- Permanent Link:
- Why Are Initial Public Offering (IPO) Prices Only Partially Adjusted?
- INCE, OZGUR S. ( Author, Primary )
- Copyright Date:
- Subjects / Keywords:
- Initial public offerings ( jstor )
Investors ( jstor )
Market prices ( jstor )
Natural logarithms ( jstor )
Price efficiency ( jstor )
Price level changes ( jstor )
Prices ( jstor )
Pricing ( jstor )
Public information ( jstor )
Tradeoffs ( jstor )
- Source Institution:
- University of Florida
- Holding Location:
- University of Florida
- Rights Management:
- Copyright Ozgur S. Ince. Permission granted to University of Florida to digitize and display this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
- Embargo Date:
- Resource Identifier:
- 496613760 ( OCLC )
This item has the following downloads:
WHY ARE INITIAL PUBLIC OFFERING (IPO) PRICES ONLY PARTIALLY
OZGUR S. SINCE
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
Ozgur $. Since
I would like to dedicate this dissertation to my parents, Filiz and Metin Ince; my brother,
Ufuk Ince; and Elise Chandon.
I would like to thank Jay R. Ritter, my dissertation committee chair, for his
guidance and invaluable comments. I also wish to thank Christopher James, Chris
Janiszewski, Jason Karceski, Richard Pettway, and seminar participants at the University
of Florida, Virginia Tech, Koc University, and ES SEC for their helpful comments.
TABLE OF CONTENTS
A C K N O W L E D G M E N T S ......... .................................................................................... iv
LIST OF TABLES ......... ...... ............................... ................. ............ vii
LIST OF FIGURES .......... .. ........................... ................... viii
A B S T R A C T ............. .. ................. ............... ................................... .. ix
1 IN T R O D U C T IO N ............................................................................. .......... .. .. ... 1
2 PARTIAL ADJUSTMENT PHENOMENON, HYPOTHESES, AND
EM PIRICAL PRED ICTION S ......................................................... ..................... 8
The Dynamic Information Acquisition Hypothesis.................................................9
The B bargaining H ypothesis ...................................................... ........ ............... 11
The Tradeoff H ypothesis .................................................. .............................. 16
3 M E T H O D O L O G Y ........................................................................... ....................18
Estim ation of the A djustm ent R atio .................................... .................................... 18
Sam ple Selection Bias ............................................................ 22
4 D A T A ................................................................................................................... 2 6
D ata S o u rc e s ............................................................................................................... 2 6
V variable C on struction ....................................................................... ....................26
5 EM PIRICA L AN ALY SIS ................................................ .............................. 33
Estimation of the Adjustment Ratio ................. .............................. ...............33
R egression A analysis ............... .... .. .............................. ... ............. ........... ...38
The Impact of Public Information on the Value and Pricing of IPO Firms ........38
Multivariate Analysis of the Efficiency of the Offer Price Adjustment Process 40
6 R OBU STN E SS ........._ ............ .............. .. .. ......... .. ............ 56
Endogeneity of Underwriter Selection ................................................. 56
Stability of the First-Stage Regression ...................... ..............56
Forw ard-Looking B ias ............................................................................. 57
Scale Effect on Efficiency ................................................ .............................. 57
Investor Sentim ent ................................................. ...... .............. 58
7 C O N C L U SIO N ............. ............................................................ ....... .... 59
L IST O F R E FE R E N C E S ............................................. ............................. ...................63
BIOGRAPHICAL SKETCH ........... ...... ..... ........... .. ...............66
LIST OF TABLES
4-1 D definition of V ariables. ................................................. .....................................3 1
5-1 Regressions of First-Day Returns on IPOs Categorized by Price Adjustment ........50
5-2 Descriptive Statistics for the Adjustment Ratio ....................................... .......... 51
5-3 Regressions of Percentage Offer Price Revision and Percentage Change in Firm
Value during the Registration Period on Public Information Measures ..................52
5-4 System Estim action .......................................................... .. .. .......... 53
LIST OF FIGURES
5-1 Mean and Median Adjustment Ratio by Cohort Year.........................................55
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
WHY ARE INITIAL PUBLIC OFFERING (IPO) PRICES ONLY PARTIALLY
Ozgur S. Ince
Chair: Jay R. Ritter
Major Department: Finance, Insurance, and Real Estate
Initial public offerings with upward offer price revisions leave disproportionately
high amounts of money on the table. In this study, we investigate the factors that
determine the efficiency of IPO offer price adjustment process. We conduct an efficiency
analysis by comparing the magnitude of actual offer price revision to the hypothetical
complete level of offer price revision for each IPO in our sample.
The analysis of 1,122 IPOs with upward price revisions during 1985-2003 indicates
that on average only 42% of the change in firm value during the registration period was
incorporated into offer prices, with this ratio dropping to 29% during the bubble period of
1999-2000. We document substantial variation in the efficiency of offer price adjustment
across IPOs and over time, and test three hypotheses concerning the incorporation of
information into offer prices.
Our findings do not support the dynamic information acquisition hypothesis of
Benveniste and Spindt (1989). We find substantial inefficiencies in the incorporation of
market returns and information spillovers from recent IPOs into offer prices. More
remarkably, the incorporation of private information is more efficient than the
incorporation of public information. The cross-sectional analysis of the offer price
adjustment efficiency reveals that the partial adjustment phenomenon is primarily due to
agency problems between the issuing firms and the underwriters, and the magnitude of
the offer price revision is strongly related to the relative bargaining powers of the two
Beginning with Hanley (1993), various empirical studies have documented that
when bookbuilding is used, the underpricing of IPOs is positively related to the offer
price adjustment. Recent studies have shown that the magnitude of offer price adjustment
is the single biggest predictor of IPO underpricing. This relationship is economically
significant: During 1985-2003, average underpricing has exceeded 57% for issues with
upward price adjustment, whereas it has been only 4.6% for issues with downward price
adjustment.1 During 1999-2000, this difference was even more dramatic: Average
underpricing was 120% for IPOs with upward price adjustment versus 8% for those with
downward price adjustment.
This positive relationship between offer price adjustment and underpricing is
primarily due to the partial adjustment of offer prices to the change in the value of the
issuing firm during its registration period. In our sample, IPOs with upward price
adjustment have experienced first-day closing prices on average 113% higher than the
midpoint of their initial filing range, while the average offer price increase for these IPOs
has remained at a relatively small 29%. During 1985-2003, an aggregate of $113 billion
1 In the rest of this article we use the term "upward price adjustment" for issues where the final offer price
is above the maximum of the original filing price range, and "downward price adjustment" for issues where
the final offer price is below the minimum of the original filing price range.
was left on the table due to underpricing.2 Approximately 76% of the total money left on
the table was from the 25% of IPOs whose offer prices were adjusted upward.
In this study we conduct a comprehensive investigation of the efficiency of the IPO
offer price adjustment process. We start by estimating an efficiency ratio for each IPO
with offer price adjustment by comparing the change in the firm value during the
registration period with the revision in the offer price. This efficiency ratio captures
actual offer price adjustment relative to the hypothetical complete level of offer price
adjustment. The change in the firm value during the registration period is unobservable
since the expected value of the firm at the time of the initial filing is unknown. In this
study, we estimate the initial expected value of the IPO firm using the observed market
price once it starts trading, the initial file price range, and the underpricing information
provided by IPOs whose offer prices are not revised. The primary advantage of our
methodology lies in its ability to identify the relation between the efficiency of the offer
price adjustment process and various firm-, underwriter-, and offer-specific
characteristics. Previous studies of IPO pricing efficiency have for the most part focused
on underpricing, which is a combination of two components: the part that is independent
from the offer price revision and common to all IPOs, for instance the initial return
required by investors for investing in the shares of the issuing firm; and the additional
underpricing due to the partial adjustment of offer prices. To the extent that the same
factors are systematically related to both components, using underpricing as the measure
of inefficiency leads to ambiguity in pinpointing the determinants of the IPO offer price
adjustment efficiency. Our approach allows us to provide a comprehensive analysis of the
2 Money left on the table is defined as the extra offer proceeds that would be attained by the issuing firm if
the shares were sold at the first-day closing market price rather than at the offer price.
cross-sectional determinants of the offer price adjustment efficiency, and to test three
hypotheses concerning the incorporation of information into the offer prices.
Our analysis reveals that from 1985 to 2003, on average only 42% of the increase
in firm value during the registration period was incorporated into offer prices by means of
an upward adjustment. In the bubble period of 1999 and 2000, this percentage dropped to
29%. Translated into dollars, conditional on an upward offer price adjustment, the
average IPO firm in our sample with $100 million initial expected offer proceeds left
$23.2 million on the table due to incomplete adjustment, equivalent to more than three
times the initial expected commission payment to the underwriters.
Next, we relate the efficiency of offer price adjustment to various firm-,
underwriter-, and offer-specific characteristics, and investigate which hypothesis of IPO
pricing best explains the variation of efficiency across IPOs with upward price revision.
Until recently the prevailing explanation for the partial adjustment of IPO offer prices
was based on the Benveniste and Spindt (1989) model of IPO pricing. This dynamic
information acquisition framework is based on the notion that potential investors have an
information advantage over the issuers and the underwriters with respect to the market
value of the issuing firms. Underwriters use underpricing as a reward to extract this
private information from the investors. As a result, underpricing and offer proceeds for
the issuers increase simultaneously, resulting in higher average underpricing for IPOs
with upward offer price adjustment.
Our findings do not support the dynamic information acquisition hypothesis. We
find that market returns and information spillovers from recent comparable IPOs (those in
the same Fama-French (1997) industry) are only partially incorporated into the offer
prices. Issuing firms that go public following recent comparable IPOs with high levels of
underpricing and during a period of high returns in the market experience less efficient
offer price adjustments. Our results indicate that, even though upward offer price
revisions in the IPO market are associated with higher market valuations for all
comparable IPOs that soon follow, this information is incorporated into the offer prices
efficiently only for offers taken public by the same underwriter. We interpret this finding
as evidence that the underwriters that are associated with upward offer price revisions in
the recent past are likely to adjust offer prices relatively more efficiently in their future
IPOs. Notably, we find that the incorporation of private information is more efficient than
the incorporation of public information. It appears that an offer price increase in excess of
that driven by public information is due to the issuer's ability to efficiently extract a
greater surplus from the underwriter and the investors, rather than due to exogenous and
costly firm-specific information production by the investors.
This is not the first study to question the dynamic information acquisition
explanation of the partial adjustment phenomenon. This explanation has recently come
under criticism, primarily due to the positive relation documented between market returns
during the registration period and the subsequent underpricing (e.g., Loughran and Ritter,
2002; Lowry and Schwert, 2004). In light of this finding, two new hypotheses of offer
price adjustment have emerged.
The bargaining hypothesis is based on the notion that there are significant agency
problems between the underwriter and the issuing firm, and suggests that the offer prices
are determined as a result of negotiations between the two parties (Loughran and Ritter,
2002; Ljungqvist and Wilhelm, 2003). In the offer price adjustment context, an
unexpected positive shock to the valuation of the firm during the registration period
results in a bigger aggregate surplus, which is then divided between the underwriter and
the issuer according to their respective bargaining powers. This leads to a partial
adjustment of IPO offer prices.
Our results are consistent with the bargaining hypothesis of IPO offer price
adjustment. We find that the efficiency of offer price adjustment is better for IPO firms
with attractive outside options and for those with bigger incentives to negotiate. On the
other hand, efficiency is lower for issues underwritten by investment banks whose
services are highly in demand at the time and by those that employ influential research
analysts, indicating a positive relation between the bargaining powers of the underwriters
and the efficiency of IPO pricing. Incorporation of market returns becomes less efficient
towards the end of the registration period. We interpret this as evidence that the issuing
firm's bargaining power weakens over time, reflecting the declining threat of withdrawal.
In line with this argument, we find that the excess surplus created at the preliminary offer
price, presumably the result of a price competition among potential underwriters during
the bake-off process, erodes during the registration period. We also find evidence
consistent with the existence of a negotiation process: Issues with longer registration
periods and those with offer prices set at fractional dollars are associated with better offer
Our findings also have implications for the validity of the Edelen and Kadlec
(2005) tradeoff hypothesis. According to the tradeoff hypothesis, the partial adjustment
phenomenon is primarily due to the issuing firm's uncertainty over the offer's successful
completion. Issuers maximize the expected surplus from going public by weighing the
probability of success against offer proceeds conditional on success. If the surplus
increases during the registration period, issuers maximize the expected surplus by seeking
a higher success probability, thereby demanding only a partial adjustment of offer prices.
The bargaining and the tradeoff hypotheses differ in regard to the role of the underwriters
in the offer price adjustment process. According to the bargaining hypothesis, partial
adjustment occurs due to the underwriters' conflict of interests with the issuers and their
ability to profit from underpricing. On the other hand, according to the tradeoff model
underwriters do not play a role in the partial adjustment of offer prices. Our findings
regarding the relation between underwriter characteristics, such as the level of demand
for their services and the employment of influential research analysts, and the efficiency
of the offer price adjustment contradict the predictions of the tradeoff hypothesis.
In addition, according to the tradeoff hypothesis, the underwriters and issuing firms
have no incentive to engage in a nickel-and-diming strategy. Therefore, whether the final
offer price is at an integer or a fraction of a dollar should not contain any information,
and should not be significantly related to the efficiency of the offer price adjustment. This
contradicts our finding that issues with offer prices set at integers experience less
complete price adjustments. Taken together, the evidence in this study does not support
the Edelen and Kadlec (2005) tradeoff model.
In sum, we document substantial inefficiencies in the IPO offer price adjustment
process. The partial adjustment of IPO offer prices imposes substantial costs on the
issuing firms and their pre-issue shareholders. The evidence suggests that the partial
adjustment phenomenon is primarily due to agency problems between the issuers and the
underwriters, and the variation of efficiency across IPOs is best explained by the
bargaining explanation of IPO offer price adjustment. Our findings indicate that the
issuing firms can increase the efficiency of the IPO pricing process by improving their
bargaining powers vis-a-vis the underwriters by taking proper actions prior to the IPO,
for instance by securing alternative financing options.
The rest of this dissertation is organized as follows. Chapter 2 introduces the partial
adjustment phenomenon, and discusses the hypotheses and empirical predictions. Chapter
3 describes the methodological approach. Chapter 4 describes the data and the variable
construction. Chapter 5 discusses the empirical results. Chapter 6 presents various
robustness checks and Chapter 7 provides a summary and conclusion.
PARTIAL ADJUSTMENT PHENOMENON, HYPOTHESES, AND EMPIRICAL
The pricing process for an initial public offering (IPO) typically starts with the
meetings between prospective underwriters and the issuing firm when underwriters
present tentative valuations. After the formation of an underwriter syndicate, the first
pricing information is provided to the public in the form of an offer price range either in
the preliminary prospectus or an amended statement filed with the Securities and
Exchange Commission (SEC). The anticipated offer price range may be adjusted upward
or downward throughout the registration period depending on the level of investor
demand for the offer. The final offer price is set by the issuing firm and the underwriters
the day before the offering, and shares are allocated to investors at this offer price. Often
the first-day closing market price is higher than the offer price, and the percentage
premium over the offer price is termed underpricing.
Hanley (1993) was the first to document a strong positive relation between the offer
price revision during the registration period and the subsequent underpricing. It appears
that underwriters do not set the final offer prices up to the level that would be necessary
to keep the underpricing constant in the face of strong demand from investors. In other
words, the increase in the firm value during the registration period is only partially
incorporated into the offer price.
Next, we introduce the three hypotheses of partial adjustment and their empirical
predictions in the context of our analysis.
The Dynamic Information Acquisition Hypothesis
The first explanation of the partial adjustment phenomenon tested in this study is
the dynamic information acquisition hypothesis of Benveniste and Spindt (1989).
According to this hypothesis, investors possess valuable private information with regards
to the fair market value of the IPO firm, and the underwriters and the issuing firm try to
induce these investors to truthfully reveal their information in order to maximize the offer
proceeds. Underpricing is used as a reward to satisfy the incentive compatibility
constraint of the investors, creating a partial adjustment of offer prices.
Empirical evidence on the validity of the dynamic information acquisition
hypothesis is mixed. A central prediction of the dynamic information acquisition
hypothesis is that underpricing and offer proceeds for the issuers increase simultaneously,
resulting in higher average underpricing among IPOs with upward offer price adjustment.
Consistent with this prediction, Hanley (1993) documents a positive relationship between
offer price adjustment and underpricing. Hanley also documents that lagged market
returns predict underpricing, but she makes no attempt to explain this pattern. Loughran
and Ritter (2002) revisit this finding and point out that it is inconsistent with the dynamic
information acquisition hypothesis since investors need to be rewarded only for their
private information, and the incorporation of public information should be costless to the
issuer. Lowry and Schwert (2004) also find a statistically significant positive relationship
between the market returns during the registration period and the subsequent
underpricing of the offer, but argue that this relationship is economically small and
conclude that the IPO pricing process is close to efficient. However, their analysis does
not take into consideration the simultaneous effect of public information on the level of
offer price adjustment in addition to the underpricing, which may lead to a potential
underestimation of the economic significance of the inefficiencies.
Benveniste et al. (2002) extend the dynamic information acquisition framework to
include potential indirect feedback from comparable IPOs. The authors argue that
underwriters smooth the costs of information acquisition across IPOs related by a
common factor to resolve potential coordination problems. According to the extended
dynamic information acquisition framework, information produced in comparable IPOs
should make the costs of price adjustment of an issue underwritten by the same
investment bank cheaper, but not altogether free. On the other hand, there should be no
cost to adjusting offer prices when information spillover originates from comparable
IPOs taken public by other underwriters, since this is public information. Thus, the
information revealed in the pricing of IPOs should be fully incorporated into the offer
prices of subsequent IPOs that are in the same industry and taken public by other
Ljungqvist and Wilhelm (2002) and Benveniste et al. (2003) examine the
relationship between information spillovers and the level of offer price adjustment. Both
studies find that information revealed in comparable IPOs affect subsequent IPO offer
prices, consistent with the dynamic information acquisition hypothesis. However, they do
not investigate the efficiency of this price adjustment. In this study, we analyze the
efficiency of offer price adjustment associated with information spillovers. In addition,
1Edelen and Kadlec (2005) solve this problem econometrically by isolating the market returns from the
offer price adjustment using orthogonalization and report a larger relationship between underpricing and
market returns during the registration period. They do not discuss the implications of this finding on the
significance of the costs associated with the partial adjustment of public information.
our analysis distinguishes between information spillovers among IPOs underwritten by
the same investment bank and by others, and thereby provides a more comprehensive test
of the dynamic information acquisition hypothesis.
The Bargaining Hypothesis
The second explanation of the partial adjustment phenomenon tested in this study is
the bargaining hypothesis (Loughran and Ritter, 2002; Ljungqvist and Wilhelm, 2003).
This hypothesis is based on the notion that the incentives of the underwriters and the
issuing firm differ, and the offer price is set as a result of negotiations between the two
parties. In the offer price adjustment context, positive information revealed during the
registration period creates a surplus that is shared between the underwriter and the issuing
firm according to their bargaining powers and incentives to bargain.
Despite the empirical and anecdotal evidence for the existence of significant
conflicts of interest between underwriters and issuing firms, empirical investigations of
the bargaining hypothesis have been limited in the IPO literature, with a few notable
exceptions. Loughran and Ritter (2002) combine the bargaining hypothesis and prospect
theory to explain the variation in IPO underpricing, and suggest that issuers bargain hard
over the offer price in bad states of the world, whereas they are pushovers in bargaining
in good states of the world. Ljungqvist and Wilhelm (2005) test the predictions of
Loughran and Ritter (2002) by examining the decisions of issuing firms subsequent to the
IPO and find evidence that managers of issuing firms that are satisfied with the IPO's
outcome according to prospect theory are less likely to switch underwriters for their first
seasoned equity offering. Ljungqvist and Wilhelm (2003) examine the relationship
between the bargaining incentives of the issuing firms' decision makers and the level of
underpricing, and find evidence consistent with bargaining in the U.S. in the second half
of the 1990s. Other studies of bargaining in the IPO context include Logue (1973), who
examines the relationship between several bargaining proxies and underpricing; and
Sternberg (1989) and Hoberg (2004) who use formal bargaining models to explain the
partial adjustment phenomenon. In this study, we develop and test the predictions of the
bargaining theory with respect to the efficiency of offer price adjustment, and provide a
comprehensive analysis of the bargaining hypothesis in the context of IPO pricing.
In our analysis of the bargaining hypothesis, we first focus on the effect of outside
options on the bargaining powers of the issuers and the underwriters. According to the
bargaining theory, an attractive outside option leads to a higher bargaining power and a
bigger share of the surplus. In the IPO pricing context, the existence of alternative
financing options for the issuing firm should lead to a more complete offer price
adjustment, due to the issuer's credible threat of withdrawal if necessary. For example, an
IPO firm that is not satisfied with the terms offered by its underwriter would be more
capable to prolong the negotiations or even withdraw and come back to the market after
hiring another investment bank if it already has an established relationship with a bank or
a private equity firm.2 Following Dunbar and Foerster (2005), we use leverage and
private equity investment in the firm prior to the IPO as proxies for alternative financing
options of the IPO firms.
Next, we examine the underwriter's outside options. We argue that the
underwriter's bargaining power is positively related to the overall demand for its services
2 Consistent with this argument, Busaba et al. (2001) and Dunbar and Foerster (2005) find that IPO firms
with greater access to alternative financing options are more likely to withdraw. In addition, Busaba et al.
(2001) show that a higher probability of withdrawal by the issuing firm leads to lower underpricing '
although the authors suggest that a higher probability of withdrawal strengthens the issuer's bargaining
power with respect to the investors, rather than the underwriters.
by other issuers. The rationale is that the opportunity cost of an issue's withdrawal to the
underwriter is negatively related to the underwriter's expected revenues from other
issues. In other words, a given offer is more valuable to the underwriter during a cold IPO
market than during a hot one. On the other hand, the opportunity cost of withdrawal is
positively related to the foregone revenues from the offer in case of a deal failure.
Therefore, the bargaining hypothesis suggests that the ratio of expected fee income from
the issue to the total expected fee income from issues completed or filed during the
registration period should be negatively related to the underwriter's bargaining power,
and consequently should be positively related to the efficiency of the offer price
Prior studies have shown that research analyst reputation plays an important role in
securing underwriting business. Dunbar (2000) and Clarke et al. (2003) report that the
market share, and the change in the market share, of underwriters are positively related to
the presence of an analyst who is a member of Institutional Investor's All-American
Research Team. Furthermore, there is evidence that the presence of an all-star analyst
affects the pricing of an initial offering. Cliff and Denis (2004) report that issues
underwritten by investment banks with all-star analysts are more underpriced, and they
interpret the higher underpricing as compensation for post-IPO coverage by highly
ranked analysts. Loughran and Ritter (2004) attribute the higher underpricing observed in
the 1990s to the increased importance of analyst coverage. To the extent that the
existence of an all-star analyst leads to a higher demand for the services of the
underwriter and to attractive outside options of revenue generation, investment banks
with all-star analysts should be associated with less complete price adjustments in their
Ljungqvist and Wilhelm (2003) argue that issuers care more about underpricing
and bargain harder when their stakes in the offer and thus their opportunity cost of
underpricing are higher. Bradley and Jordan (2002) and Loughran and Ritter (2004) use
share overhang, the ratio of retained shares to the public float, to examine the relationship
between the opportunity cost of underpricing and first-day returns.3 We include share
overhang in the analysis as a proxy for the issuers' incentives to bargain. The bargaining
hypothesis predicts a negative relationship between share overhang and the efficiency of
the offer price adjustment.
The pricing of an IPO starts with the bake-off process, during which potential
underwriters compete for the issuer's business and present their preliminary valuations.
As a result, the initial file price range is determined in a highly competitive setting. On
the other hand, the final offer price is set at a time when the issuer's probability of
switching underwriters is very low (therefore, a threat to withdraw is not credible) due to
the high levels of sunk costs incurred during a long registration period. Therefore, the
bargaining hypothesis suggests that the bargaining power of the issuing firms decline
during the registration period. Barondes (2005) argues that between the selection of an
investment bank and the final IPO pricing, the negotiating advantage shifts in favor of the
investment banks. Hall and Robbins-Roth (1992) state that "Pricing is a factor that caught
3 An alternative interpretation of the relationship between underpricing and share overhang is provided by
Loughran and Ritter's (2002) prospect theory model of offer price adjustment. Loughran and Ritter argue
that issuers tend to sum the opportunity cost of underpricing with the wealth gain on retained shares as
prices increase substantially in the aftermarket. As a result of this behavior, issuers do not bargain as hard
as they should, especially when share overhang is high.
many managers by surprise. By the time the pricing meeting occurs, companies typically
are not negotiating from a position of strength." Therefore, the bargaining hypothesis
predicts less efficient price adjustment towards the end of the registration period than in
the beginning. We test this prediction by investigating whether market returns are
incorporated into the offer prices more completely in the beginning of the registration
period compared to towards the end. We also examine whether the surplus created at the
initial filing due to the high level of competition erodes away during the registration
period by means of a less efficient offer price adjustment.
In the IPO pricing process, a potential offer price is proposed by underwriters after
observing the demand for the offer by the investors, and the issuer has the option to
accept the proposed price or negotiate for better terms. Stronger negotiations by the
issuing firm results in a longer registration period due to the SEC regulations that require
an amendment to be filed and the need to reconfirm investor demand at the new price
(Barcaskey, 2005). Insistence of the issuing firm during the negotiations to increase the
offer price should result in higher offer proceeds and a more efficient offer price
adjustment. Therefore, the bargaining hypothesis predicts a positive relationship between
the length of the registration period and the efficiency of the price adjustment.4
The bargaining hypothesis suggests that an increase in firm value during the
registration period creates a surplus that is shared among the issuer, investors, and
underwriters. A distinctive feature of the IPO pricing process is that the share of surplus
that accrues to investors and underwriters is much less certain than the share that accrues
4 Busaba et al. (2001) interpret the registration period length as an indicator of weaker investor interest. In
unreported work we find that, conditional on upward offer price adjustment, longer registration periods are
associated with higher offer price adjustments. This suggests that a longer pre-market does not indicate a
weak demand for offers with upward offer price adjustment.
to the issuing firm. The reason for this asymmetry is that expected profits to investors
depend on the price at which they sell their shares in the aftermarket, which is uncertain
at the time the offer price is set, whereas offer proceeds for the issuing firm are known
Harris's (1991) costly negotiation hypothesis suggests that the propensity to use a
rounded set of prices rather than fractions is positively related to uncertainty about gains.
Therefore, the costly negotiation hypothesis suggests that offer prices are more likely to
be set at fractions as a result of stronger negotiations by issuing firms that face lower
uncertainty relative to underwriters and investors, and predicts a lower efficiency of price
adjustment for issues that are priced at integers. Consistent with this prediction, Bradley
et al. (2004) find that average underpricing for IPOs with integer offer prices is
significantly higher than those priced at a fraction. However, they do not examine the
efficiency of the offer price adjustment process. Mola and Loughran (2004) find higher
discounts at integer offer prices for seasoned offerings, and interpret this as evidence of
investment bank pricing power.
The Tradeoff Hypothesis
The third explanation of the partial adjustment phenomenon tested in this study is
the tradeoff hypothesis of Edelen and Kadlec (2005). In their model, issuers maximize
the expected surplus from going public by weighing the probability of success against
offer proceeds conditional on success. If the surplus increases during the registration
period, issuers maximize expected surplus by seeking a higher success probability,
thereby demanding only a partial adjustment of offer prices.
Most empirical predictions of the tradeoff hypothesis are similar to those of the
bargaining explanation of offer price adjustment. For example, both hypotheses contend
that public information is only partially incorporated into offer prices. In addition, the
offer price that the issuing firm accepts is a function of the firm's opportunity cost of
withdrawal. However, there are several predictions that distinguish the bargaining from
the tradeoff hypothesis.
A central difference between the tradeoff and the bargaining hypotheses is related
to the role of the underwriters in the pricing process. According to the bargaining
hypothesis, partial adjustment occurs due to the underwriters' conflict of interests with
the issuers, and their ability to profit from underpricing. On the other hand, underwriters
do not play a role in partial adjustment in the tradeoff hypothesis since issuers demand
partial adjustment of offer prices in order to maximize their expected surplus. Therefore,
contrary to the predictions of the bargaining hypothesis, the level of demand for the
underwriters' should not be related to the efficiency of the offer price adjustment. In
addition, according to the Edelen and Kadlec (2005) model, underwriters and issuing
firms have no incentives to engage in a nickel-and-diming strategy. Therefore, whether
the final offer price is set at an integer or a fraction of a dollar should not contain any
information, and should not be significantly related to the efficiency of offer price
All three hypotheses may partly explain the pattern observed in IPO offer price
adjustments, and one important goal of this study is to ascertain the relative importance
of the three explanations.
Estimation of the Adjustment Ratio
In this study we examine the level and determinants of efficiency in the IPO offer
price adjustment process. We define efficiency as the magnitude of the offer price
adjustment relative to the hypothetical complete level of offer price adjustment. We posit
that the complete level of offer price adjustment is equal to the change in the issuing
firm's value during the registration period. In other words, if the underwriters revise the
offer price as much as the change in the firm value, then the offer price adjustment is
complete and the eventual underpricing is equal to the underpricing that would have been
observed if the issuing firm's value had not changed.
We construct the Adjustment Ratio (AR), which is defined as the estimated
percentage change in an IPO firm's share value divided by the percentage offer price
revision during the registration period:1
AR. -o (3-1)
1 There is no theoretical reason for using the ratio of the two as the efficiency measure. For example, using
the difference between the percentage change in the firm value and the percentage offer price revision is
theoretically valid. However, empirical evidence indicates that the frictions associated with the offer price
adjustment process are multiplicative rather than additive. Additive frictions, such as a constant difference
regardless of the level of offer price revision, would lead to a negative relation between the offer price
revision and the underpricing which is not the case as documented in numerous studies. In addition, a valid
efficiency measure should be free of scale effects. In other words, the efficiency should not be related to the
magnitude of the offer price revision. In unreported results we find that the ratio measure is not related to
the offer price revision ceteris paribus, whereas the difference measure is positively related to it. Therefore,
we conclude that the ratio is a valid efficiency measure.
where OPi is the offer price, OPio is the original expected offer price which is taken as the
midpoint of the original file price range, MVi is the first-day closing share price of the
IPO firm, and MVio is the estimated market price of shares at the time of the original
filing. AR is an inefficiency measure as higher values indicate a less efficient offer price
adjustment process. A perhaps more intuitive measure of efficiency would be the
reciprocal of the Adjustment Ratio. However, the change in the firm value measure has
undesirable characteristics as the denominator since it takes values close to zero.
The expected market price of shares at the time of the original filing, MVio, is
required to calculate AR, but is not readily observable. Our methodology allows us to
estimate MVio using the information provided by the pricing of IPOs without subsequent
offer price revisions. We posit that MVio is the expected fair value of the shares that the
underwriters and the issuers anticipate at the time of the initial filing, and the percentage
difference between MVio and OPio is the underpricing that would have been observed if a
substantial change in the firm value that requires an offer price revision had not occurred.
Following this logic, we estimate the unobserved initial underpricing of IPOs with
subsequent offer price revisions using the information provided in the pricing of IPOs
without offer price revisions.
We first determine the relationship between the underpricing and firm-,
underwriter-, and offer-specific characteristics for IPOs whose final offer price is set in
the original file price range using the following first-stage regression model:
1 = UP1 = f (l,, I,-, Underwriter-, and Offer characteristics) (3-2)
We use the coefficients from this OLS regression to estimate what the underpricing
would have been for IPOs with eventual offer price revisions if the values of the issuing
firms had not changed during the registration period and the offer prices were not
adjusted. This analysis is essentially equivalent to decomposing the total underpricing of
IPOs with offer price revisions into two: (i) initial expected underpricing common to all
IPOs, independent of the offer price revision, and (ii) additional underpricing due to the
partial offer price adjustment. Then, we use the estimated initial underpricing of IPOs
with offer price revisions to calculate the expected market price of shares at the time of
the original filing using
MVO = (UP0 +1)OP0 (3-3)
where OPiO is the midpoint of the original file price range, and is the estimated initial
underpricing using the coefficients from the model in Equation (2). Then we plug-in
MVio in Equation (1) to calculate AR for all IPOs with offer price revisions.
The Adjustment Ratio measures the inefficiency of IPO offer price adjustments. If
offer prices are adjusted fully such that the percentage offer price revision is equal to the
percentage change in the issuing firm's value during the registration period, then AR
equals one. Otherwise, the higher the AR, the lower the adjustment of the offer price
relative to the change in the market value, and hence the lower the efficiency.
After calculating ARi for all deals with offer price revisions using the methodology
explained above, we investigate the determinants of variation in ARi across deals using
the OLS regression
AR. f (Firm-, Underwriter-,
Offer-specific characteristics, Public Information Proxies)
There are two primary concerns with regard to the validity of our efficiency
measure that need to be addressed. First, are the variables that are used to calculate the
efficiency ratio measured correctly? In particular, whether the midpoint of the initial file
price range and the estimated initial share value can be regarded as unbiased estimates of
their true values may be material to draw correct inferences from the analysis.
Lowry and Schwert (2004) find that the offer price revisions are predictably related
to firm and offer characteristics and market returns leading up to the initial filing. They
conclude that the file price ranges are low-balled by the underwriters and therefore are
not unbiased estimates of the final offer price. This upward bias in the magnitude of the
offer price revision can make the IPO pricing process seem more efficient than it really is
if this bias is not accounted for in the change in the firm value measure.2 However, our
methodology captures the inefficiency associated with the low-balling of the initial file
price range. Since we estimate the expected firm value at the time of the original filing
using the midpoint of the file price range, a low-balled price range leads to a larger value
for the change in the firm value measure as well and therefore results in a larger AR and
lower efficiency. The validity of our efficiency measure also depends on the accuracy of
the first-stage regression estimates of initial underpricing. We control for the stability of
the first-stage regressions using alternative specifications in Section 6.
The second concern is whether a complete offer price adjustment (100% efficiency)
is in fact attainable in an optimal IPO pricing process. A less than complete offer price
2 It should be noted that even though the magnitude of the offer price revision may be biased upward, the
final offer price should eventually be lower than what it otherwise would be for the underwriters to benefit
from low-balling. This implies that the issuers give importance to the magnitude of the offer price revision
independent from the level of the final offer price and proceeds. This is consistent with the prospect theory
explanation of Loughran and Ritter (2002).
adjustment may be due to unavoidable frictions in addition to avoidable inefficiencies.
For example, Booth and Smith (1986) argue that overpricing an IPO is undesirable to
both the issuing firm and the underwriters. To the extent that the fair market value of the
issuing firm is not known with certainty at the time of the offer, it may be in the best
interest of risk-averse underwriters and issuers to underadjust the offer price, especially
for issues with high valuation uncertainty and during volatile markets. We control for this
possibility by including firm-specific and general market uncertainty measures and
various control variables in the cross-sectional regressions. Another issue is the effect of
investor sentiment on the market value of the firm. If the underwriters price IPOs with
long-run value in mind, then the offerings with high investor overreaction in the
aftermarket would seem to have less efficient offer price adjustments. This "leaning
against the wind" hypothesis implies a negative relation between the post-IPO returns of
the issuing firms and the efficiency of the offer price adjustment. We test the "leaning
against the wind" hypothesis and examine whether investor sentiment plays a role in the
efficiency of the offer price adjustment in Section 6.
Sample Selection Bias
In this study, we concentrate our efforts on analyzing the efficiency of upward offer
price revisions primarily for two reasons. First, an overwhelming proportion of the
aggregate money left on the table is associated with IPOs with upward price revisions.
Therefore, an investigation of the factors that drive the efficiency of downward price
revisions is not economically as interesting. Second, there is very little systematic
variation in the pricing efficiency of IPOs when the offer prices are revised downward.3 It
3 Table II in Section 5.1 shows that the R square of the regression of underpricing on various firm-, offer-,
and underwriter-specific characteristics is 4.2% for IPOs with downward price revisions, whereas it is
appears that the price setting of IPOs with downward offer price revisions is
fundamentally different from the rest. In fact, there is evidence of widespread price
stabilization by the underwriters when the offer prices are revised downward. The
proportion of IPOs with exactly zero underpricing is 28.4% when the offer prices are
revised downward, whereas it is only 1.45% when the offer prices are revised upward.
This price stabilization renders the pricing of those IPOs uninformative for the purposes
of this study.
However, regression analysis using a subsample of IPOs is complicated by a
potential sample selection bias. If the selection criteria for the subsample is not random,
ordinary least squares (OLS) regressions produce biased and inconsistent coefficient
estimates. IPOs with upward price revisions do not constitute a random subsample of all
IPOs since the direction of the offer price revision is systematically related to various
factors and can be reliably predicted using information available prior to the offer date.
Moreover, the same factors are likely to play a role in the efficiency of the offer price
adjustment. For example, public information revealed during the registration period
affects both the efficiency of the offer price adjustment process and the direction of the
offer price revision. An upward offer price revision despite unfavorable market
conditions and information spillovers is possibly due to a favorable private information
revelation by the investors. Consequently, low market returns and information spillovers
may be associated with high levels of private information production in the subsample,
even if public and private information productions are completely orthogonal in the full
sample of IPOs. This may contaminate the coefficient estimates of the public information
41.1% for IPOs with upward price revisions. In addition, only industry affiliation and a time dummy are
proxies since the coefficients may capture additional information unrelated to publicly
We control for this potential sample selection bias using the Heckman (1979)
maximum likelihood estimation (MLE) procedure.4 Heckman MLE procedure is the
joint estimation of the following two econometric models:
Upward upward X upward upward (3-5)
AR = o; XAR + 3IAR IMR + R (3-6)
where Jpward is an indicator function for an upward offer price revision and is equal to
one for IPOs with final offer prices higher than the maximum of the initial file price
range and zero otherwise. Xupward includes the public information proxies and various
control variables, and the model is estimated using the profit analysis. The correction for
the sample selection bias is enabled by the inclusion of the inverse Mills ratio in the AR
model. The inverse Mills ratio (IMR) is:
IMR T (0( upward Xupward (3
IM/R = ol ) (3-7)
where au"'ardX"upard are the predicted probabilities of inclusion in the upward offer price
revision subsample from the probit regression of I"upard on Xupward, p is the standard
normal density function, and D is the standard normal distribution function. IMR is a
monotonically decreasing function of the predicted probability of inclusion in the
4 Heckman MLE is more efficient than the Heckman two-step procedure since the selection and the
prediction models are estimated jointly in the MLE.
A statistically significant coefficient for IMR indicates that the subsample is indeed
not randomly chosen and the OLS coefficient estimates are biased. The sign of the
coefficient estimate for IMR is also informative. The coefficient on IMR is given by:
3IMIR p( ward, AR) O( AR) (3-8)
where p is the correlation between the two error terms and a is the standard deviation of
the error term from the AR regression. Therefore, the sign of the coefficient on IMR is
the same as the sign of the correlation between the two error terms.
The error term in the probit regression captures the effect of information that is
relevant to the valuation of the issuing firm and independent from the publicly available
information when Xupward includes the public information proxies. According to the
dynamic information acquisition hypothesis, this private information is only partially
incorporated into the offer prices. Therefore, p("upwa, AR) should be positive since high
levels of private information should lead to a less complete offer price adjustment and a
higher AR. On the other hand, a higher offer price revision than predicted by the public
information measures may be due to the issuing firm's ability to demand and realize a
better pricing in the IPO. In that case, issuing firms with high (low) bargaining powers
will be associated with higher (lower) upward and lower (higher) AR, and p(upward, AR)
will be negative. Therefore, a positive coefficient on IMR is consistent with the dynamic
information acquisition hypothesis, whereas a negative coefficient is consistent with the
The sample of IPOs comes from data provided by Jay Ritter. This database
contains 8,042 IPOs between 1985 and 2003. Excluding ADRs (American Depository
Receipts), closed-end funds, REITs (real estate investment trusts), banks and savings &
loans, partnerships, unit offers, reverse leveraged buyouts, IPOs with an offer price below
$5.00 per share, and IPOs with missing first-day close share price results in a sample of
This IPO database is supplemented with information from the new issues database
of Securities Data Corporation (SDC). For each IPO, we collect information on the filing
date, institutional investment prior to the issue, and the underwriting syndicate. We
assign a ranking (RANK) to each lead manager using Loughran and Ritter's (2004)
classification, which is based on Carter and Manaster (1990). Founding dates come from
the Field-Ritter database (see Field and Karpoff (2002) and Loughran and Ritter (2004)).
Total leverage prior to the offering is taken from Compustat (Leverage data are available
for 86% of the final sample). Industry and individual firm returns are from the Center for
Research in Securities Prices (CRSP).
Variable names and their descriptions are listed in Table I.
[Insert Table I about here]
The two dependent variables in the analysis are underpricing, measured using the
offer price and the first-day closing price, and the adjustment ratio. The first set of
explanatory variables control for the riskiness of the issuing firm and the uncertainty in
the market during the registration period. The second set includes proxies for public
information revealed during the registration period, and the third set includes proxies for
the bargaining powers of the underwriters and the issuers, and the proxies for their
incentives to bargain.
The first risk proxy we use is the volatility of the issuing firm's stock price after the
offering as a proxy for ex ante risk (Firm-specific uncertainty) (Ritter (1984, 1987).
Asquith et al. (1998) report that price stabilization by underwriters continues until four
weeks after the offering. In order to leave out the effects of price stabilization on stock
return volatility, we use the standard deviation of returns over the first six months after
the offering, excluding the first month after the offer date. In addition to firm-specific
uncertainty, uncertainty in the market prior to the offering may affect the adjustment of
offer prices. To capture the general uncertainty in the market we use the volatility of
market returns during the registration period using an equally weighted index of firms in
the same Fama-French industry as the issuing firm (Market uncertainty).1 We also add
technology (Tech Dummy) and internet (Net Dummy) dummies to capture the
component of firm risk associated with those industries. To control for firm size, we use
the twelve month trailing sales of the issuing firm prior to going public. Firms with zero
1 Fama-French industry specifications are provided by Ken French at
http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/Data Library. We use the specifications updated in
2005 that form 49 industries using 4-digit SIC codes.
sales are assigned a value of $10,000. To adjust for inflation, we convert sales to 2003
dollars using the Consumer Price Index, and transform it using logarithms (Ln (Sales)).
To test the implications of the dynamic information acquisition hypothesis, we
include in the regression the underpricing and the offer price adjustment of comparable
firms taken public during the registration period by an IPO's own lead underwriter (UP
(Same Underwriter), PrcAdj (Same Underwriter)), and by different underwriters (UP
(Other Underwriters), PrcAdj (Other Underwriters)), as separate explanatory variables.2
If an issuer has more than one lead underwriter we add up the offers of all lead
underwriters when calculating the information spillover measures. To examine the
relationship between market returns and the adjustment ratio, and the potential change in
that relation over the duration of the registration period, we include in the analysis two
market-return variables. First, we calculate the return on an equally weighted index of all
publicly-traded firms in the same Fama-French industry as the IPO firm, for the duration
of the registration period excluding the last 15 trading days, standardized to a 30-day
return as in Busaba et al. (2001) (Industry Ret (Beginning)). Second, we measure the
return on the same index during the last 15 trading days before the offering (Industry Ret
The third set of explanatory variables tests the bargaining hypothesis of offer price
adjustment. The bargaining hypothesis suggests that the value of outside options is
positively related to bargaining power. We use pre-issue leverage of IPO firms and the
existence of institutional investment at the time of the offering as proxies for the issuing
firm's alternative financing options. Information on leverage comes from COMPUSTAT.
2 Following Ljungqvist and Wilhelm (2002), we set the spillover measures to zero if there are no
Following Busaba et al. (2001), firm leverage is defined as the ratio of total debt (short-
and long-term as well as subordinated) to total assets in the most recent quarter prior to
the issue. We use a dummy variable which equals one if the firm's leverage ratio is above
the median leverage ratio of the sample (High Debt Dummy). We test the relationship
between institutional investment in the IPO firm and the efficiency of the offer price
adjustment by including a dummy variable that equals one if at least one venture capital
or private equity firm owns a stake in the firm at the time of the offering (Institution
Dummy). IPO firms with institutional investment are identified from SDC's New Issues
As a proxy for the underwriter's outside options, we use the ratio of the
underwriter's expected fee income from a particular offer to the total actual and expected
fee income from all other IPOs completed or filed by the underwriter during the offer's
registration period (UW Relative Income).3 If an issuer has more than one lead
underwriter we take the mean of UW Relative Income across the lead underwriters. Prior
studies have shown that the quality of analyst coverage provided by the underwriters
affects their market shares and the pricing of their IPOs (Dunbar (2000), Clarke et al.
(2003), Cliff and Denis (2004)). To examine the relationship between analyst coverage
and the efficiency of the price adjustment, we use a dummy variable (All-star Dummy)
that equals one if the lead underwriter (or at least one lead if there are co-leads) has an
all-star analyst in that year's survey of Institutional Investor (first-, second-, or third-
team), and if the issuing firms is covered by that all-star analyst in the year after the
offering. This information is from Jay Ritter's IPO database, and is available between
3Expected fee income from an IPO is calculated using the expected offer proceeds at the time of the initial
filing and the gross spread as reported by the SDC.
1993 and 2003. In our sample, the all-star analyst dummy is equal to one for
approximately 19% of the offers in that time period. To examine the link between
underwriter prestige and the cost of offer price adjustment, we use the underwriter
rankings of Loughran and Ritter (2002), which builds upon the Carter and Manaster
(1990) and Carter, Dark, and Singh (1998) rankings (Underwriter Rank).
We include share overhang, the ratio of retained shares to the public float, in the
analysis as a proxy for the issuers' incentives to bargain (Overhang).4 To examine the
relation between the initial surplus at the preliminary offer price and the subsequent
efficiency of the offer price adjustment, we use the ratio of the issuer's price-to-sales
ratio to the mean price-to-sales ratio of comparable firms in the market (Initial P/S
Multiple). We calculate the issuing firm's initial price-to-sales ratio using the midpoint of
the preliminary file price range and the total revenue in the twelve months prior to the
offering. We calculate the mean price-to-sales ratio of comparable firms using the share
price and the twelve month revenues of firms in the same Fama-French industry as the
issuing firm, as of the latest fiscal year end prior to the offering. To examine the
relationship between the strength of the negotiations and the efficiency of the price
adjustment we include the natural logarithm of the length of the registration period
(Registration Length). As an additional indicator of negotiation between the underwriter
and issuer during the adjustment of offer prices, we include a dummy variable that equals
one when the offer is priced at an integer (Integer OP).
4 Loughran and Ritter (2i 4 question whether the percentage of shares offered (the measure used by
Ljungqvist and Wilhelm (2003) and related to the reciprocal of overhang) is the correct measure of how
much "skin in the game" management has, rather than the dollar value of the shares owned. Because
valuations were much higher in 1999-2000 than previously, the patterns diverge when these alternative
definitions are used.
Table 4-1. Definition of Variables.
Ln (1 +Age)
UP (Same Underwriter)
UP (Other Underwriters)
PrcAdj (Same Underwriter)
PrcAdj (Other Underwriters)
Equals MV / OP, where MV is the firm's first-day close share price,
and OP is the offer price.
Adjustment Ratio. Equals the ratio of the change in a firm's value per
share during the registration period to the offer price adjustment.
The natural logarithm of one plus firm age, defined as the difference
between the IPO calendar year and the founding year. Founding dates
are from the Field-Ritter dataset available on Jay Ritter's website.
The natural logarithm of total firm assets prior to the offering, in 2003
The natural logarithm of firm's trailing twelve month sales prior to
going public, in 2003 purchasing power.
Equals 1 if the firm operates in a technology industry excluding
internet firms, as identified in Loughran and Ritter (2004).
Equals 1 if the firm operates in an interet-related industry, as
identified in Loughran and Ritter (2004).
Percentage daily standard deviation of the issuer's stock returns, for
the five month period starting one month after the offer date.
Percentage daily standard deviation of stock returns for firms in the
same Fama-French industry as the issuer, during the issuer's
Lead underwriter prestige ranking, using Loughran and Ritter's (2002)
classification, based on Carter and Manaster (1990). If there is more
than one lead underwriter, we use the highest ranking.
The average underpricing of same-industry IPOs taken public by the
same lead underwriter, completed during the registration period of the
issue. If there are no same-industry IPOs in that time period, it is set to
The average underpricing of same-industry IPOs taken public by a
different lead underwriter, completed during the registration period of
the issue. If there are no same-industry IPOs in that time period, it is
set to zero.
The average offer price adjustment of same-industry IPOs taken public
by the same lead underwriter, completed during the registration period
of the issue. If there are no same-industry IPOs in that time period, it is
set to zero.
The average offer price adjustment of same-industry IPOs taken public
by a different lead underwriter, completed during the registration
period of the issue. If there are no same-industry IPOs in that time
period, it is set to zero.
Table 4-1. Continued
Industry Ret (Beginning)
Industry Ret (End)
High Debt Dummy
Initial P/S Multiple
Registration Ii gilll
j\N it Dummy
Average market return of firms in the same Fama-French industry as
the issuer, during the issuer's registration period, excluding the last 15
trading days before the offering. We standardize Industry Ret
(Beginning) to a 30-day equivalent.
Average market return of firms in the same Fama-French industry as
the issuer, during the last 15 trading days of the issuer's registration
Equals 1 if the issuer's ratio of total debt to total assets prior to the
offer is larger than 15% (median of the sample).
Natural logarithm of the ratio of expected spread income from the
offer to the lead underwriter's expected spread income from all offers
completed or filed during the issue's registration period. If there is
more than one lead underwriter, we take the average across
underwriters before taking the natural logarithm.
Ratio of retained shares to public float.
The ratio of the issuer's price-to-sales ratio to the mean price-to-sales
ratio of all firms in the same Fama-French industry as the issuer. The
issuer's price-to-sales ratio is calculated using the midpoint of the
preliminary file price range and the twelve month trailing revenues
prior to the IPO. The mean price-to-sales ratio of comparable firms is
calculated using the share price and the twelve month trailing revenues
as of the latest fiscal year end prior to the offering.
The natural logarithm of the duration of the registration period.
Equals 1 if a venture capital or private equity firm owns shares of the
IPO firm at the time of the offering.
Equals 1 if the offer price is set at an integer dollar.
Equals 1 if the lead underwriter has an all-star analyst covering the
firm in the one year following the offering.
Equals 1 if the offer year is between 1990 and 1998.
Equals 1 if the offer year is 1999 or 2000.
Equals 1 if the offer year is between 2001 and 2003.
In this section we investigate the level and determinants of efficiency in IPO offer
price adjustments. We first calculate the adjustment ratio for IPOs as described in Section
3. Second, we test the predictions of the dynamic information acquisition hypothesis
(Benveniste and Spindt (1989), Benveniste et al. (2002)) by comparing the effect of
public information revealed during the registration period on the offer price adjustment
and the change in firm value. Third, we investigate the determinants of the variation in
the adjustment ratio across IPOs and through time.
Estimation of the Adjustment Ratio
In this sub-section, we estimate the initial underpricing of IPOs with subsequent
offer price revisions. In order to estimate the initial underpricing, we first investigate the
determinants of underpricing for IPOs with no subsequent offer price revisions. In this
analysis we use explanatory variables that are commonly used in other IPO underpricing
studies. In order to prevent a look-ahead bias we only use information that is available at
the time of the initial filing.
Table II presents the ordinary least squares regression results. The coefficients from
the underpricing regression of IPOs with a final offer price inside the original file price
range are used to estimate the adjustment ratio in the next step of the analysis. In
addition, Table II reports the regression results separately for all IPOs, IPOs with
negative price adjustments, and IPOs with upward price adjustments.
[Insert Table II about here]
The regression results from Table II suggest that the level and variation of
underpricing depend significantly on offer price adjustment. Notably, for IPOs with
negative price adjustment only the technology industry dummy and a dummy variable for
IPOs from 1990-1998 are significantly related to underpricing. In addition, the adjusted R
square of 4.2% reveals that the explanatory power of the model is low. On the other hand,
the determinants of IPO pricing for issues with upward price adjustment show
considerable differences compared to other issues. The last row in Table II shows that
underwriter rank, share overhang, total assets, and firm age have substantially more
impact on underpricing when the offer price is adjusted upward.1 Internet firm IPOs are
underpriced an additional 23% compared to other IPOs when offer prices are adjusted
upward, but this relationship does not exist in offers with no upward price adjustment.
The regression results also reveal that the extreme underpricing in the bubble period of
1999 and 2000 can be attributed for the most part to IPOs with upward price adjustment.
The bubble dummy for IPOs with upward price adjustment is 68% whereas it is 17% for
those with no price adjustment and an insignificant 2% for those with negative price
Next, we estimate the initial underpricing of IPOs with subsequent offer price
revisions using the following model from Table II:
1 Habib and Ljungqvist (2001) and Fernando et al. (21""4) argue that the underwriter rank variable is
endogenous in underpricing regression. Loughran and Ritter (2-""14 control for endogeneity using an
instrumental variable regression and find no difference in the underwriter rank coefficient in the
instrumental variable regression compared to an OLS regression. We conduct a similar analysis in Section
6 to control for potential endogeneity between the underwriter prestige and the efficiency of the offer price
UPO = 8.68 0.32 *Rank +1.66 Overhang -1.63 Ln(Assets)+ 0.53 *Ln(Sales)
0.73 Ln(1 + Age)+ 2.39 Tech Dummy +3.17 Net Dummy +1.01 Inst Dummy (5-1)
+ 6.80 Nineties Dummy +16.74 Bubble Dummy + 5.65 *Post Dummy
Using the estimated initial underpricing from Equation (9) we calculate the
adjustment ratio, the ratio of the percentage change in firm value during the registration
period to the percentage offer price adjustment, as described in Section 3. The adjustment
(MVi (UP +1)OP0)/
AR. = oi p/ (5-2)
1 (OP; -OP")
where MVi is the first-day closing share price of the IPO firm OPi is the offer
price, OPiO is the midpoint of the file price range used as a proxy for the original
expected offer price, and is the initial underpricing estimated using Equation (9).
Table III presents descriptive statistics for the adjustment ratio, categorized by the
direction of the offer price adjustment and by offer price adjustment quartile. Panel A
shows the descriptive statistics for IPOs with downward offer price revisions. The mean
AR is 1.26, suggesting that on average 79% ([1/1.26]* 100) of the decrease in the firm
value during the registration period is incorporated into the offer prices. This partial
downward adjustment is surprising in that it appears as if the underwriters are setting the
final offer prices too high after revising the preliminary expected offer prices downward.
There are several reasons why we may observe only partial adjustment of offer prices in
IPOs with downward offer price revisions. First, the costs to the underwriter of slightly
overpricing an IPO may be smaller than the costs of withdrawing the offer altogether if
the expected final offer price at full adjustment is below the reservation price of the
issuing firm. Second, the initial underpricing before the downward offer price revision
may be too high and therefore the final offer price may be at a fair value after the partial
adjustment. This is consistent with the findings of Lowry and Schwert (2004) who
suggest that the initial file price ranges are low-balled by the underwriters. Third, the
partial adjustment finding may be due to a truncation bias in that we only observe IPOs
that successfully go public. The adjustment may be full on average, but less than full
conditional on offer completion if IPOs with more complete adjustments are more likely
to withdraw. On the other hand, the downward adjustment may be even more partial than
suggested by the results in Table III. The underwriters appear to engage in price
stabilization especially for IPOs with downward offer price revisions. The proportion of
IPOs with exactly zero underpricing is 28.4% when the offer prices are revised
downward, whereas it is only 1.45% when the offer prices are revised upward. If the
underwriters support the share price of the issuing firm in the aftermarket so that it does
not fall below the final offer price, then AR is upward biased and the downward offer
price revision is even less complete.
[Insert Table III about here]
Panel B presents the descriptive statistics for IPOs with upward offer price
revisions. The mean AR is 2.36 suggesting that on average only 42% ([1/2.36]*100) of
the increase in the firm value during the registration period is incorporated into the offer
prices. This partial adjustment is highly costly for the issuers and their pre-issue
shareholders. As an example, a firm with $100 million in expected filing proceeds and a
40% increase in the expected market value during the registration period increases its
final offer proceeds by only $16.8 million ([40%] 0.42 $100 million = $16.8 m.), and
leaves an additional $23.2 million ($40 million $16.8 million) on the table due to the
partial adjustment of its offer price. Given that the median underwriter spread is 7%, the
indirect costs due to the partial offer price adjustment for this issuing firm is more than
three times the initial expected direct costs of going public.
We also observe a negative relation between the efficiency of the upward offer
price adjustment and the magnitude of the offer price revision. The mean AR for IPOs in
the first quartile (offer price revision below 15.2%) is 1.90, whereas it is 3.06 for IPOs in
the fourth quartile (offer price revision above 35%). However, in Section 6 we show that
this relation is spurious: IPOs in the upward offer price adjustment sample and with
higher valuation uncertainty tend to have higher offer price revisions, but such issues also
experience less complete offer price adjustments.
Figure 1 depicts the mean and median adjustment ratio for IPOs with upward offer
price adjustments by year. As illustrated in the figure, the efficiency of offer price
adjustment exhibits substantial variation over time, reaching a maximum of 3.53 in 1999.
The cost of upward offer price adjustment is clearly larger in the bubble period of 1999-
2000. The average AR during 1999 and 2000 is 3.46, which is equivalent to incorporating
only 28.9% of the change in the firm value into the offer prices. For the firm in the
example above, this means increasing the offer proceeds by only $11.6 million and
leaving an additional $28.4 million on the table.
[Insert Figure 1 about here]
The costs associated with the partial adjustment of IPO offer prices are very high
and point to substantial inefficiencies in the IPO pricing process. In the next section, we
examine the relationship between the adjustment ratio of IPOs with upward offer price
revisions and firm-, underwriter-, and offer-specific characteristics, as well as IPO market
conditions, and we investigate the determinants of the inefficiencies in the IPO offer price
The Impact of Public Information on the Value and Pricing of IPO Firms
We first examine how our public information proxies are related to the offer price
adjustment and the change in firm value during the registration period. This analysis
allows us to verify that our proxies indeed capture publicly available information relevant
to the pricing the IPO firms, and to examine whether the information revealed during the
registration period affect the offer prices and the value of the IPO firms differently. The
results are presented in Table IV.
The dependent variable in the first column is the percentage offer price adjustment,
measured as the ratio of the final offer price to the midpoint of the original filing price
range. The dependent variable in the second column is the estimated percentage change
in firm value during the registration period. The independent variables include measures
of information spillovers from recent comparable IPOs, and the market returns of
comparable public firms during the registration period. Information spillovers from
recent comparable IPOs underwritten by the same investment bank and from those
underwritten by other investment banks are examined separately following Benveniste et
al. (2002). The market returns in the beginning and towards the end of the registration
period are also examined separately in order to capture any change in the incorporation of
information during the registration period.
It is well documented that the underpricing of IPOs is strongly related to their offer
price revisions and both the underpricing and the offer price revisions are strongly related
to contemporaneous market returns. Therefore our information spillover measures are
highly correlated with each other and with the market return variables. In order to isolate
the influence of each public information measure, we orthogonalize the average
underpricing of recent comparable IPOs variables with respect to the average offer price
adjustment and market return variables, and the average offer price adjustment of recent
comparable IPOs variables with respect to the market return variables. White's
heteroskedasticity-consistent standard errors are provided in parentheses below the
[Insert Table IV around here]
The results from Table IV demonstrate that both the offer price adjustment and the
change in the value of the IPO firm during the registration period are strongly and
positively related to information spillovers from recent comparable IPOs and market
returns of comparable public firms. More specifically, higher average underpricing and
higher average offer price adjustment in the IPO market lead to higher final offer prices
and total valuation for comparable IPO firms that are going through the IPO pricing
process. In addition, higher market returns lead to higher offer prices and valuation for
IPO firms that recently filed for an IPO and also for those that are about to complete their
On the other hand, the findings suggest that public information revealed during the
registration period has a bigger impact on the value of the firm than on the magnitude of
the offer price adjustment. The coefficients of the regressors on column (2) are larger
than the respective coefficiencts on column (1), and the differences are statistically highly
significant. This finding suggests that information revealed in recent comparable IPOs,
both from those underwritten by the same investment bank and by others, and the market
returns during the registration period are only partially incorporated into IPO offer prices.
Therefore, offer price adjustment in response to public information revealed during the
registration period is inefficient and costly to the issuer. The evidence from this analysis
is contrary to the predictions of the dynamic information acquisition hypothesis, but
consistent with both the bargaining and tradeoff hypotheses.
Multivariate Analysis of the Efficiency of the Offer Price Adjustment Process
Having shown that the public information proxies used in this study indeed
influence the valuation of the issuing firm and the pricing of the offer, we next conduct a
comprehensive analysis of the efficiency of the IPO offer price adjustment process. In
particular, we investigate the factors that lead to a more efficient offer price adjustment
process and we determine which hypothesis of IPO offer price adjustment best explains
Table V presents the regression results. The dependent variable, the adjustment
ratio, is defined as the ratio of the change in firm value during the registration period to
the offer price adjustment. Therefore, the dependent variable is a measure of inefficiency
and a positive coefficient in the regression analysis means that the factor is negatively
related to the efficiency of offer price adjustment. Model (1) presents the base ordinary
least squares regression results and includes valuation uncertainty measures, proxies for
public information revealed during the registration period, bargaining hypothesis proxies,
and various control variables as the explanatory variables. Model (2) includes the all-star
analyst dummy, which is available between 1993 and 2003.
[Insert Table V around here]
The findings suggest that IPOs that are harder to value experience less complete
offer price adjustments: The coefficient on the volatility of issuing firm post-IPO returns,
Firm-specific uncertainty, is positive and significant. It appears that underwriters are risk-
averse and they tend to underadjust the offer prices when there is a relatively high
probability of valuation reversal. This finding is consistent with Booth and Smith (1986)
who argue that underwriters use their reputation to certify that an issue is not overpriced.
In addition, Nanda and Yun (1997) find that overpriced offerings are associated with a
decrease in the lead underwriter's market value and Dunbar (2000) finds that investment
banks lose market share if they are associated with overpriced IPOs. Even though the
aversion to overprice leads to lower offer proceeds, the issuers may also benefit from it.
The market may take overpricing as a negative signal with respect to the ability and the
incentives of the firm's managers, which may especially be damaging to issuers that plan
to come back to the market for a seasoned equity offering and to insiders that are
planning to sell their shares in the near future. On the other hand, the volatility of the
comparable public firm returns during the registration period, Market uncertainty, is not
significantly related to the efficiency of the offer price adjustment. Underwriters do not
seem to be concerned about the uncertainty in the market when they are setting the IPO
Next, we test the predictions of the dynamic information acquisition hypothesis by
investigating the relationship between the measures of public information revealed during
the registration period and the adjustment ratio. According to the extended dynamic
information hypothesis of Benveniste and Spindt (1989) and Benveniste et al. (2003), all
public information measures should be negatively related to the adjustment ratio (hence
positively related to the efficiency of offer price adjustment) and the coefficients on the
measures of information spillover from recent IPOs taken public by the same underwriter
should be less negative than those taken public by different underwriters. The relation
between information spillovers from recent IPOs and the offer price adjustment have
been documented in previous studies.2 However, this is the first study to examine how
efficiently this information is incorporated into offer prices.3 In addition, ours is the first
study to differentiate the source of the information spillover with regards to the identity
of the lead underwriters as suggested by Benveniste et al. (2003).
According to the regression results in model (1), incorporation of information
revealed in recent comparable IPOs taken public by other underwriters is inefficient: the
coefficients on UP (Other Underwriters) and PrcAdj (Other Underwriters) are positive
and statistically significant. In other words, IPO markets with high average underpricing
2 See Ljungqvist and Wilhelm (2002), Ljungqvist and Wilhelm (2003), Benveniste et al. (2003), and Edelen
and Kadlec (2005).
3 Edelen and Kadlec (2005) document a positive relationship between information spillover measures and
subsequent underpricing. However, they do not discuss the implications of this result.
and high average offer price revisions are followed by IPOs that experience less efficient
offer price adjustments. This finding is inconsistent with the dynamic information
acquisition hypothesis since the information regarding the pricing of recent IPOs is
publicly available and should be completely incorporated into the offer prices. On the
other hand, incorporation of information revealed in recent comparable IPOs taken public
by the same underwriter is more efficient: the coefficient PrcAdj (Same Underwriter) is
negative and significant. Taken together with the finding that the average offer price
revision of recent comparable IPOs of other underwriters do not lead to a more efficient
offer price adjustment, this result suggests that the efficiency of the offer price adjustment
process is not related to the type of information that is revealed, but rather to the identity
of the underwriters. More specifically, underwriters with a history of upward offer price
revisions tend to adjust offer prices in future IPOs more efficiently. Together, these
results are inconsistent with the extended dynamic information acquisition hypothesis.
The regression results also suggest that the efficiency of the offer price adjustment is
negatively related to the average market return of comparable public firms during the
registration period. This inefficiency is especially prevalent towards the end of the
registration period: Industry Ret (End) is positive and highly significant, while Industry
Ret (Beginning) is negative but insignificant. This inefficiency associated with the
incorporation of public information is similar to the findings of Loughran and Ritter
(2002), and consistent with both the bargaining and tradeoff hypotheses, but not the
dynamic information acquisition hypothesis.
Next, we examine the relation between the bargaining hypothesis proxies and the
efficiency of the offer price adjustment. Coefficient estimates for all eight bargaining
proxies have the predicted sign and are highly significant. The coefficient on High Debt
Dummy is negative and significant, indicating that issuers with higher leverage at the
time of the offering experience more complete offer price adjustments. This relation is
consistent with Helwege and Packer (2004) who argue that firms with high levels of
leverage are more concerned over the pricing of their equity. Additionally, Busaba et al.
(2001) show that firms with higher leverage are more likely to withdraw under
undesirable circumstances, and attribute this finding to a better access to alternative
sources of financing for such firms. Similarly, the coefficient on Institution Dummy is
negative and significant, suggesting that issuers that have an established relationship with
institutional investors (venture capitalists and private equity funds) prior to their IPOs
experience more efficient offer price adjustments. Taken together, we interpret these
results as evidence that issuing firms with access to alternative financing sources during
the IPO pricing process have higher bargaining powers in their negotiations with the
underwriters as a result of a more credible threat to withdraw, which leads to more
complete offer price adjustments. Overhang is positively related to the adjustment ratio,
consistent with a lower opportunity cost of money left on the table associated with
smaller offerings. In other words, issuers that have lower incentives to bargain for higher
offer prices experience less efficient offer price adjustments.
The bargaining hypothesis is based on the notion that potential conflicts of interest
between the issuers and underwriters play a significant role in the setting of the IPO offer
prices. More specifically, the efficiency of the offer price adjustment process is related to
not only the bargaining power of the issuers but also to that of the underwriters. In
particular, according to the bargaining hypothesis, underwriters whose services are highly
in demand by other issuing firms at the time should be associated with less efficient offer
price adjustments since the potential withdrawal of a single IPO is financially less costly.
The regression results support this argument. The coefficient on UW Relative Income is
negative, which implies that when the expected fee income from an individual IPO is
relatively small compared to the fee income from recently completed offers and from
offers expected to be completed in the near future, the efficiency of the offer price
adjustment for that particular IPO is lower.4 Model (2) includes the All-star analyst
dummy, which equals one if the lead underwriter employs an all-star analyst that is
specialized in the industry of the IPO firm and provides coverage for the IPO firm within
a year of the offer date. Dunbar (2000) and Clarke et al. (2003) find that the demand for
the services of an underwriter is positively related to the presence of an analyst who is a
member of Institutional Investor's All-American Research Team. As predicted by the
bargaining hypothesis, the coefficient on the All-star analyst dummy is positive and
highly significant. In other words, there is less adjustment of the offer price when the lead
underwriter employs an all-star analyst who subsequently covers the company going
public. Overall, the negative relation between the efficiency of the offer price adjustment
process and the demand for the underwriter's services is consistent with the bargaining
hypothesis. On the other hand, this result is inconsistent with the dynamic information
acquisition and the tradeoff hypotheses.
4 The same argument could be made for underwriters with high prestige. In fact, the prestige of the lead
underwriter, as measured by its Carter-Manaster rank, is negatively and statistically significantly related to
the offer price adjustment efficiency when UWRelative Income is excluded from the regressions. The fact
that the influence of the underwriter prestige disappears after controlling UWRelative Income suggests that
it is the short-term demand that determines the underwriter's bargaining power in its deals at the time rather
than its long-term prestige, as should be expected.
The initial pricing in the beginning of the registration period influences the
efficiency of subsequent offer price adjustment: the coefficient on Initial P/S Multiple is
positive and statistically significant. Therefore, issuers with relatively high initial offer
prices experience less efficient offer price adjustments. It appears that the initial surplus
created as a result of competition among potential underwriters during the bake-off
process erodes away during the registration period. Taken together with the finding that
the efficiency of the incorporation of the market returns worsens towards the end of the
registration period, this result supports the idea that the bargaining powers of the issuers
decline during the IPO pricing process.
The coefficient on Integer OP is positive and significant, suggesting that the issues
priced at fractional dollars experience more complete price adjustments. We interpret this
as evidence that price setting by the issuer instead of the underwriter, which indicates a
higher bargaining power, results in a more efficient offer price adjustment. This finding is
inconsistent with the tradeoff hypothesis since in the Edelen and Kadlec (2005)
framework issuing firms set the offer prices that maximize the unconditional expected
offer proceeds and consequently the efficiency of offer price adjustment should not be
related to whether the final offer price is set at the fractions of a dollar or at an integer.
The coefficient on the length of the registration period, Registration Length, is negative
and significant, suggesting that longer negotiations result in more complete adjustments.5
The ordinary least squares analysis in models (1) and (2) assumes that the sub-
sample of IPOs with upward price revisions is representative of all IPOs. However, this
5 The distribution of ,ir. di is skewed towards right due to a group of IPOs with uncharacteristically long
registration periods. In order to ensure that our results are not driven by outliers, we repeat the analysis
after tiunaiingl ,. ,a ih to 180 days. The results are qualitatively and quantitatively identical.
assumption is not valid if the selection criteria for the sub-sample of IPOs is not random,
and OLS regressions may produce biased coefficient estimates and may lead to incorrect
inferences. We correct for this potential sample-selection bias using the Heckman (1979)
MLE procedure. The MLE procedure consists of the joint estimation of two econometric
systems. The first system is a probit model of inclusion in the sub-sample using the full
sample of IPOs, where the dependent variable equals one if an IPO experiences upward
price revision and zero otherwise. The second system is the main specification that
parallels the OLS regressions in models (1) and (2), with the addition of the inverse Mills
ratio (IMR) which is derived using the predicted probabilities of inclusion in the sub-
sample estimated in the first-step probit regression.
Model (3) presents the first-step probit regression of inclusion in the sub-sample.
The independent variables include proxies for valuation uncertainty, information
spillover measures and market returns during the registration period, firm size as proxied
by Ln (Sales), industry affiliation, and two instrumental variables, Ln (Expected
Proceeds) and Pure Primary Dummy, to ensure good identification.
Not surprisingly, more positive information revealed during the registration period
leads to a higher probability of upward offer price revision: the coefficients on five of the
six public information measures are positive and statistically significant. The exception is
the UP (Same Underwriter) variable which has a positive but insignificant coefficient.
The probit regression results also suggest that IPOs are less likely to experience upward
offer price revision during volatile markets. On the other hand, firm-specific valuation
6 In unreported results, we find that Ln (Expected Proceeds) and Pure Primary Dummy are not significantly
related to the efficiency of the offer price adjustment. Therefore, they satisfy the required criteria of valid
uncertainty is not significantly related to the likelihood of upward offer price revision,
perhaps because offer prices of issuers that are harder to value are equally likely to be
revised downward or upward.
Models (4) and (5) report the Heckman MLE results from the joint estimation of
the two econometric systems. The coefficient estimate for the inverse Mills ratio (IMR) is
statistically significant and thus indicates the presence of a sample selection bias. The
biggest differences between the OLS and the Heckman MLE results are associated with
the public information proxies. The coefficient on PrcAdj (Other Underwriters) remains
large and positive but turns insignificant after correcting for the sample selection bias. In
addition, the coefficient on Industry Ret (Beginning) becomes weakly significant, which
leads to a more apparent reversal in the efficiency of the offer price adjustment over the
duration of the registration period. Overall, the inferences drawn from the OLS regression
results remain largely unchanged after correcting for the sample selection bias.
The sign of the coefficient on the inverse Mills ratio is especially interesting and
informative. The statistically significant negative coefficient estimate on AR indicates
that the correlation between the error terms of the probit regression and the inefficiency
regression is also negative. In other words, issues with upward offer price revisions that
are driven primarily by private information production are associated with more complete
offer price adjustments. This finding is inconsistent with the dynamic information
acquisition hypothesis which predicts that private information production leads to less
complete offer price revisions. On the other hand, it is consistent with the bargaining
hypothesis. It appears that issuing firms that achieve higher offer prices than suggested by
publicly available information are associated with more complete offer price adjustments,
which is consistent with the predicted relationship between the bargaining power of the
issuers and the efficiency of the offer price adjustment process.
Table 5-1. Regressions of First-Day Returns on IPOs Categorized by Price Adjustment*
Underwriter Ln Ln Ln Tech Net Institution Nineties Bubble Post R2ad
Intercept Rank Overhang (Assets) (Sales) (1+Age) Dummy Dummy Dummy Dummy Dummy Dummy
All -4.46 1.39 4.21 -2.94 0.61 -1.99 6.80 32.04 0.03 9.72 33.74 5.57 0.277
N= 4,825 (2.00) (0.29) (0.51) (0.53) (0.41) (0.53) (1.43) (4.85) (1.48) (0.75) (2.98) (1.69)
OP < LO 5.06 -0.30 0.04 0.02 -0.59 0.03 3.87 17.95 -2.61 3.39 1.66 3.39 0.042
N= 1,151 (1.99) (0.45) (0.34) (0.47) (0.51) (0.39) (1.97) (13.66) (2.30) (0.66) (4.12) (2.38)
LO < OP
N= 2,471 (1.85) (0.24) (0.43) (0.35) (0.31) (0.47) (0.93) (4.20) (0.93) (0.72) (2.48) (1.47)
HI < OP -9.51 4.94 7.33 -8.01 0.54 -3.87 4.72 23.43 -0.60 17.18 68.80 12.45 0.411
N= 1,203 (10.48) (1.28) (0.94) (1.56) (1.25) (1.75) (3.72) (7.30) (4.03) (3.01) (6.48) (5.50)
*The sample includes 4,825 US operating firm IPOs over 1985-2003. Unit offers, REITs, closed-end funds, banks and S&Ls, ADRs, reverse
LBOs, and IPOs with minimum of the price range below $8.00 are excluded. OP is the final offer price. LO is the minimum of the initial filing
price range. HI is the maximum of the initial filing price range. Underwriter Rank is the updated Carter-Manaster prestige measure of the lead
underwriter. Overhang is the ratio of retained shares to public float (the number of shares issued). Ln (Assets) is the natural logarithm of the pre-
issue book value of assets, expressed in millions of dollars of 2003 purchasing power using the CPI. Ln (Sales) is the natural logarithm of the total
pre-issue revenues during the prior twelve months, expressed in millions of dollars of 2003 purchasing power using the CPI. Ln (1+Age) is the
natural logarithm of one plus the number of years since the firm's founding date as of the IPO. Tech Dummy takes a value of one (zero otherwise)
if the firm is in the technology business, and Net Dummy is similarly defined for internet firms. Institution Dummy takes a value of one (zero
otherwise) if a venture capital or private equity firm owns shares of the IPO firm at the time of the offering. Nineties Dummy takes a value of one
(zero otherwise) if the IPO occurred during 1990-1998. Bubble Dummy takes a value of one (zero otherwise) if the IPO occurred during 1999-
2000. Post Dummy takes a value of one (zero otherwise) if the IPO occurred during 2001-2003. White's heteroskedasticity consistent standard
errors are in parentheses. ***, **, and denote significance at the 1, 5, and 10 percent level, respectively.
Table 5-2. Descriptive Statistics for the Adjustment Ratio
Panel A: IPOs with downward offer price adjustment
Offer price Mean (Median) offer Mean (Median) change Mean (Median)
adjustment quartile N price adjustment in the firm value PAR
1 (low) 245 -38.8% -43.5% 1.12
(-36.8%) (-42.9%) (1.13)
2 264 -26.2% -31.1% 1.19
(-25.9%) (-32.5%) (1.24)
3 230 -20.1% -26.2% 1.31
(-20.0%) (-26.1%) (1.30)
4 (high) 261 -13.8% -19.7% 1.43
(-14.3%) (-20.4%) (1.45)
All 1000 -24.6% -30.0% 1.26
(-23.1%) (-29.8%) (1.24)
Panel B: IPOs with upward price adjustment
Offer price Mean (Median) offer Mean (Median) change Mean (Median)
adjustment quartile N price adjustment in firm value PAR
1 (low) 254 12.6% 24.1% 1.90
(13.0%) (19.3%) (1.54)
2 293 18.2% 38.8% 2.14
(18.2%) (32.8%) (1.75)
3 285 27.6% 65.7% 2.34
(27.3%) (52.9%) (2.01)
4 (high) 287 59.8% 190.2% 3.06
(50.0%) (138.0%) (2.55)
All 1119 29.7% 80.9% 2.36
(22.2%) (42.6%) (1.94)
The sample includes 2119 US operating firm IPOs with either upward or downward offer price
revisions during 1985-2003. Unit offers, REITs, closed-end funds, banks and S&Ls, ADRs,
reverse LBOs, and IPOs with minimum of the price range below $8.00 are excluded. Offer price
revision is the offer price minus the midpoint of the initial filing range, divided by the midpoint of
the initial filing range. Change in the firm value is the first-day close share price minus the
original expected market price, divided by the original expected market price. Adjustment ratio
(AR) is defined as the ratio of the percentage change in a firm's share value during the
registration period to the percentage offer price adjustment.
Table 5-3. Regressions of Percentage Offer Price Revision and Percentage Change in
Firm Value during the Registration Period on Public Information Measures
Dependent Variable: Test: Coefficients are
(1) Percentage (2) Percentage equal across the two
.> equal across the two
Offer Price Change in Firm m (
models (p value)
Intercept -0.01 0.06**
UP (Same Underwriter) 0.04** 0.19** < 0.001
UP (Other Underwriters) 0.08 0.56 < 0.001
PrcAdj (Same Underwriter) 0.28*** 0.82*** < 0.001
PrcAdj (Other 0.31** 1.04** < 0.001
Industry Ret (Beginning) 1.70*** 3.69** < 0.001
Industry Ret (End) 1.04** 3.39*** < 0.001
Adjusted R2 0.233 0.243
Test: All coefficients = 0
(p value) < 0.001 < 0.001
Number of Observations 4300 3997
The sample includes 4300 US operating firm IPOs during 1985-2003. Unit offers, REITs,
closed-end funds, banks and S&Ls, ADRs, reverse LBOs, and IPOs with minimum of the price
range below $8.00 are excluded. Offer price revision is the offer price minus the midpoint of the
initial filing range, divided by the midpoint of the initial filing range. Change in firm value is the
first-day close share price minus the original expected market price, divided by the original
expected market price. White's heteroskedasticity consistent standard errors are in parentheses.
***, **, and denote significance at the 1, 5, and 10 percent level, respectively.
Table 5-4. System Estimation
Column: (1) (2) (3) (4) (5)
AR Dummy =1 if AR
Dependent Variable: AR AR Dummy 1 if AR AR
Depende(1993-2003) OP > HI (1993-2003)
Estimation Method: OLS OLS S n Heckman ML Heckman ML
Public Information Proxies
UP (Same Underwriter)
UP (Other Underwriters)
PrcAdj (Same Underwriter)
PrcAdj (Other Underwriters)
Industry Ret (Beginning)
Industry Ret (End)
High Debt Dummy
UW Relative Income
Initial P/S Multiple
Table 5-4. Continued
Column: (1) (2) (3) (4) (5)
AR Dummy =1 if AR
Dependent Variable: AR AR Dummy 1 if AR AR
Depend ar(1993-2003) OP > HI A(1993-2003)
Estimation Method: OLS OLS e Heckman ML Heckman ML
Ln (Sales) -0.00 -0.01 0.04 -0.03 -0.04
(0.04) (0.05) (0.01) (0.04) (0.05)
TechDummy -0.01 -0.07 0.30** -0.16 -0.21
(0.12) (0.14) (0.05) (0.14) (0.16)
Net Dummy 0.28 0.28 0.55** 0.09 0.11
(0.21) (0.21) (0.07) (0.21) (0.22)
Underwriter Rank -0.09 -0.10 -0.10 -0.11
(0.07) (0.08) (0.07) (0.07)
Bubble Dummy 0.49* 0.37 0.49* 0.36
(0.26) (0.26) (0.26) (0.26)
Ln (Expected Proceeds) 0.14*
Pure Primary Dummy -0.21**
Intercept 4.43 4.77 -1.52 5.49 5.75
(0.87) (0.96) (0.11) (0.93) (1.02)
Inverse Mills Ratio -0.58 -0.55
Adjusted R2/ Pseudo R2 0.277 0.271 0.138
All coefficients = 0
(n-value) <0.001 <0.001 <0.001 <0.001 <0.001
Number of Observations 902 777 3100 902 777
*The sample includes 902 US operating firm IPOs with upward offer price adjustment over
1985-2003. Unit offers, REITs, closed-end funds, banks and S&Ls, ADRs, reverse LBOs, and
IPOs with minimum of the price range below $8.00 are excluded. Adjustment ratio (AR) is
defined as the ratio of the change in a firm's share value during the registration period to the offer
price adjustment. The definition of independent variables is as in the Appendix. Columns (1) and
(2) are estimated using OLS. Column (4) and (5) are estimated using Heckman Maximum
Likelihood methodology, with Column (3) as the selection model. Column (2) and Column (5)
include the All-star Dummy,which is available for the period 1993-2003. Underpricing spillover
variables are orthogonalized with respect to the price adjustment spillover variables and the
market return variables. Price adjustment spillover variables are orthogonalized with respect to
the market return variables. White's heteroskedasticity consistent standard errors are in
parentheses. *** **, and denote significance at the 1, 5, and 10 percent level, respectively.
1985 1987 1989 1991 1993 1995 1997 1999 2001 2003
--Mean AR- -*- -Median AR
Figure 5-1. Mean and Median Adjustment Ratio by Cohort Year. Adjustment Ratio (AR)
is the ratio of the change in a firm's share value during the registration period
to the offer price adjustment. The sample is composed of 1,119 IPOs from
1985 to 2003 with upward price adjustment, excluding unit offers, REITs,
closed-end funds, banks and S&Ls, ADRs, reverse LBOs, and IPOs with the
minimum of the price range below $8.
In this section we conduct various experiments to ensure that our findings are
Endogeneity of Underwriter Selection
Habib and Ljungqvist (2001) and Fernando et al. (2004) argue that the reputation
ranking of the lead underwriter is endogeneous in underpricing regressions. A similar
endogeneity bias may affect the analysis in this study if issuers choose underwriters
based on the bargaining powers of the two sides. We test this possibility using the
Durbin-Wu-Hausman test. We include in the AR regression the residual from a first-stage
regression of the underwriter prestige on various firm- and offer-specific characteristics.
A significant coefficient estimate on the residual indicates an endogeneity bias. In our
test, we cannot reject the null hypothsis that the underwriter prestige is exogeneous
Stability of the First-Stage Regression
The adjustment ratio is a function of initial underpricing, which is estimated using a
first-stage regression of underpricing on a set of variables. To test the stability of this
analysis, we repeat the estimation of initial underpricing by adding other explanatory
variables to the first-stage regression. More specifically, in unreported work we include
the firm-specific volatility, high-leverage dummy, integer offer price dummy, the natural
logarithm of expected proceeds, and the pure primary dummy to the first-stage
estimation. The results remain the same and our conclusions remain unchanged.
The estimation of the initial underpricing is done in the first-stage regression of
IPOs with no offer price revisions that went public throughout the whole sample period
of 1985 to 2003. Therefore, AR for IPOs earlier in the sample period is calculated using
information provided in future IPOs. Even though the time dummies in the first-stage
regression capture this time effect in the intercept, a forward-looking bias could affect the
results if the coefficients differ significantly over time. We repeat the analysis using
rolling-window regressions in the first-stage estimation. More specifically, we estimate
the initial underpricing for each IPO with upward offer price revision using only the
information provided in the pricing of IPOs that went public in the last three years. The
results are qualitatively identical.
Scale Effect on Efficiency
Table III demonstrates a potential scale effect on the efficiency of the offer price
adjustment: higher offer price revisions (output) seem to be achieved only at lower
efficiencies. It is important to ensure that the explanatory variables in the main regression
analysis are not in fact capturing this scale effect. For instance, IPOs with integer offer
prices experience on average higher offer price revisions. Therefore, it is possible that the
inefficiency associated with the pricing of IPOs with integer offer prices is in fact due to
this scale effect. In unreported results we investigate the relation between the efficiency
and the magnitude of the offer price revision in more detail. We find that the positive
relation between the revision and AR disappears after controlling for other factors. In
particular, in a regression of AR on the offer price revision and firm-specific uncertainty,
the coefficient estimate of the offer price revision becomes insignificant and remains
insignificant after other factors are included. It appears that the relation between the offer
price revision and the efficiency of offer price adjustment is spurious: IPOs with a higher
valuation uncertainty tend to have higher offer price revisions conditional on having an
upward offer price adjustment, but at the same time their offer price revisions are less
complete. As an additional test, we include the magnitude of the offer price revision in
our regressions. The coefficient estimate is insignificant and the other coefficients remain
the same. Therefore, we conclude that the findings are not driven by a scale effect.
Investment banks sometimes claim that they have long-run value in mind when
they set IPO offer prices. Therefore, if they believe the market to be overoptimistic in
regards to the valuation of an IPO they may not increase the offer price up to the market
clearing level. Loughran and Ritter (2002) call this the "leaning against the wind" theory.
We test whether the inefficiencies we document in this study are in part due to the
cautious behavior of the underwriters when setting the offer prices. Leaning against the
wind theory predicts a positive relation between the efficiency of the offer price
adjustment and long-run post-IPO returns of the issuing firm: overreaction by the market
results in a seemingly inefficient offer price adjustment process and low returns in the
aftermarket. In unreported tests, we include the returns of the issuing firm during the one
year period following the IPO in our main regression. We find that the coefficient
estimate is insignificant and the other coefficients do not change materially. We conclude
that the partial adjustment phenomenon is not driven by the cautious price setting of the
underwriters in response to the investor sentiment.
How efficiently do the underwriters adjust offer prices in response to the change in
the expected market valuations of the IPO firms? How does this efficiency vary across
IPOs? Which hypothesis of IPO offer price adjustment best explains the pattern?
In this study we examine the level and determinants of the efficiency of the IPO
offer price adjustment process. More specifically, we compare the magnitude of the offer
price adjustment relative to the change in the firm value during the registration period of
IPOs with upward price adjustment. We document that between 1985 and 2003 only 42.0
% of the increase in firm value prior to the IPO was incorporated into offer prices. In the
bubble period, the level of price adjustment relative to the change in firm value dropped
to 28.9%. Inefficiency of offer price adjustment at these levels points to significant costs
and inefficiencies in the IPO pricing process.
We evaluate three hypotheses related to the offer price adjustment process. Under
the extended dynamic information acquisition hypothesis (Benveniste and Spindt (1989),
Benveniste et al. (2002)), underwriters use their discretion to extract information from
investors, thereby reducing the average level of underpricing. This hypothesis implies
that the change in the firm value due to publicly observable industry returns and
information spillovers from comparable IPOs in the same industry should be fully
incorporated into offer prices.
Our results indicate that IPOs whose upward offer price revisions are primarily
driven by private information production rather than publicly available information
experience more efficient offer price adjustments. This finding is inconsistent with the
central implication of the dynamic information acquisition hypothesis that the partial
adjustment phenomenon is due to the partial incorporation of private information. We
find that market returns during the registration period and information spillovers from
recent comparable IPOs are only partially incorporated into the offer prices. Issuing firms
that go public following recent comparable IPOs with high levels of underpricing and
during a period of high returns in the market experience less efficient offer price
adjustments. In addition, the information provided through the offer price adjustment of
recent IPOs is incorporated into the offer prices inefficiently, except when the source of
the information is from the underwriter's own offerings. Overall, our findings do not
support the dynamic information acquisition hypothesis as the explanation of the partial
We next test the predictions of the bargaining hypothesis (Loughran and Ritter
(2002), Ljungqvist and Wilhelm (2003)). The bargaining hypothesis suggests that the
offer prices are determined as a result of a bargaining process between underwriters and
issuers. In the offer price adjustment context, positive information revealed during the
registration period creates a surplus that is divided between the underwriter and the
issuing firm according to their bargaining powers and incentives to bargain.
We show that the efficiency of offer price adjustment is higher for IPO firms with
alternative financing options and with high opportunity cost of underpricing, as suggested
by the bargaining hypothesis. On the other hand, efficiency is lower for issues
underwritten by investment banks with high demand for their services in the short-term
and by those that employ influential research analysts, indicating that attractive outside
options increase the bargaining powers of the underwriters vis-a-vis the issuing firms. We
document a decline in the efficiency of the offer price adjustment over the duration of the
registration period. In addition, a bigger surplus for the issuing firm created at the
preliminary offer price at the beginning of the registration period is taken back through a
subsequent less efficient offer price adjustment. These findings are consistent with a
decline in the bargaining powers of the issuing firms during their IPO pricing processes,
presumably reflecting their lack of options towards the end. We also find that issues with
longer registration periods and those with offer prices set at fractional dollars are
associated with better offer price adjustments, which we interpret as an evidence for the
existence of a negotiation process in the setting of the IPO offer prices.
Our findings do not support the Edelen and Kadlec (2005) tradeoff hypothesis as a
full explanation of the partial adjustment phenomenon. According to the tradeoff
hypothesis partial adjustment is demanded by issuers, and consequently the efficiency of
the price adjustment is independent from the actions and the characteristics of
underwriters. However the evidence on the negative relationship between the demand for
the underwriters' services and the efficiency of the offer price adjustment suggests
significant conflicts of interest between the underwriters and the issuing firms. In
addition, according to the tradeoff hypothesis, the underwriters and issuing firms have no
incentives to engage in a nickel-and-diming strategy. Therefore, whether the final offer
price is set at an integer or a fraction of a dollar should not contain any information, and
should not be significantly related to the efficiency of offer price adjustment. Contrary to
this prediction, we find that issues with offer prices set at integers experience less
complete price adjustments, suggesting the existence of a negotiation process between the
underwriters and the issuers.
Overall, our analysis reveals significant inefficiencies in the IPO offer price
adjustment process and we document a systematic variation in this inefficiency across
IPOs and over time. Our findings do not support the distinct implications of the dynamic
information acquisition hypothesis or the tradeoff hypothesis. The evidence in this study
is consistent with the hypothesis that offer prices are determined as a result of a
bargaining process between underwriters and issuers, as hypothesized by Loughran and
Ritter (2002) and Ljungqvist and Wilhelm (2003).
LIST OF REFERENCES
Asquith, D., J.D. Jones, and R. Kieschnick, 1998, "Evidence on Price Stabilization and
Underpricing in Early IPO Returns," Journal of Finance, 53, 1759-1773.
Barcaskey, M.J., 2004, "Do SEC Regulations Constrain Offer Price Revisions of IPOs?"
working paper, College of William and Mary.
Barondes, R.D.R, 2005, "Correcting the Empirical Foundations of IPO-Pricing
Regulation," working paper, University of Missouri-Columbia.
Benveniste, L.M., and P.A. Spindt, 1989, "How Investment Bankers Determine the Offer
Price and Allocation of New Issues," Journal of Financial Economics, 24, 343-361.
Benveniste, L.M., W.Y. Busaba, and W.J. Wilhelm, Jr., 2002, "Information Externalities
and the Role of Underwriters in Primary Equity Markets," Journal of Financial
Intermediation, 11, 61-86.
Benveniste, L.M., A. Ljungqvist, W.J. Wilhelm, Jr., and X. Yu, 2003, "Evidence of
Information Spillovers in the Production of Investment Banking Services," Journal
of Finance, 58, 577-608.
Booth, J., and R. Smith, 1986, "Capital raising, underwriting and the certification
process," Journal of Financial Economics, 15, 261-281.
Bradley, D.J., J.W. Cooney, B.D. Jordan, and A.K. Singh, 2004, "Negotiation and the
IPO Offer Price: A Comparison of Integer versus Non-Integer IPOs," Journal of
Financial and Quantitative Analysis, 39, 517-540.
Bradley, D.J., and B.D. Jordan, 2002, "Partial Adjustment to Public Information and IPO
Underpricing," Journal of Financial and Quantitative Analysis, 37, 595-616.
Busaba, W.Y., L.M. Benveniste, and R.J. Guo, 2001, "The Option to Withdraw IPOs
During the Premarket," Journal of Financial Economics, 60, 73-102.
Carter, R.B., F.H. Dark, and A.K. Singh, 1998, "Underwriter Reputation, Initial Returns,
and the Long-run Performance of IPO Stocks," Journal of Finance, 53, 285-311.
Carter, R.B., and S. Manaster, 1990, "Initial Public Offerings and Underwriter
Reputation," Journal of Finance, 45, 1045-1067.
Clarke, J.E., C. Dunbar, and K.H. Kahle, 2003, "All-star Analyst Turnover, Investment
Bank Market Share, and the Performance of Initial Public Offerings," working
paper, Georgia Tech University.
Cliff, M., and D.J. Denis, 2004, "Do IPO Firms Purchase Analyst Coverage with
Underpricing?" Journal of Finance, 59, 2871-2901.
Dunbar, C., 2000, "Factors Affecting Investment Bank Initial Public Offering Market
Share," Journal of Financial Economics, 55, 3-41.
Dunbar, C., and S. Foerster, 2005, "Second Time Lucky? Underwriter Switching and the
Performance of Withdrawn IPOs that Return to Market," working paper.
Edelen, R.M., and G.B. Kadlec, 2005, "Issuer Surplus and the Partial Adjustment of IPO
Prices to Public Information," Journal of Financial Economics, 77, 347-373.
Fama, E., and K. French, 1997, "Industry Costs of Equity," Journal of Financial
Economics, 43, 153-193.
Field, L.C., and J. Karpoff, 2002, "Takeover Defenses of IPO Firms," Journal of Finance,
Hall, C., and C. Robbibs-Roth, 1992, "Going Public Without Panic," Recorder, May 6.
Hanley, K., 1993, "The Underpricing of Initial Public Offerings and the Partial
Adjustment Phenomenon," Journal of Financial Economics, 34, 231-250.
Harris, L., 1991, "Stock Price Clustering and Discreteness," Review of Financial Studies,
Heckman, J., 1979, "Sample Selection Bias as a Specification Error," Econometrica, 47,
Helwege, J., and F. Packer, 2004, "Private Matters," working paper.
Hoberg, G., 2004, "Strategic Underwriting in Initial Public Offerings," working paper,
University of Maryland.
James, C., and P. Wier, 1990, "Borrowing Relationships, Intermediation, and the Cost of
Issuing Public Securities," Journal of Financial Economics, 28, 149-171.
Ljungqvist, A., 2005, "Conflicts of Interest and Efficient Contracting in IPOs,"
unpublished working paper, New York University.
Ljungqvist, A., and W.J. Wilhelm, 2002, "IPO Allocations: Discriminatory or
Discretionary?" Journal of Financial Economics, 65, 167-201.
Ljungqvist, A., and W.J. Wilhelm, 2003, "IPO Pricing in the Dot-Com Bubble," Journal
of Finance, 58, 723-752.
Ljungqvist, A., and W.J. Wilhelm, 2005, "Does Prospect Theory Explain IPO Market
Behaviour?" Journal of Finance, 60, 1759-1790.
Ljungqvist, A., T.J. Jenkinson, and W.J. Wilhelm, 2003, "Global Integration in Primary
Equity Markets: The Role of U.S. Banks and U.S. Investors," Review of Financial
Studies, 16, 63-99.
Ljungqvist, A., V. Nanda, and R. Singh, 2004, "Hot Markets, Investor Sentiment, and
IPO Pricing," Journal of Business, forthcoming.
Logue, D., 1973, "Premia on Unseasoned Equity Issues, 1965-69," Journal of Economics
and Business, 25, 133-141.
Loughran, T., and J.R. Ritter, 2002, "Why Don't Issuers Get Upset About Leaving
Money on the Table in IPOs?" Review of Financial Studies, 15, 413-443.
Loughran, T., and J.R. Ritter, 2004, "Why Has IPO Underpricing Increased Over Time?"
Financial Management, 33 (3), 5-37.
Lowry, M., and G.W. Schwert, 2004, "Is the IPO Pricing Process Efficient?" Journal of
Financial Economics, 71, 3-26.
Mola, S., and T. Loughran, 2004, "Discounting and Clustering in Seasoned Equity
Offering Prices," Journal of Financial and Quantitative Analysis, 39, 1-23.
Nanda, V., Y. Yun, 1997, "Reputation and financial intermediation: an empirical
investigation of the impact of IPO mispricing on underwriter market value,"
Journal of Financial Intermediation, 6, 39-63.
Ritter J.R., 1984, "The 'Hot Issue' Market of 1980," Journal of Business, 57, 215-240.
Ritter J.R., 1987, "The Costs of Going Public," Journal of Financial Economics, 19, 269-
Sternberg, T., 1989, "Bilateral Monopoly and the Dynamic Properties of Initial
Public Offerings," working paper, Vanderbilt University.
Ozgir (Ozzie) Ince earned his Bachelor of Science in mechanical engineering from
Istanbul Technical University in 1999. He earned his Master of Science in mechanical
engineering with a concentration on fluid dynamics and heat transfer from the University
of Michigan in 2001. The requirements for the degree of Doctor of Philosophy in finance
were completed in May 2006 at the University of Florida.
viiLISTOFTABLESTable page 4-1DefinitionofVariables.............................................................................................315-1RegressionsofFirst-DayReturnsonIPOsCategorizedbyPriceAdjustment........505-2DescriptiveStatisticsfortheAdjustmentRatio.......................................................515-3RegressionsofPercentageOfferPriceRevisionandPercentageChangeinFirmValueduringtheRegistrationPeriodonPublicInformationMeasures..................525-4SystemEstimation....................................................................................................53
viiiLISTOFFIGURESFigure page 5-1MeanandMedianAdjustmentRatiobyCohortYear..............................................55
18CHAPTER3METHODOLOGYEstimationoftheAdjustmentRatioInthisstudyweexaminethelevelanddeterminantsofefficiencyintheIPOofferpriceadjustmentprocess.Wedefineefficiencyasthemagnitudeoftheofferpriceadjustmentrelativetothehypotheticalcompletelevelofofferpriceadjustment.Wepositthatthecompletelevelofofferpriceadjustmentisequaltothechangeintheissuingfirmsvalueduringtheregistrationperiod.Inotherwords,iftheunderwritersrevisetheofferpriceasmuchasthechangeinthefirmvalue,thentheofferpriceadjustmentiscompleteandtheeventualunderpricingisequaltotheunderpricingthatwouldhavebeenobservediftheissuingfirmsvaluehadnotchanged.WeconstructtheAdjustmentRatio(AR),whichisdefinedastheestimatedpercentagechangeinanIPOfirmssharevaluedividedbythepercentageofferpricerevisionduringtheregistrationperiod:1 0(MVMV)ii 0 MV i 0(OPOP)ii 0 OP i ARi(3-1) 1Thereisnotheoreticalreasonforusingtheratioofthetwoastheefficiencymeasure.Forexample,usingthedifferencebetweenthepercentagechangeinthefirmvalueandthepercentageofferpricerevisionistheoreticallyvalid.However,empiricalevidenceindicatesthatthefrictionsassociatedwiththeofferpriceadjustmentprocessaremultiplicativeratherthanadditive.Additivefrictions,suchasaconstantdifferenceregardlessofthelevelofofferpricerevision,wouldleadtoanegativerelationbetweentheofferpricerevisionandtheunderpricingwhichisnotthecaseasdocumentedinnumerousstudies.Inaddition,avalidefficiencymeasureshouldbefreeofscaleeffects.Inotherwords,theefficiencyshouldnotberelatedtothemagnitudeoftheofferpricerevision.Inunreportedresultswefindthattheratiomeasureisnotrelatedtotheofferpricerevisionceterisparibus,whereasthedifferencemeasureispositivelyrelatedtoit.Therefore,weconcludethattheratioisavalidefficiencymeasure.
20WeusethecoefficientsfromthisOLSregressiontoestimatewhattheunderpricingwouldhavebeenforIPOswitheventualofferpricerevisionsifthevaluesoftheissuingfirmshadnotchangedduringtheregistrationperiodandtheofferpriceswerenotadjusted.ThisanalysisisessentiallyequivalenttodecomposingthetotalunderpricingofIPOswithofferpricerevisionsintotwo:(i)initialexpectedunderpricingcommontoallIPOs,independentoftheofferpricerevision,and(ii)additionalunderpricingduetothepartialofferpriceadjustment.Then,weusetheestimatedinitialunderpricingofIPOswithofferpricerevisionstocalculatetheexpectedmarketpriceofsharesatthetimeoftheoriginalfilingusing 00 0 MV(UP1)OP ii i (3-3)whereOPi0isthemidpointoftheoriginalfilepricerange,andistheestimatedinitialunderpricingusingthecoefficientsfromthemodelinEquation(2).Thenweplug-inMVi0inEquation(1)tocalculateARforallIPOswithofferpricerevisions.TheAdjustmentRatiomeasurestheinefficiencyofIPOofferpriceadjustments.Ifofferpricesareadjustedfullysuchthatthepercentageofferpricerevisionisequaltothepercentagechangeintheissuingfirmsvalueduringtheregistrationperiod,thenARequalsone.Otherwise,thehighertheAR,thelowertheadjustmentoftheofferpricerelativetothechangeinthemarketvalue,andhencethelowertheefficiency.AftercalculatingARiforalldealswithofferpricerevisionsusingthemethodologyexplainedabove,weinvestigatethedeterminantsofvariationinARiacrossdealsusingtheOLSregressionARi=f(Firm-,Underwriter-,Offer-specificcharacteri stics,PublicInformationProxies) (3-4)
24proxiessincethecoefficientsmaycaptureadditionalinformationunrelatedtopubliclyavailableinformation.WecontrolforthispotentialsampleselectionbiasusingtheHeckman(1979)maximumlikelihoodestimation(MLE)procedure.4HeckmanMLEprocedureisthejointestimationofthefollowingtwoeconometricmodels: upwardupwardupwardupward = IX+ (3-5) ARARIMRAR =ARX+IMR (3-6)whereIupwardisanindicatorfunctionforanupwardofferpricerevisionandisequaltooneforIPOswithfinalofferpriceshigherthanthemaximumoftheinitialfilepricerangeandzerootherwise.Xupwardincludesthepublicinformationproxiesandvariouscontrolvariables,andthemodelisestimatedusingtheprobitanalysis.ThecorrectionforthesampleselectionbiasisenabledbytheinclusionoftheinverseMillsratiointheARmodel.TheinverseMillsratio(IMR)is: upwardupward (X) upwardupward (X) =IMR (3-7)whereupwardXupwardarethepredictedprobabilitiesofinclusionintheupwardofferpricerevisionsubsamplefromtheprobitregressionofIupwardonXupward,isthestandardnormaldensityfunction,andisthestandardnormaldistributionfunction.IMRisamonotonicallydecreasingfunctionofthepredictedprobabilityofinclusioninthesubsample. 4HeckmanMLEismoreefficientthantheHeckmantwo-stepproceduresincetheselectionandthepredictionmodelsareestimatedjointlyintheMLE.
51Table5-2.DescriptiveStatisticsfortheAdjustmentRatio* PanelA:IPOswithdownwardofferpriceadjustment OfferpriceadjustmentquartileNMean(Median)offerpriceadjustmentMean(Median)changeinthefirmvalueMean(Median)PAR 1(low)245-38.8%-43.5%1.12(-36.8%)(-42.9%)(1.13)2264-26.2%-31.1%1.19(-25.9%)(-32.5%)(1.24)3230-20.1%-26.2%1.31(-20.0%)(-26.1%)(1.30)4(high)261-13.8%-19.7%1.43(-14.3%)(-20.4%)(1.45)All1000-24.6%-30.0%1.26(-23.1%)(-29.8%)(1.24) PanelB:IPOswithupwardpriceadjustment OfferpriceadjustmentquartileNMean(Median)offerpriceadjustmentMean(Median)changeinfirmvalueMean(Median)PAR 1(low)25412.6%24.1%1.90 (13.0%)(19.3%)(1.54)229318.2%38.8%2.14(18.2%)(32.8%)(1.75)328527.6%65.7%2.34(27.3%)(52.9%)(2.01)4(high)28759.8%190.2%3.06(50.0%)(138.0%)(2.55)All111929.7%80.9%2.36(22.2%)(42.6%)(1.94) *Thesampleincludes2119USoperatingfirmIPOswitheitherupwardordownwardofferpricerevisionsduring1985-2003.Unitoffers,REITs,closed-endfunds,banksandS&Ls,ADRs,reverseLBOs,andIPOswithminimumofthepricerangebelow$8.00areexcluded.Offerpricerevisionistheofferpriceminusthemidpointoftheinitialfilingrange,dividedbythemidpointoftheinitialfilingrange.Changeinthefirmvalueisthefirst-dayclosesharepriceminustheoriginalexpectedmarketprice,dividedbytheoriginalexpectedmarketprice.Adjustmentratio(AR)isdefinedastheratioofthepercentagechangeinafirmssharevalueduringtheregistrationperiodtothepercentageofferpriceadjustment.
52Table5-3.RegressionsofPercentageOfferPriceRevisionandPercentageChangeinFirmValueduringtheRegistrationPeriodonPublicInformationMeasures*DependentVariable: (1)PercentageOfferPriceRevision(2)PercentageChangeinFirmValueTest:Coefficientsareequalacrossthetwomodels(pvalue) Intercept-0.010.06*** (0.00) (0.01) UP(SameUnderwriter)0.04***0.19***<0.001(0.01)(0.04)UP(OtherUnderwriters)0.08***0.56***<0.001(0.01)(0.04)PrcAdj(SameUnderwriter)0.28***0.82***<0.001(0.03)(0.09)PrcAdj(Other Underwriters) 0.31***1.04***<0.001(0.03)(0.08)IndustryRet(Beginning)1.70***3.69***<0.001(0.08)(0.27)IndustryRet(End)1.04***3.39***<0.001(0.06)(0.18)AdjustedR20.2330.243Test:Allcoefficients=0(pvalue)<0.001<0.001NumberofObservations43003997 *Thesampleincludes4300USoperatingfirmIPOsduring1985-2003.Unitoffers,REITs,closed-endfunds,banksandS&Ls,ADRs,reverseLBOs,andIPOswithminimumofthepricerangebelow$8.00areexcluded.Offerpricerevisionistheofferpriceminusthemidpointoftheinitialfilingrange,dividedbythemidpointoftheinitialfilingrange.Changeinfirmvalueisthefirst-dayclosesharepriceminustheoriginalexpectedmarketprice,dividedbytheoriginalexpectedmarketprice.Whitesheteroskedasticityconsistentstandarderrorsareinparentheses.***,**,and*denotesignificanceatthe1,5,and10percentlevel,respectively.
54Table5-4.ContinuedColumn:(1)(2)(3)(4)(5) DependentVariable:ARAR(1993-2003)Dummy=1ifOP>HIARAR(1993-2003)EstimationMethod:OLSOLSSelectionModelHeckmanMLHeckmanML ControlVariablesLn(Sales)-0.00-0.010.04***-0.03-0.04(0.04)(0.05)(0.01)(0.04)(0.05)TechDummy-0.01-0.070.30***-0.16-0.21(0.12)(0.14)(0.05)(0.14)(0.16)NetDummy0.280.280.55***0.090.11(0.21)(0.21)(0.07)(0.21)(0.22)UnderwriterRank-0.09-0.10-0.10-0.11(0.07)(0.08)(0.07)(0.07)BubbleDummy0.49*0.370.49*0.36(0.26)(0.26)(0.26)(0.26)InstrumentalVariablesLn(ExpectedProceeds)0.14***(0.03)PurePrimaryDummy-0.21***(0.04)Intercept4.43***4.77***-1.52***5.49***5.75***(0.87)(0.96)(0.11)(0.93)(1.02)InverseMillsRatio-0.58***-0.55***(0.20)(0.23)AdjustedR2/PseudoR20.2770.2710.138----Allcoefficients=0 (p value) <0.001<0.001<0.001<0.001<0.001NumberofObservations9027773100902777 *Thesampleincludes902USoperatingfirmIPOswithupwardofferpriceadjustmentover1985-2003.Unitoffers,REITs,closed-endfunds,banksandS&Ls,ADRs,reverseLBOs,andIPOswithminimumofthepricerangebelow$8.00areexcluded.Adjustmentratio(AR)isdefinedastheratioofthechangeinafirmssharevalueduringtheregistrationperiodtotheofferpriceadjustment.ThedefinitionofindependentvariablesisasintheAppendix.Columns(1)and(2)areestimatedusingOLS.Column(4)and(5)areestimatedusingHeckmanMaximumLikelihoodmethodology,withColumn(3)astheselectionmodel.Column(2)andColumn(5)includetheAll-starDummy,whichisavailablefortheperiod1993-2003.Underpricingspillovervariablesareorthogonalizedwithrespecttothepriceadjustmentspillovervariablesandthemarketreturnvariables.Priceadjustmentspillovervariablesareorthogonalizedwithrespecttothemarketreturnvariables.Whitesheteroskedasticityconsistentstandarderrorsareinparentheses.***,**,and*denotesignificanceatthe1,5,and10percentlevel,respectively.
55Figure5-1.MeanandMedianAdjustmentRatiobyCohortYear.AdjustmentRatio(AR)istheratioofthechangeinafirmssharevalueduringtheregistrationperiodtotheofferpriceadjustment.Thesampleiscomposedof1,119IPOsfrom1985to2003withupwardpriceadjustment,excludingunitoffers,REITs,closed-endfunds,banksandS&Ls,ADRs,reverseLBOs,andIPOswiththeminimumofthepricerangebelow$8. 00.511.522.533.541985198719891991199319951997199920012003 MeanAR MedianAR