-leung, S. L. Cheung, and Richard Y. Ho, 1993, Listing requirements, uncertainty, and
underpricing of IPOs, Working paper (City Polytechnic of Hong Kong, Hong Kong).
Christie, William and Paul Schultz, 1994, Why do NASDAQ market makers avoid odd eight
quotes?, Journal of Finance, forthcoming.
French, Kenneth R. and Robert McCormick, 1984, Sealed bids, sunk costs, and the process of
competition, Journal of Business 57, 47-441.
Garfinkel, Jon A., 1993, IPO underpricing, insider selling and subsequent equity offerings: Is
underpricing a signal of quality?, Financial Management 22, 74483.
Grinblatt, Mark and C.Y. Hwang, 1989, Signalling and the pricing of unseasoned new issues,
Journal of Finance 44, 393-420.
Hanley, Kathleen Weiss, 1993, The relationship of the offer price to the preliminary file range and
the use of the overallotment option in initial public offerings, Journal of Financial Economics 34,
231-250.
Hanley, Kathleen Weiss and William J. Wilhelm,1995, Evidence on the strategic allocation of initial
public offerings, Journal of Financial Economics 37, 239-257.
Ibbotson, Roger G. and Jay Ritter, 1994, Initial public offerings, in: Handbook of operations
research and management science: Finance (North-Holland, Amsterdam).
Jegadeesh, N., M. Weinstein, and I. Welch, 1993, An empirical investigation of IPO returns and
subsequent equity offerings, Journal of Financial Economics 34, 153-175.
Koh, Francis and Terry Walter, 1989, A direct test of Rock’s model of the pricing of unseasoned
issues, Journal of Financial Economics 23, 251-272.
Levis, M., 1990, The winner’s curse problem, interest costs, and the underpricing of initial public
offerings, Economic Journal, 76-89.
Loughran, T., Jay Ritter, and K. Rydqvist, 1994, Initial public offerings: International insights,
Pacific-Basin Finance Journal 2, 165-199.
McDonald, John G. and Bertrand C. Jacquillat, 1974, Pricing of initial equity issues: The French
sealed-bid auction, Journal of Business 47, 37-47.
Megginson, W. and K. Weiss, 1991, Venture capitalist certification in initial public offerings, Journal
of Finance 46, 879-903.
Merton, Robert C., 1987, A simple model of capital market equilibrium with incomplete informa-
tion, Journal of Finance 42, 4833510.
Ritter, Jay R., 1984, The ‘hot’ issue market of 1980, Journal of Business 32, 2155240.
Ritter, Jay R., 1987, The costs of going public, Journal of Financial Economics 19, 269-282.
Ritter, Jay R., 1991, The long-run performance of initial public offerings, Journal of Finance 46,
3-27.
Rock, Kevin, 1982, Why new issues are underpriced, Unpublished Ph.D. dissertation (University of
Chicago, Chicago, IL).
Rock, Kevin, 1986, Why new issues are underpriced, Journal of Financial Economics 15, 187-212.
Smith, Clifford, 1986, Investment banking and the capital acquisition process, Journal of Financial
Economics 15, 3-30.
Welch, Ivo, 1992, Sequential sales, learning, and cascades, Journal of Finance 47, 6955732.
White, H., 1980, A heteroskedasticity-consistent covariance matrix estimator and a direct test for
heteroskedasticity, Econometrica 22, 817-838
Wilson, Robert, 1977, A bidding model of perfect competition, Review of Economic Studies 4,
511-518.