Walden University

Walden University
COLLEGE OF MANAGEMENT AND TECHNOLOGY
This is to certify that the doctoral dissertation by
Zachary A. Smith
has been found to be complete and satisfactory in all respects,

and that any and all revisions required by
the review committee have been made.
Review Committee
Dr. Mohammad Sharifzadeh, Committee Chairperson,
Applied Management and Decision Sciences Faculty
Dr. William H. Brent, Committee Member,

Dr. Robert T. Aubey, Committee Member,

Chief Academic Officer
Denise DeZolt, Ph.D.
Walden University
2008
ABSTRACT
An Empirical Investigation of Initial Public Offering (IPO) Price Performance
by
Zachary Alexander Smith
M.B.A., Capella University, 2005

B.S., Old Dominion University, 2004
Dissertation Submitted in Partial Fulfillment

of the Requirements for the Degree of
Doctor of Philosophy
Applied Management & Decision Science
Walden University
August 2008
ABSTRACT
For decades, researchers have disagreed on the magnitude and predictability of abnormal
securities’ price performance generated by initial public offerings (IPOs). The problem
researched in this study was (a) whether IPOs generate abnormal price performances
during their initial years of public trading and (b) if IPOs systematically generate
abnormal price performance, then the identification of predictable abnormal price
performance is in conflict with the theory of market efficiency. The purpose of this study
was to identify the best specified and most powerful method of abnormal performance
detection and apply that method to examine the price performance of IPOs. Matched by
size, industry, and book-to-market ratios this study explored which of the resulting seven
portfolios and matched-firm methods of abnormal performance detection produced the
best specified and most powerful test statistics. Additionally it considers if IPOs generate
abnormal performance. The researcher used the event study approach for the research
design along with the buy and hold abnormal return (BHAR) method of aggregating
returns to conduct this analysis. The findings were that (a) all of the matched-firm
methods of abnormal performance detection were best specified and most powerful and
(b) the IPOs generated statistically significant abnormal price performances in all of the
hypotheses tested. These findings are in conflict with the theory of market efficiency and
imply the existence of social injustice in the IPOs’ premarket pricing and allocation
process. Revisions to the IPO premarket pricing and allocation method may lead to
significant positive social change; the revisions could prevent unfair advantages gained
by those investors with privileged premarket IPO allocations and could provide equal
opportunities for investors to benefit from the IPO process.

An Empirical Investigation of Initial Public Offering (IPO) Price Performance
by
Zachary Alexander Smith
M.B.A., Capella University, 2005

B.S., Old Dominion University, 2004
Dissertation Submitted in Partial Fulfillment

of the Requirements for the Degree of
Doctor of Philosophy
Applied Management & Decision Science
Walden University
August 2008
3325352
3325352
2008
DEDICATION
I dedicate this dissertation to my children—Cooper and Harris. Without your
unconditional love and understanding, I do not think I would have completed this
process. Thank you and I love you.

ACKNOWLEDGMENTS
I would first like to thank my mother for giving me the drive to attempt to obtain
my Ph.D., and my father for being there for me throughout completion of my degree. In
addition, I would like to thank my dissertation committee for their assistance: Dr.
Mohammad Sharifzadeh, CFA (Faculty Mentor/Dissertation Committee Chair), Dr.
William Brent (Dissertation Committee Member), and Dr. Robert Aubey (Dissertation
Committee Member). Finally, I would like to thank my wife of eight years, Tamara, for
suffering through this and my other academic adventures.
ii
TABLE OF CONTENTS
DEDICATION.................................................................................................................... ii
TABLE OF CONTENTS................................................................................................... iii
LIST OF TABLES............................................................................................................ vii
LIST OF FIGURES ......................................................................................................... viii
CHAPTER 1: INTRODUCTION TO THE STUDY...........................................................1

Introduction....................................................................................................................1
Background of the Study ...............................................................................................4
Problem Statement .........................................................................................................6
Purpose of the Study ......................................................................................................7
Research Questions and Hypotheses .............................................................................7
Section 1: Specification and Power ........................................................................ 8
Section 2: Initial & Subsequent Short-Term Performance ..................................... 9
Section 3: Longer-Term Performance .................................................................. 12
Section 4: The Quiet Period and Lockup Expiration ............................................ 12
Theoretical Base...........................................................................................................13
Definition of Terms......................................................................................................16
Assumptions.................................................................................................................18
Limitations ...................................................................................................................19
Significance of the Study.............................................................................................20
Social Change ..............................................................................................................21
Summary......................................................................................................................22
CHAPTER 2: LITERATURE REVIEW ...........................................................................23

IPO Process..................................................................................................................23
Anomalous IPO Events................................................................................................25
Initially Abnormally Positive Performance .......................................................... 25
Long-term Negative Performance......................................................................... 27
The Quiet Period ................................................................................................... 29
Lockup Expiration ................................................................................................ 30
Securities’ Pricing, Estimating Returns, and Efficient Markets ..................................31
Methodological debate.................................................................................................33
Measuring Normal Performance........................................................................... 33
Benchmark Construction ...................................................................................... 34
Benchmark Construction ................................................................................ 35
Measuring Abnormal Performance....................................................................... 40
BHAR vs. CAR............................................................................................... 41
Summary......................................................................................................................43
iii
CHAPTER 3: RESEARCH METHOD .............................................................................45
Introduction..................................................................................................................45
Research Design and Methodology .............................................................................45
Justification for Using the Event Study Methodology.......................................... 45
Approach to Implementing the Event Study Methodology .................................. 49
Setting and Sample ......................................................................................................49
Canvassing a Significant Target Population ......................................................... 50
Data Collection ............................................................................................................51
Data Analysis ...............................................................................................................52
The Variables ........................................................................................................ 52
The Hypotheses of Research................................................................................. 53
Section 1: Specification and Power Analysis ................................................. 53
Section 2: Initial and Subsequent Short-term Performance............................ 56
Section 3: Longer-term Performance.............................................................. 58
Section 4: The Quiet Period & Lockup Expiration......................................... 59
Summary......................................................................................................................61
CHAPTER 4: RESULTS...................................................................................................63
Introduction..................................................................................................................63
Specification and Power ..............................................................................................63
Test of the Hypotheses.................................................................................................64
Section 1: Test for Specification........................................................................... 64
Section 2: Tests of Power ..................................................................................... 68
Section 3: Initial Performance .............................................................................. 74
IPO Performance (Pre-issuance)..................................................................... 75
Initial Day of Public Trading .......................................................................... 77
Section 3: Long-term Abnormal Performance...................................................... 78
Section 4: Quiet and Lockup Expiration............................................................... 80
Conclusions..................................................................................................................81
CHAPTER 5: SUMMARY, CONCLUSION, AND RECOMMENDATIONS ...............83
Introduction........................................................................................................................83
Overview......................................................................................................................83
Interpretation of findings .............................................................................................85
Specification and Power Analyses........................................................................ 86
Biases ............................................................................................................. 86
Matched-Firm vs. Matched-Portfolio Approach............................................. 87
Matched Firm—Specification and Power....................................................... 89
Short-term IPO Performance ................................................................................ 90
Long-term IPO Performance................................................................................. 92
Event-related IPO Performance ............................................................................ 93
Implications for Social Change....................................................................................95
Short-term IPO Performance and Income Disparity in the United States. ........... 95
iv
Chinks in the Armor of Market Efficiency ........................................................... 97
Integration and Summary...................................................................................... 99
Validity and Reliability of the Research Results .......................................................101
Recommendations for Action ....................................................................................104
Recommendations for Further Study .........................................................................105
Conclusion .................................................................................................................107
REFERENCES ................................................................................................................109
APPENDIX A:.................................................................................................................115
Power Analysis—1-Year Event Horizon—n=50.......................................................115
APPENDIX B ..................................................................................................................116
Power Analysis—1-Year Event Horizon—n=500.....................................................116
APPENDIX C ..................................................................................................................117
Power Analyses—2-Year Event Horizon—n= 50.....................................................117
APPENDIX D..................................................................................................................118
Power Analyses—2-Year Event Horizon—n= 500...................................................118
APPENDIX E ..................................................................................................................119
APPENDIX F...................................................................................................................120
APPENDIX G..................................................................................................................121
APPENDIX H..................................................................................................................122
APPENDIX I ...................................................................................................................123
Average Initial Day IPO Performance by Year and Month.......................................123
APPENDIX J ...................................................................................................................126
Yearly Average of Pre-Trade Performance 1997-2007 .............................................126
APPENDIX K..................................................................................................................127
Yearly Abnormal Performance at the Expiration of the Quiet Period.......................127
APPENDIX L ..................................................................................................................128
Abnormal Performance & Lockup Expiration...........................................................128
APPENDIX M .................................................................................................................129
v
Popular Benchmarks used to Represent Normal Performance ..................................129
APPENDIX N..................................................................................................................130
Methods used in Long-term Event Studies................................................................130
CURRICULUM VITAE..................................................................................................131
vi
LIST OF TABLES
Table 1. Specification analysis I ..................................................................................... 663

Table 2. Specification analysis II.................................................................................... 674
Table 3. Pre-public trading averaged monthly IPO performance ..................................... 73
vii
LIST OF FIGURES
Figure 1. One-year power curve. ...................................................................................... 69

Figure 2. One-year power curve (n=500). ........................................................................ 70
Figure 3. Four-year power curve (n=50)........................................................................... 71
Figure 4. Four-year power curve (n=500)......................................................................... 72
Figure 5. Long-term abnormal performance..................................................................... 79
Figure 6. BHAR (IPO V. Matched-firm).......................................................................... 80
viii
CHAPTER 1:
INTRODUCTION TO THE STUDY
Introduction
In this study, I have analyzed events that influence the price performance of
unseasoned initial public offerings (IPOs). Recently researchers have spent considerable
time debating whether IPOs produce abnormal performance (see Brav, Geczy, &
Gompers, 2000; Brown & Weinstein, 1985; Cheng, Chueng, & Po, 2004; Schultz, 2003).
Researchers have formed a substantial empirical base to support the hypotheses
concluding that IPOs have generated abnormal market performances (see Affleck-
Graves, Hedge, & Miller, 1996; Ibbotson, 1975; Loughran & Ritter, 2004; Reily &
Hatfield, 1969). In this study, I tested the specification and power of different
methodological approaches used to identify abnormal performance and the prevalence of
abnormal performance associated with events occurring in response to the issuance of
unseasoned IPOs. Both of these concepts are critical components of the research project.
Therefore, I will address them individually in the following paragraphs.
Researchers choose to use benchmarks or market proxies to model normal
performance when examining IPO performance because IPOs lack historical performance
data. Without historical performance data, researchers have limited ability to produce
reliable estimates of expected return. To construct a model of normal performance that
does not rely on the correlation of the market and event firms’ historical performance,
researchers need to find a method that captures systematic and idiosyncratic risks
involved in a security’s price performance. Researchers have substituted models based

2
upon firm-specific characteristics to gauge normal performance (referred to in this
document as the matched-firm approach). To ensure that researchers are using the best
criteria to model normal performance they must conduct specification and power
analyses on the methods they use to detect abnormal performance. The majority of the
problems affecting the results of event studies occur when the event horizon is
lengthened, however, Barber and Lyon (1997) and Lyon, Barber, and Tsai (1999)
provided examples of biases that influence the results of short-term analyses.
The biases that are likely to influence the results of longer-term analyses are,
according to Lyon, Barber, and Tsai (1997) the rebalancing, skewness, and survivorship
biases. I will discuss each of these biases in chapter 2 and make modifications to the
models of normal performance in an attempt to minimize these biases impact on the test
statistics’ accuracy. In addition, Lyon et al. (1997) found that “the analysis of long-run
abnormal returns is treacherous” (p. 198), because of these biases; therefore, the
importance of addressing each of these biases and ensuring that the statistical method
used accurately identifies abnormal performance is a critical component of research
projects testing market efficiency.
After I select a well-specified method that can accurately identify abnormal
performance, I will begin testing hypotheses related to events that occur as companies
issue unseasoned equity shares. The hypotheses related to unseasoned equity issuance
include (a) short-term outperformance, (b) longer-term underperformance, and (c)
abnormal performances surrounding the expiration of the quiet and lockup periods.
Researchers have tested all of these hypotheses using different samples and
3
methodologies. The reasons why I have reexamined these hypotheses include evidence of
biased benchmarks, empirical disagreement regarding the significance of the abnormal
performance, small-sample biases, and a need for researchers to update some of the tests.
Previous studies conducted on IPO performances did not combine tests of the
methodology used to detect abnormal performance and tests of the hypotheses previously
mentioned. This combination is necessary to evaluate the accuracy of the conclusions
derived from the analyses, according to Lyon et al. (1999).
The hypotheses that I have evaluated in this study are as follows. First, IPOs
produce substantially positive abnormal performances, generally between the offer of the
issue and close of trade on the first day of public trading (see Krigman, Shaw, &
Womack, 1999; McDonald & Fisher, 1972; Loughran & Ritter, 2004; Reily & Hatfield,
1969). The magnitudes of these gains are significant; Ritter (2003) found an average
initial IPO return of 18.4% for IPOs listed from 1960-2000. Second, IPOs, on average,
underperform standard benchmarks in studies of longer-term performance. Ritter (1991)
found, that IPOs underperform standard market indices by 27.39% over a 3-year period;
this is equivalent to an investor underperforming the benchmark by 9.13% a year. Third,
events occur in the initial year of trading following an IPO that produce abnormal
performances. These events are the expiration of the quiet and lockup periods. For
example, Bradley, Jordan, and Ritter (2003) showed that the expiration of the quiet
period may cause a significant 4.1% positive performance movement in IPOs (p. 1), and
Field and Hanka (2001), calculated a significantly negative 1.5% return as the lockup
period expires (p. 471). In these cases, abnormal performances seem to occur
4
systematically; if I substantiate the systematic nature of these anomalous events, investors
should adjust their risk/return expectations for these offerings and resulting from these
events.
Background of the Study
The historic price performances of IPOs traded on the New York Stock Exchange
(NYSE), National Association of Securities Dealers Automated Quotation (NASDAQ),
and the American Stock Exchange (AMEX) produce interesting questions regarding what
return investors should require for their invested capital. First, shortly after companies
offer the IPOs to the market, they generate abnormal returns; the magnitude is, according
to Ritter (2003), 18.4%. Performance of this magnitude indicates that the market is either
misvaluing IPOs by pushing the price above its fundamental value, or the underwriters of
IPOs are consistently undervaluing these issues; both of these outcomes do not fit well
within the efficient market paradigm. Second, the mysterious performances of IPOs seem
to continue as they mature; researchers have shown that the IPOs, when compared against
standard benchmarks, obtain longer-term negative performances. For example, Ritter
(1991) found that if investors use a buy-and-hold investment strategy, investing in all
IPOs that companies have issued to the public and holding these issues for 3-years, they
would have obtained a -27.39% BHAR.
Along these same lines, at the conclusion of the quiet and lockup periods, IPOs
exhibit abnormal performances in a systematic manner, trending in one direction or
another (positive or negative performance). Bradley, Jordan, and Ritter (2003)
documented that at the expiration of the quiet period IPOs produce an abnormal return of
5
4.1%. Field, and Hanka (2001) and Garfinkle, Malkiel, and Bontas (2002), found that at
the expiration of an IPO’s lockup period, firms produce negative performances ranging
from 1.5 to 4.47%. If investors know that at the end of the lockup period the trading
volume of a security increases because there is an increase in the liquidity provided for
the issue, then investors should incorporate the knowledge of this future event into the
value of the offering at the IPO’s issuance.
Researchers have attempted to document the anomalous nature of IPO
performance, but flaws in the statistical methodologies used to test for abnormal return
cause major disagreements in the interpretation of their findings and conclusions (see
Mitchell & Stafford, 2000; Barber & Lyon, 1997). The flaws in researchers’
methodologies are biases that are evident when researchers use different methods to
aggregate cumulative returns and different portfolios constructions to attempt to
approximate normal performance. Researchers contaminate portfolios because they fall
prey to the rebalancing, skewness, new issue, and survivorship biases, which I will
address briefly in chapter 2. Researchers also disagree on the methodology of aggregation
of abnormal returns generated by IPOs. The two competing methodologies are the BHAR
and the Cumulative Abnormal Return (CAR), which I will also examine in chapter 2. In
addition, the research base is plagued with small sample biases and fragmentation. To
begin to correct the preceding biases and potentially damaging methodological
constructions, I propose the following.

6
Problem Statement
The problems that I will address in this dissertation are threefold: (a) there is
fragmentation in IPO performance literature, (b) researchers have relied on small sample
sizes in examining IPO performance (quiet and lockup expiration), and (c) previous
analyses are discredited because the methodological approach used to detect abnormal
performance is misspecified and or not powerful. The questions addressed in this study,
using a sample of unseasoned IPOs issued from 1985-2002, are:
1. Does the pairing of the matched-firm benchmark approach with the BHAR
methodology produce well-specified test statistics for identifying
abnormal performance in event studies? Inquiry was based upon the
research findings presented in Barber and Lyon (1997), Mitchell and
Stafford (2000), and Gutts, Mills, and Roberts (1997).
2. I relied on the results and conclusions reached in Ritter and Welch (2002)
and Cheng, Cheung, and Po (2004) to examine questions related to
whether the returns generated shortly after the unseasoned IPO’s issuance
is positive and significant and whether this abnormal performance is
constrained to the period prior to public trading was based upon results
and conclusions reached in.
3. Is the relationship between long-term negative abnormal performance and
IPO issuance real? If it is, is this negative abnormal performance
significant? I used a similar approach to Ritter (1991) methodology to
construct my approach to analyze longer-term abnormal performance.

7
4. Is there significant positive abnormal performance at the conclusion of the
quiet period? I based my inquiries upon the findings and conclusions
reached in Bradley, Jordan, and Ritter (2003) and Bradley, Jordan, Ritter,
and Wolf (2004).
5. Does a negative abnormal performance at the conclusion of the lockup
period exist? If it does, what is the significance of it? I have based the
analysis of abnormal performance at the expiration of the lockup period
upon the findings and conclusions generated in Field and Hanka (2001)
and Garfinkle, Malkiel, and Bontas (2002).
After I identified the best-specified approach to use for identifying abnormal returns, the
hypotheses were tested and I evaluated the hypothesized relationships using a
consistently applied method.
Purpose of the Study
The purpose of this study is to examine the price performance of IPOs using a
well-specified and powerful method of detecting abnormal performance: (a) at its public
offering and (b) at subsequent time intervals in response to this offering. The goals of this
research project was to investigate theories related to the performance of IPOs, and to test
these theories using historical pricing data for IPOs that were issued from January 1985
to 2002.
Research Questions and Hypotheses
In the ensuing section, I have outlined the hypotheses of this research project.
There are four sections: (a) specification and power analysis, (b) short-term performance,
8
(c) longer-term performance, and (d) events related to IPO performance. The results of
the first section influenced the remainder of the analyses. Therefore, readers should
evaluate the hypotheses in the order that they are presented and then collectively.
Section 1: Specification and Power
In this section, I have outlined how I executed the specification and power
analyses in this research project. Two different techniques functioned as measures of
normal performance for the event firm, these approaches were: (a) the matched-firm
approach and (b) the portfolio-matching approach. Using both techniques, I matched the
event firms based upon a combination of the three different firm-specific variables: (a)
market capitalization, (b) industry affiliation, and (c) market-to-book ratios.
After the specification tests were completed, I chose the best-specified benchmark
of normal performance and used this method to conduct the remainder of the analyses.
For my first hypothesis test, I took data from 1985-2002 and conducted 10 nonrepeating
random samples of 50 companies per year; with 180 samples distributed evenly across
the 18- years. For each benchmark and sample, I matched the benchmark return against
the simulated event firm and the BHAR was calculated.
Hypothesis 1:
H0: Mean BHAR is zero.

H1: Mean BHAR is not zero.
After running the preceding tests on the 180 samples of 50 companies, I
calculated the Empirical Size (ES) statistic. I then compared the ES statistic against the
constructed confidence interval and checked the specification of the method. Next, I
9
combined the results of this analysis with the ensuing power analysis to select a
benchmarking method.
In the power analysis, I took the same samples that were used in the specification
analysis and simulated abnormal performances at different intervals to determine when
each of the six methods reacted to the simulated abnormal performances of +/- 1, 5, 10,
15, 30, 50, and 75%.
Hypothesis 2:
H0: Mean BHAR is zero

H1: Mean BHAR is not zero
I then combined the results of the individual tests, 180 samples of 50 companies, and
calculated the Empirical Power (EP)—I calculated this statistic by dividing the number of
times I rejected the null hypothesis by the number of tests conducted.
Section 2: Initial & Subsequent Short-Term Performance
In section 2, I have evaluated whether IPOs, upon their initial offering, generate
substantial abnormally positive performance results--when compared against the
expected return on the IPO. There are dissenting academic opinions regarding when the
initial abnormal IPO performance occurs, whether it occurs, and which statistical
methods can identify these abnormal performances. I have contrasted Loughran and
Ritter (2004) against Cheng, Cheung, and Po (2000) to provide a basis to explore the first
debate, and Mitchell and Stafford (2000) compared with Ritter (1991) and Barber and
Lyon (1997) to provide detailed information pertaining to the second debate.
In reference to when abnormal IPO performance occurs, Loughran and Ritter
(2004) claim that the initial underperformance is generated in the first day of trading,
10
whereas Cheng, Cheung, and Po (2000) conclude that the abnormal trading profits are
only available prior to the opening of public trading. I have provided evidence to quell
this debate and provided significant results that should illustrate whether initial
abnormally positive performance is obtainable between the offer and the first day of
trade, during the first day of trade, or during both periods.
When researchers attempt to provide evidence as to whether the abnormal
performance actually occurs, a disagreement sometimes arises. This disagreement
generally stems from the researcher’s choice of using the BHAR or the CAR method to
calculate the aggregated abnormal performance. Based upon the formulaic properties
used to calculate the two methods of aggregation, which I have discussed in chapter 2, I
have chosen to use the BHAR method over the CAR. I made this decision because the
BHAR better represents the abnormal return an investor (buy and hold) would obtain
over an identified period; researchers’ compound BHARs in their calculations, however,
they do not compound CARS.
Researchers have shown the BHAR method of aggregating returns produces
overstated test-statistics. This occurs when researchers use portfolios as proxies for
normal returns, according to Barber, and Lyon (1997). However, when researchers use
the matched-firm technique their results do not suffer from the same statistical problems.
I have illustrated how these biases influence the results of the study in chapter 2 and
design a method in chapter 3 to test the power and statistical significance of the different
methods.
11
In Hypothesis 3, I have determined whether IPOs produce abnormal performance
in the period prior to public trading using the following hypothesis test.
Hypothesis #3:
H 0 : R IPOAverage ( Offering → InitialTrade ) ≤ 0

H 1 : R IPOAverage ( Offering → InitialTra de ) > 0
In the second test of this section, Hypothesis 4, I have compared the aggregate average
monthly performance obtained investing in IPOs prior to public trading against the
average performance of the best performing, in this period, standard market index. I
relied on standard indices to estimate normal performance because the performances
occur prior to public trade; I cannot match the event firm to a portfolio or matched-firm
prior to public trade.
Hypothesis 4:
H 0 : R IPOMonthly (Offering → InitialTrade ) ≤ R DJIAMonthly

H 1 : R IPOMonthly (Offering → InitialTrade ) > RDJIAMonthly
By testing Hypothesis 5, I have determined whether the abnormal performances
that occur prior to the issuance of unseasoned equity continue after companies issue their
shares. In this analysis, I have matched the returns of the IPO against the best-specified
and most powerful benchmark identified in Hypotheses 1 and 2.
Hypothesis 5:
H 0 : R IPO ( Day 1) ≤ RIndustyMat chedFirm

H 1 : R IPO ( Day 1) > R IndustyMat chedFirm
12
After Hypotheses 2 through 5 were tested, I had determined whether initial abnormally
positive performance was generally constrained to the period preceding the issuance of
shares.
Section 3: Longer-Term Performance
If the hypothesized relationships in Hypotheses 3 and 4 are correct, then using
either the offering price or the initial trade as the starting price will affect the actual
estimate of long-term abnormalities in IPO performance. Therefore, I did not include the
performance obtained in these two periods to analyze longer-term IPO performance. The
general question I was interested in answering was whether IPOs significantly
underperform my benchmarks, and when the underperformance occurs.
Hypothesis 6:
H 0 : BHAR ( TradingDay 2 − 350 ) = 0

H 1 : BHAR ( TradingDay 2 − 350 ) ≠ 0
I chose to focus on three years of data because the metrics tested in Hypotheses 1 and 2
deteriorate significantly, in terms of power, as I increased the event horizon beyond 3-
years (see Appendix E - H).
Section 4: The Quiet Period and Lockup Expiration
In this section of the analysis, I was interested in determining whether IPOs
exhibit significant abnormal performance in the 5-day period occurring at the end of the
quiet and lockup expiration. I have used the five trading-day period surrounding the
expiration of the quiet and lockup periods as the measurement period. First, for the quiet
period, at the conclusion of the quiet period analyst can issue opinions pertaining to the
13
newly issued shares fundamentals, the analysts’ opinions are overwhelmingly positive
and induce a significantly positive performance at the conclusion of this period.
Therefore, I proposed the following hypothesis test:
Hypothesis 7:
H 0 : R Quiet ≤ R IndustyMatchedFirm
H 1 : R Quiet > R IndustyMatchedFirm
As the shares of the newly issued IPO approach the expiration of the lockup period, the
officers of the company and other powerful shareholders can sell their shares on the open
market. Prior to this expiration, they are constrained. Because of the increase in volume
of shares trading, of the available for public trade shares, and the likelihood that powerful
shareholders will tend to diversify their holdings, it was my conjecture that the shares
will generate a significantly negative performance event.
Hypothesis 8:
H 0 : R Lock −up ≥ R IndustyMat chedFirm

H 1 : R Lock −up < RIndustyMatchedFirm
By forwarding, theses hypotheses I was interested in addressing whether or not the
markets react to events related to IPO issuance. I will now discuss how this concept of
market efficiency evolved, and how researchers have applied event studies to detect
market inefficiencies.
Theoretical Base
The theoretical basis of this evaluation begins with the conceptual market
framework suggested by Bachelier (1900/1964), Cowles and Jones (1937), and

14
Alexander (1961) and defined by Fama (1965), which resulted in the concept of market
efficiency. Bachelier (1900/1964) wrote, “At any given time, the market believes in
neither a rise nor a fall of true prices” (p. 26). Bachelier’s work went relatively
unrecognized until the 1960s, but other researchers independently discovered the
sentiments of this theorist’s conjectures. Thirty-seven years later, Cowles and Jones
(1937) found that the aggregate monthly stock price changes exhibited specific patterns
from 1835-1935; namely, if the market rose (fell) in the preceding month, it had a 65%
chance of rising (falling) in the current month (pp. 282-283). From this, after some
careful analysis, Cowles et al. (1937) concluded that there are systematic performances
obtained in the market that contrast with the fair-game market concept. Alexander (1961)
took this conclusion further, he illustrated that investors could generate profitable trading
strategies using knowledge of these patterns.
Fama (1965) agreed with some of the conclusions reached by Alexander (1961).
Specifically, that securities performances exhibit certain dependencies. However, Fama
claimed that when investors consider the profit derived from trading strategies used to
exploit the aforementioned departures from normality, the benefits are minuscule.
Moreover, Fama stated that Alexander presupposes that the investor applying his trading
rules would buy (sell) the securities at their lowest (highest) values; this is a faulty
assumption because securities markets are difficult, if not impossible, to time. As
researchers discovered anomalies that do not fit within theoretical boundaries of the
concept of market efficiency, they are continuously questioning the concept of market
efficiency. However, this concept of market efficiency provides the fundamental basis
15
from which researchers base their evaluation of securities’ price performance; it would
suffice to state that it is the bedrock of modern finance.
In the current research project, I define Fama’s efficient market hypothesis (EMH;
Fama, 1976) as follows. First, at any given time, securities’ prices reflect their
fundamental value. Second, if information is available, the market (the market in this case
would be all investors) readily consumes and interprets the information correctly. In the
words of Fama (1970), the implied relationship between market efficiency and expected
returns rests on:
The assumption that the conditions of market equilibrium can be stated in

terms of expected returns and that equilibrium expected returns are formed
on the basis of (and thus “fully reflect”) the information set Φt have a
major empirical implication--they rule out the possibility of trading
systems based only on information Φt that have expected profits or returns
in excess of equilibrium expected profits or returns. (p. 384)
This hypothesis is highly controversial because researchers have illustrated that there
exist potential flaws in the theory of market efficiency (e.g., IPO performance). Since
Fama (1976) articulated the EMH, analysts and academics alike have been carefully
analyzing market efficiency, some finding trading rules and evidence against the efficient
market paradigm and some reporting results in support of this hypothesis.
Two subsets of research project’s analyzing the efficient market quandary have
received a permanent position in the literature focused on testing market efficiency: (a)
those attempting to build an appropriate methodological platform from which they may
test market efficiency, and (b) those interested in testing market efficiency. Ball and
Brown (1968) pioneered the first group of research and Fama, Fisher, Jensen, and Roll
(1969) designed the application of the event study methodology to test market efficiency.
16
This event study methodology provided the framework to test different models that
attempt to measure abnormal returns. The second groups of researchers have tested
whether events generate abnormal performance (see Affleck-Graves, Hedge, & Miller,
1996; Ibbotson, 1975; Loughran & Ritter, 2004; Reily & Hatfield, 1969 for examples of
tests run on IPO-related events). Because any tests of market efficiency are joint tests of a
model of normal performance combined with tests of market efficiency it is appropriate
to deal with each separately. Market efficiency and models of normal performance form
the foundation from which this analysis will progress. In chapter 2, I will thoroughly
analyze IPO performance, ensuring to consider elements from both subsets of the
following questions: (a) how should empiricists measure normal and abnormal
performance and (b) how might IPOs generate abnormal performance?
Definition of Terms
Efficient market hypothesis: “A market in which prices always ‘fully reflect’
available information is called ‘efficient’” (Fama, 1970, p. 383). Fama proceeded to rank
the forms of market efficiency in terms of the information that concern analysts. In the
weak form of this hypothesis, the information component is historical stock prices. In the
semi-strong form, the information the analyst is concerned with is whether the market
adjusts to other information that is available to the public. Finally, in the strong form,
researchers testing market efficiency are concerned with certain investors who possess
monopolistic access to certain information (p. 383).

17
Initial public offering (IPO): The IPO is an event in the history of a company
where a private company first sells its shares to the public and thus changes its ownership
structure from a private enterprise to a publicly traded company.
Issuance (of the IPO): After the 20-day ‘waiting period’, which is given to
investors to analyze the registration statement (Simon, 1989, p. 297), the SEC allows
companies to issue their shares to the public and list on a designated exchange. The
issuance occurs the day of the IPO, or when shares begin to trade on their public
exchange.
Lockup period (of the IPO): An agreement between the underwriter and the issuer
that ensures that any shares held by insiders (management, directors, employees, and
affiliated parties) after the offering, will not be sold until 180 days after the registration
statement becomes effective1.
Offering (of the IPO): The offering of price of a security is the price that an initial
investor must pay to receive an allocation of shares in the offering2. This occurs after the
registration statement has been filed with the SEC and occurs on the day prior to or the
day of the public offering.
Quiet period (of the IPO): The quiet period is a period of time that extends from
the time a company files its registration statement with the Securities and Exchange
Commission (SEC) to the time the SEC declares the statement is effective. During this
period the SEC places restrictions on how the newly issued companies’ staff and its
1
see: http://www.sec.gov/investor/pubs/analysts.htm
2
see: http://moneycentral.hoovers.com/global/msn/index.xhtml?pageid=1954
18
representative are to communicate to the public3; such restrictions allow the market and
investors enough time to determine the value of the security without influence from the
firm’s management or analysts (Bradley, Jordan, Ritter, & Wolf, 2004, p. 3).
Underwriter (underwriting syndicate): The underwriting syndicate is comprised
of potentially a lead underwriter, managers or comanagers, and book-runners, although,
for the purpose of this study, unless noted, it is not important to distinguish who
underwrote the IPO, just that the company issued shares.
Assumptions
In this research project, I have attempted to refrain from imposing any undue
restrictions or assumptions on the data analyzed or tools used to analyze the data.
Although there are obvious underlying assumptions, I have addressed the majority of
these in order of importance. A standard assumption of the normative approach used to
analyze security’s price performance is that when analyzing concepts related to a
security’s risk and return the assumed relationship is linear. In addition, researchers
generally assume that a security’s returns are identically and independently distributed
and drawn from a normal distribution. This concept means that trading based upon
technical trading rules would provoke investors to expect the market to distribute their
returns equally through time.
The independence and normal distribution concepts are the basic assumptions
underlying the normative approach to securities’ analysis. These concepts also explain
the performance behavior an investor would likely expect if markets were efficient.
3
see: http://www.sec.gov/answers/quiet.htm
19
These assumptions allowed me to illustrate the differences of how the world actually
works and how it should work. I have imposed mechanisms to model normal
performance (i.e. match-firm preferred to the matched portfolio approach to measuring
abnormal returns) that have attempted to correct for departures from this neat
interpretation of the world.
I have also generalized the results of sensitivity analysis run on a randomly
sampled population to a nonrandom sample. I have concluded that canvassing the
population of IPOs has provided a more appropriate measure of abnormal performance
than could have obtained by conducting a random sample of this population. I decided to
canvass this population because the data is difficult to obtain, and one of the goals of this
research project was to increase the size of samples used to test for abnormal
performance. Therefore, I have assumed that if a testing procedure produces test statistics
that are well specified and powerful in random samples they will behave similarly in
nonrandom samples.
Limitations
I have restricted these analyses to a limited set of data, including only those
companies that issued unseasoned shares from 1985 to 2002 on the U.S. financial
exchanges. Because I have chosen to focus solely on analyzing unseasoned equity
offerings, I have not attempted to generalize the conclusions reached in this analysis to all
public offerings--specifically seasoned equity offerings (SEOs). Researchers have
provided evidence that SEOs underperform standard benchmarks as much as unseasoned
issuances (see Jegadeesh, 2000; Spiess & Afflect-Graves, 1995). I have decided not to
20
broaden the scope of this analysis to include SEOs. However, I believe by combining the
outcomes of studies attempting to identify abnormal performances generated by SEOs
and IPOs researchers may find better mechanisms to analyze longer-term performance,
and find greater linkages between the anomalies then currently exist.
When analyzing the results and significance of conclusion reached in conducting
event studies, researchers should consider the sample size as either a limitation or
strength. It would be optimal if the researcher analyzed the performance of all
unseasoned IPOs listed on the U. S. financial markets historically. A study of this
magnitude would take an extraordinary amount of time to execute, but the rationale
behind not executing this strategy is access to data. Even access to bulk data on IPOs
prior to 1995, is costly and difficult to obtain. Therefore, I have chosen to use the
available data that is readily attainable in this analysis.
Significance of the Study
This study is significant to the academic and professional communities because
the results will provide insight as to why researchers have labeled IPO performance as an
anomaly. Studies that have preceded the current research project used different samples,
event horizons, and methods to examine the abnormal performance of IPOs. This
fragmentation has generated incredible disagreement among the authors and theorists.
Studies that attempt to explain these anomalies need to integrate the tests of methods and
procedures used to identify abnormal performance with actual tests on IPO performance
to help interested parties to access the accuracy of conclusions reached.

21
The intent of this study is to reintegrate tests of methods and tests of abnormal
performances related to unseasoned IPO issuance, while using a significant sample of
existing data (1985-2002). The concepts that will be tested are that (a) IPOs produce
abnormally positive performance prior to trading publicly, (b) in the longer-term they
produce abnormally negative performance, (c) IPOs produce abnormally positive
(negative) performance at the expiration of their quiet (lockup) periods and perform
abnormally over 1-, 2-, and 3-year time horizons.
Social Change
Abnormal IPO performance affect investors and management or firm affiliates in
different ways. In this paragraph, I will briefly outline the social problems abnormal IPO
performance may generate. The first problem is that the process of IPO issuance does not
allow investors equal opportunity to economic outcomes. Specifically, the underwriting
syndicate, whose purpose in the issuance process is to allocate shares of the offering,
selectively allocates shares of the issue to privileged clients, without allowing the market
to bid for the offering. Second, the actions taken by the underwriting syndicate stifle the
efficiency of the issuance process in the following ways: (a) selective allocation of
shares, (b) determining a price without reference to market forces, and (c) I have
compared the ensuing positive abnormal performances to fixing a price ceiling on an
asset or commodity. I have developed these two social change concepts in chapter 5 and
they have serious implications for how investors think about market efficiency.
Revision in the IPO premarket pricing and allocation will eliminate IPOs’
abnormal performances enjoyed by a few investors and provides equal opportunities for
22
all investors to benefit from the IPO process. This will lead to significant positive social
change in the distribution of wealth and income in the society.
Summary
Based on the assumptions embedded in the normative financial theory, the
financial markets are efficient and, therefore, at any time a security’s price reflects its
fundamental or intrinsic value. To value a security, analysts use various sources of
information; this information set provides the basis for valuation. If information, based
upon historical analyses of securities’ price performances, provides evidence that the
market will react to a preplanned event in a specific manner, causing the price of a
security to trend in one direction or another, the market should integrate this information
into the price of a security. Thus, the security’s price should reflect this information when
the information becomes available. Therefore, if investors are aware that significant
performance movements will occur in the future, they should adjust the current price by
discounting the future value of the firm. In terms of IPO pricing and performance, it does
not seem as though this adjustment is being made by analysts or investors.
The focus of this study is to evaluate the legitimacy of anomalies uncovered by
researchers in regards to IPO performance. In chapter 2, I have reviewed literature related
to: (a) the actual process of issuing shares, (b) anomalous events related to IPO
performance, (c) securities pricing, and (d) the methodology used to detect abnormal
performance. In chapter 3, I have formulated the research methodology for testing the
hypotheses presented in the proceeding chapter. In chapter 4, I have presented the results
of the study, and in chapter 5, I summarized and concluded the analysis.

CHAPTER 2:
LITERATURE REVIEW
In this chapter, I have examined the fundamental concepts embedded within a
model of abnormal performance and identified key events that occur because of the
unseasoned issuance of shares on the U.S. financial exchanges. I have accomplished this
in two stages, by examining the IPO process and events related to this process and
conducted research addressing methodological difficulties that arise when researchers
attempt to measure abnormal returns. In this section, I have initially examined events that
researcher have historically suggested to generate abnormal IPO performance and
described various models that have been used to measure abnormal performance.
Examinations of these two areas of research provide the theoretical basis from which I
have conducted the remainder of the analyses.
IPO Process
The process of issuing unseasoned securities is complex. First, there are conflicts
of interest between the organization’s managers, investors, and underwriters. Second,
there is a complex underlying structure (legal and informational), which needs to be
adhered to when the company issues its shares. The conflicts and interrelationships
between the aforementioned policies and parties and their influence on the price
formulation of IPOs are beyond the scope of this section of this paper.
The process of equity issuance begins with the construction of a letter of intent,
which is a document outlining the agreement pertaining to the process of equity issuance
between the issuer and the underwriting syndicate. This document outlines the
24
underwriters’ compensation schedule, provides a price range for the IPO, and
approximates the number of shares available for issue (see Dalton, Certo, & Daily, 2003).
The underwriters’ services are obtained for numerous reasons during the IPO process; the
initial reason is to draft a preliminary prospectus, which is also referred to as the Red
Herring or the S-1 Statement (per SEC forms; Dalton et al., 2003, p. 292). In addition, the
underwriters offer services related to share distribution, market stabilization, and market-
making activities for the securities offered.
Twenty days after the filing of the preliminary prospectus, this document becomes
the official prospectus; at this time, the SEC has another 20 days to review the official
prospectus (see Ellis, Michaely, & O’Hara, 1999). Once the prospectus is official, the
underwriting syndicate plays a more traditional role of the investment banker and
conducts a road show. In this road show, the underwriting syndicate attempts to solicit,
create, and evaluate the public’s interest in the shares of their offering; typically, this
process takes three to four weeks (see Ellis et al., 1999). Once the prospectus is effective
and approved, the SEC evokes another 20-day waiting period, which the SEC may
discretionarily waive (see Ellis et al., 1999).
After the SEC has approved the waiver or once the issuer’s 20-day waiting period
has elapsed, the underwriter and the issuer can begin to place their offering. The issuing
company and underwriters generally meet between the close of trade on the day prior to
the intended date of issuance before the company issues its shares to decide on the exact
date of issuance. At this time, the underwriter and the issuer reach a firm agreement about
25
all of the details of the offering (e.g. the price, share amount of shares offered, & the
price of each share).
From the filing of the initial registration to the offering and issuance, through a
period ranging from 25 to 40 days after the offering occurs, the officers and analysts
affiliated with the company are not to make any forward-looking statements or issue
guidance other than the guidance contained in the official prospectus (see Ritter, 1998).
In addition, the underwriter and the issuer normally agree on a lockup period, which
typically, lasts for 180 days after the company issues the security (see Wiggenhorn &
Madura, 2005). Both of these periods allow market forces to determine an appropriate
market value for the security without interference from market professionals or corporate
insiders.
Anomalous IPO Events
I have provided a synopsis of the events IPOs face in their initial year of
seasoning, in this section I shifted the attention to anomalies found within this period and
thereafter. I have examined the following in this study: (a) the performance IPOs obtain
from offering to their initial trade, and from their initial trade to close of trade on the first
day of trading, (b) longer-term performance, and (c) performance occurring at the
expiration of the quiet and lockup periods. Now I will turn to examining the empirical
conjectures pertaining to the actual IPO events.
Initially Abnormally Positive Performance
The most visible abnormality that currently exists in studies of IPO performance
is that IPOs tend to produce extremely abnormally positive performance results a short
26
duration after going public. This excess abnormal return occurs either in the preissuance
period or in the 1-day performance of the post-offering period (see Krigman, Shaw, &
Womack, 1999; Loughran & Ritter, 2004; McDonald & Fisher, 1972; Reily & Hatfield,
1969). Miller and Reilly (1987) found that the extent of this underperformance was
approximately 9.87% (p. 34), and Ibbotson, Sindelar, and Ritter (1994) reiterated this
sentiment by concluding that “first–day returns average 10-15%” (p. 66).
Cheng, Cheung, and Po (2004) found, while studying IPO price performance on
the Hong Kong financial market, that no trading profits were obtainable once IPOs begin
trading publicly (p. 853), this finding contrasts those reached in Miller and Reilly (1987),
an analysis of IPOs listed in the U.S. markets. Historically, researchers seem to have
assumed that IPOs obtained profits in the first trading day. Alternatively, perhaps, they
have ignored the negative social and process implications attached to an empirical finding
that the positive IPO performance is constrained to the pretrading period. If the abnormal
performance is constrained between the offer and issuance, then the distributions of
shares, and whom the shares are distributed, become a more fundamental question, in
regards to fairness of distribution of these shares. This question is relevant because the
underwriting syndicate holds an unfair informational advantage over the majority of the
investing public.
The pertinent questions are does the underwriting syndicate exploit the
informational advantage and is the initial abnormally positive performance a result of an
under pricing / rebating scheme (see Ritter & Welch, 2002)? However, the investing
public may create abnormal performance because they are acting irrationally when
27
attempting to value these IPO. This irrational analysis may occur because the investors
know about the historical pricing anomaly (short-term abnormally positive performance),
and in turn demand for new issues are exacerbated and unwarranted optimism, in the
post-issuance performance capability of these securities, increases, thus pushes the share
price away from its fundamental value in the aftermarket (see Garfinkle, Malkiel, &
Bontas, 2002).
Case in point, Purnanandam and Swaminathan (2004) found in a study of 2,000
IPOs that went public from 1980 to 1997 the median IPO was actually overvalued when
compared against firms matched by operating characteristics-- they approximate
overvaluation at approximately 14% to 50% (p. 812). Since the investment bankers, who
are underwriting these offerings, are considered market experts, it seems that the market
would consider a systematic underpricing a disadvantageous finding for the EMH. If
investors are responsible for this abnormal performance then researchers should seriously
question market efficiency because the markets are irrationally pricing individual
securities in a reaction to systematic events related to the process of issuing unseasoned
equity shares.
Long-term Negative Performance
Researchers have also provided evidence in support of the theory that IPOs suffer
from long-term price underperformance when measured against standard benchmarks
(see Affleck-Graves, Hedge, & Miller, 1996; Ibbotson, 1975; Loughran, & Ritter, 1995;
Ritter, 1991). Ritter (1989) found that, in his sample of IPOs issued from 1975-84, IPO’s
3-year holding period returns (HPR) underperformed portfolios matched based upon
28
market capitalization and industry characteristics by 27.39% (p. 4); Ibbotson, Sindelar,
and Ritter (1994) found similar results analyzing IPO data from 1970-1990. Ritter (1989)
and Ibbotson (1994) suggested that on average IPOs underperform standard benchmarks
from the end of the initial trading day to at least the firm’s five-year publicly traded
anniversary.
Recently, a debate about how researchers apply the event study methodology to
measure long-term abnormal performance has instigated questions about whether the IPO
process actually generates abnormal performance. The fight for the concept of market
efficiency is alive and well. Researchers attempt to dismiss anomalies found that run
counter to the conceptual framework of the EMH by conducting significance and power
analyses on the methods previously used to test for abnormal performances--thus,
criticizing the applicability of tests applying the event study methodology to measure
longer-term performance. Explicitly, Schultz (2003), illustrated how the number of IPOs
issued increases as the market peaks, he conjectures that the managers of private firms
use pseudo-timing to time their issuances.
This pseudo-timing phenomenon illustrates how IPOs that are initially priced in a
hot market, to become overpriced in the aftermarket. For example, the fair intrinsic value
of the offering of ABC Corp. is $6 million, but the market has inflated this value to $6.5
million because supply is constrained and the company does not allow investors to bid for
allocations of the company. Due to the limited supply, the excess demand for shares of
ABC pushes the price past its fundamental value; as the market recognizes their error
there is an increased probability that they will correct this error. According to Schultz
29
(2003) the probability of a decrease in stock prices following the issuance of unseasoned
equity shares at a market peak is significantly greater than 50%, and the probability of an
increase in the new issues stock price is significantly less than 50%. Therefore, because
the market bids up the value of the share in the short-term, the probability that the shares
will decrease in value increases because the market has pushed the shares away from
their fundamental value due to excess demand. In a complementary study, Brav, Geczy,
and Gompers (2000) analyzed long-term IPO performance and hypothesize that the
underperformance found in the long-term performance of these issues is not an anomaly.
Brav et al. (2000) reasoned that IPO underperformance is not an IPO phenomenon, but
part of a broader systematic movement based upon a firm’s size and book-to-market ratio
(p. 246) or idiosyncratic risks. These two theories are representative of research projects
have been focused on attempting to reconcile the potential anomalous behavior of longer-
term IPO performance with the theory of market efficiency.
The Quiet Period
The quiet period is the market’s terminology for SEC regulation #5180, enacted
in 1971, it states that companies are not to issue forecasts, or predictions related to
revenues, income, and earnings per share, or publish “opinions concerning values” (see
Bradley, Jordan, & Ritter, 2003, p. 5). The rationale behind the enactment of the quiet
period is that, according to Bradley, Jordan, Ritter, and Wolf (2004) it provides investors
with the necessary time to value the company without insider interference or influence.
The quiet period facilitates the investor’s search for a fair value of the underlying assets
owned by the company.

30
At the conclusion of the quiet period, the SEC allows investment firms to initiate
coverage of a security. The reason why this period is so interesting is that Bradley,
Jordan, and Ritter (2003) have found that from 1996-2000, for all IPOs issued, analysts
initiated coverage on 76% of firms, and of these 76%, analysts initiated coverage on 96%
of these issues as a strong buy or a buy (p. 33). This is not what I would expect;
structurally, I would prefer to see a distribution that, from a probabilistic standpoint,
firms rated would just as likely receive a positive rating as a negative rating. According to
Bradley et al. (2003) when analysts initiate coverage, immediately after the quiet period,
the IPOs affected by this event experienced a significantly positive abnormal return of
4.1% in a five-day window surrounding the event (p. 33). If analysts left the newly issued
IPOs uncovered at the conclusion of their quiet period, firms experienced an insignificant
abnormal return of 0.1% (see Bradley et al., 2003, p. 33). In 2004, Bradley, Jordan,
Ritter, and Wolf (2004) attempted to expand this study to include IPOs that went public
from January 2001 through July 2002; the impact of the expiration of the quiet period
during this time horizon was insignificant (p. 11). In this dissertation, I endeavored to
answer why the two research projects differ and if so analyze whether abnormal
performance is significant during the expiration of the quiet period.
Lockup Expiration
Researchers, in the past, have not built a solid case to declare that abnormal
performance occurs as the lockup period expires. However, Field, and Hanka (2001)
found that from 1988 to 1997, during the expiration period investors experienced a three-
day abnormally negative performance of 1.5% (p. 471). The results from Garfinkle,
31
Malkiel, and Bontas (2002) were in agreement with Field et al. (2001), although the
Garfinkle et al. (2002) found that negative performance experienced during the expiration
of the lockup period was to 4.47%. The two different percentages vary remarkably and
the methods that the researchers used to calculate abnormal returns are questionable. It is
my goal to add clarity and specificity to this potential anomaly.
Securities’ Pricing, Estimating Returns, and Efficient Markets
In contemporary finance literature, analyst estimate the value of securities based
upon two sets of information: (a) the price of the security at a historical reference point
and (b) information that may affect a firm’s valuation from that reference point until the
time that the investor is interested in valuing the firm. This information set contains a
combination of events or states of the world; events included in this information set could
be public, economic, firm specific, industry affiliation, and international variables. Using
algebraic notation, and the most basic assumptions, the market arrives at prices as
follows, according to Fama (1975), the “joint distribution for security prices at time t” (p.
167) are as follows:
f m ( p1,t + r ,..., p n,t + r Φ tm−1 ) , (1)

Φtm−1 : is the information that the market uses to determine a securities’ price at
time t.
p1,t +r ,..., pn,t +r : is the different possible prices for a given security at time period
t+r, where r>0. (Fama, 1976, p. 134)
This formula illustrates that the market evaluates information received prior to attempting
to value a security, and once investors’ analyze the pertinent information contained in the
information set they then determine the appropriate price using probabilistic assessments.
32
Because I am interested in the performance of securities, I need to change emphasis from
the pricing of securities, to the returns obtained by investing in securities.
Since I have focused on estimating, the ex post expected market return on IPOs it
is appropriate to comment on the general model of estimation before discussing error
terms, and the actual calculation of abnormal returns in regards to IPO performance. The
following is the general market model used to estimate expected return on a security (see
Campbell, Lo, & MacKinlay, 1997):
E ( Rˆ it ) = α i + β i R mt + ∈
ˆ it
(2)
Ε[∈it ] = 0
α i
: is the intercept
~ ~
cov( Rit , Rmt )
βi = ~ (Fama, 1976, p. 67)
σ 2 ( R mt )
R mt
: is the Return on the Market Proxy
~
Rit : is the return on security i in time period t
~
∈ it
: estimated error of forecast
By adjusting this equation, analysts can construct a model to evaluate abnormal returns
against normal returns. I define this relationship according to Fama (1976):
~ =R~
∈it it − (αi + βi Rmt ) (3)
Under normal conditions, analysts can use this model or a more complicated factor model
to estimate the normal expected return on a security. Although, without the historical data
required to use these general formulas I used a more complicated model to estimate these
parameters—this is where the matched-firm and portfolio-matching strategies come into
consideration.
33
Methodological debate
Event studies were made popular by Ball and Brown (1968) and Fama, Fisher,
Jensen, and Roll (1969); the intended purpose of these studies were to test market
efficiency. The event study methodology is the historically accepted method used when
attempting analyzing a IPOs performance from both short- and long-term event windows
(see Bradley, Jordan & Ritter, 2003; Ibbotson, 1975; Ritter, 1991). In the present research
project, I define an event study as an analysis of the reaction of a securities’ price or
performance to an event or information (e.g. dividend changes, IPOs, SEOs, M&A, etc.).
Researchers that have chosen to focus on analyzing the accuracy of methodological
processes involved in conducting event studies have been remarkably active within the
last decade. The reasons for the increasing skepticism in the results of studies that attempt
to measure abnormal returns are problems with the following: (a) the use of biased
benchmarks and (b) the aggregation of IPO performance across time.
Measuring Normal Performance
Because IPOs do not possess historical returns from which a firm’s idiosyncratic
risk may be measured, researchers need to create proxies to estimate their expected
performance. Campbell, Lo, and MacKinlay (1997) provided a possible solution, to use a
market-adjusted-return model, which takes the formula in Equation #2 and sets the
intercept, α to 0, and, the beta coefficient, β to 1. However, by setting β = 1 ,
researchers would be expecting securities to produce a return equivalent to the market’s
return, and this implies that the risk-free rate of return is 0%. These two assumptions will
cause the error term of this equation to capture firm-specific risk, and the percentage rate
34
of return that is obtainable investing in a risk-free security, in addition to abnormal
performance. Researchers choose to avoid this method of measuring abnormal
performance because it is ineffective and inefficient. In the remainder of this section, I
have provided an evaluation of methods and biases generated by researchers using
different approaches to model normal performance.
Benchmark Construction
Researchers have shown that models used to measure the long-term performance
of stock returns are prone to misspecification (Lyon, Barber, & Tsai, 1999, p .165). The
cause of misspecification differs depending on the methods researchers use to model
normal performance. Researchers can approximate how poor a model of normal
performance is by taking samples of historic performance data and conducting
specification and power analyses on the model to test whether the model accurately
identifies abnormal performance in simulations. Studies related to methodological test of
this sort have made remarkable strides lately. Initially, researchers compared IPO returns
to standard benchmarks (e.g. Russell 3000 Index, or S&P 500), but this was ineffective
when they attempted to analyze IPO performance because IPOs lack the historical
performance data that would be used to measure the strength of the relationship between
a firm and the matching index. To create an accurate benchmark, without using historical
data, researchers have constructed portfolios or matched the event firm to a non-event
firm. The portfolio- and firm-matching approaches are more accurate than simply
matching the event firm to a standard market index, and reducing the beta coefficient to
one and the intercept to zero.

35
Benchmark Construction
There are many different benchmarks researchers use to attempt to simulate
normal performance. The basic elements that differentiate these benchmarks are: (a)
whether researchers construct the benchmark using a portfolio or a matched-firm
approach, (b) what factors researchers use to model normal performance, and (c) how
researchers weight securities in portfolios (market- or value-weight). The popular
benchmarks used in recent research have been: (a) traditional indices, (b) the Fama and
French Three-Factor Model (this method is used when researchers choose to use the CAR
method instead of the BHAR method to detect abnormal performance), (c) reference
portfolios, and (d) matched-firm approaches. In this research project it was unnecessary
make complex adjustments to portfolio-matching techniques because the firm-match
approach was well specified and powerful. Therefore, I will only explain the portfolio-
and firm-matching strategies used in this research project and the biases that affect the
accuracy of these benchmarking techniques.
Skewness bias. The skewness bias is prevalent in each of these matching
techniques; however, this bias has a more pronounce influence on the specification and
power analyses conducted on the matched-portfolio approach to benchmarking.
According to Barber and Lyon (1997) when reviewing a population of longer-term
securities’ returns researchers will find more returns above 100% than they would find in
excess of -100%. This occurs because, when an investor holds a long position in a
security, it is impossible to lose more than 100% of the investment. When using standard
tests researchers expect as many observations to fall beneath the -100% threshold, as they
36
would expect to observe over the 100% threshold. The existence of extreme positive
observations, in excess of 100%, will increase a sample’s standard deviation; these
extreme events are then overrepresented if we assume the underlying distribution is
normal (see Barber et al., 1997, p. 347). Barber et al. (1997) determined that in
simulations with observations drawn from a distribution with a known distributional
mean, in 6.6% of samples the researcher would conclude that the population mean is
significantly less than its known mean because of this skewness bias (p. 348).
Matched-portfolios. The matched-portfolio approach used to act as a benchmark
for normal performance in event studies is the most popular. It is this popular because
once the researcher determines how the portfolios should be matched there are, typically,
few portfolios to construct. In contrast, when using the matching firm technique
researchers have to match each event firm to a matched firm; for portfolios, researchers
would have an index based upon deciles of market capitalization or industry affiliation.
The matching portfolio approaches to benchmarking are traditionally prone to
misspecification due to the following benchmark biases: (a) new-listing, (b) rebalancing,
(c) survivorship and (d) skewness biases. The new-listing bias occurs when recently
issued securities are included in the portfolios used to gauge normal performance.
Empirical studies have shown that new issues produce positive (negative) abnormal
performances at different times during their maturation process, which leads researchers
to either underestimate (overestimate) abnormal returns experienced by the event firm
(see Lyon, Barber, & Tsai, 1999, p. 169). The rebalancing bias inflates the returns of the
matching portfolio, therefore, creating a negative impact on the event firm’s abnormal
37
performance (see Lyon et. al., 1999). The survivorship bias may cause researchers to
underestimate (overestimate) abnormal performance experienced by an event firm if the
researcher determines that the event firm outperforms (underperforms) the matching
portfolio.
Matched-firm. Barber and Lyon (1997), Lyon, Barber, and Tsai (1999), and Ang
and Zhang (2004), examined whether the 3-Factor, 4-Factor, carefully constructed
reference portfolios, and matched-firm approach are well specified and powerful methods
used to identify abnormal performance. The results of the aforementioned researchers’
statistical tests indicate that the matched-firm approach is better specified and more
powerful than the competing methods.
Currently, researchers are using some combination of a firm’s market
capitalization, book-to-market ratio, industry affiliation, or correlation to detect abnormal
performance using the matched firm approach (Ang & Zhang, 2004; Barber & Lyon,
1997; Lyon, Barber, & Tsai). To implement the matched-firm approach I needed to
consider how to identify the most applicable benchmark. When researchers use different
methods to identify their benchmarks, the results of their studies may differ vastly. For
example, if market capitalization and industry affiliation are the variables used to create
the benchmark then researchers need to identify, which filter (size or industry) to run first
and the boundaries placed around the filter. The results of analyses will vary depending
on which firm attribute researchers used first to sort the matched firms—market
capitalization or industry affiliation.

38
Historically researchers have successfully employed filtering techniques to the
preceding variables to measure the abnormal performance of a security with one
exception--matching by industry. Matching by industry is problematic because
researchers typically accompany this method by initially sorting the event-firms based on
other characteristics--namely, market capitalization. Ritter (1991), founds that for IPOs
listed from 1975-1984 only 36% of IPOs could be matched based their three-digit
Standard Industrial Classification Code (SIC), 21% were matched based upon their two-
digit SIC, and the remainder of the sample had to be matched to complementary or like
industries at Ritter’s discretion.
Loughran and Ritter (1995) explained why researchers would choose not to match
firms based upon the industry criteria. Loughran et al. (1995) reasoned that if firms go
public to take advantage of an industry-wide misvaluation, then matching firms based on
industry affiliation hinders the metric’s ability to uncover abnormal performance (p. 27).
Researchers have shown that this objection is generally faulty, and that the waves of
public offerings are not industry-specific. Normally, there exists a broader underlying
market trend. In addition, Loughran et al. (1995) felt that there are too few publicly
traded firms occupying the same industry to incorporate further filtering mechanisms. For
example, researchers may have problems attempting to filter first based upon market
capitalization and then industry affiliation, which is a reasonable pairing of variables.
Researchers need to conduct further research on how to better account for the
industry effect. Spiess and Affleck-Graves (1995) found that the industry-effect accounts
for one-third of long-term underperformance generated by SEOs. Since SEOs generate

39
abnormal performances comparable to IPOs I will use matching firm and portfolio
techniques based upon industry affiliation in the ensuing specification and power
analyses. Researchers that attempt to identify longer-term abnormal performance have to
overcome many obstacles when using the matched-firm approach sorted by industry
affiliation, but as Spiess et al. (1995) concluded, the trouble involved in overcoming this
obstacle may be worth the effort.
It is important to note that the majority of specification tests using the control firm
matching technique have focused on using the market capitalization and book-to-market
ratio firm attributes to estimate normal performances by sorting. Barber and Lyon (1997),
Lyon, Barber, and Tsai (1999), and Ang and Zhang (2004) all used market capitalization
and market-to-book ratios to detect abnormal performance, while Ang et al. (2004)
additionally examined using an event firm’s correlation with a non-event firm. Again,
researchers need to conduct further specification and power tests on different matching
techniques to evaluate whether additional factors lead to statistically better specified and
higher power tests of abnormal performance in random and non-random testing
situations. These additional factors include, according to Lyon, Barber, and Tsai (1999)
return performance, earnings surprises, and a company’s price-to-earnings ratio.
One criticism of the matched-firm approach was raised by Brav, Geczy, and
Gompers (2000) they reasoned, if researchers match firms that have not had an
unseasoned issuance (see Loughran & Ritter, 1995) within the five-year period preceding
the event, studies using data pre-dating 1978 may include long-term loser stocks in their
matched-firms. I define this long-term loser hypothesis as follows: long-term winners

40
will underperform long-term loser by 25% in the 3-period following the date of
calculation using securities’ performance data from 1926 to 1982 (De Bondt & Thaler, p.
804). The criticism has many assumptions supporting it; two important ones that make
this criticism less relevant to the current study are: (a) the matched-firms have to be long-
term losers and (b) I am not analyzing securities’ performances that pre-date 1980.
To summarize, Barber and Lyon (1997) concluded that researcher minimize the
new listing, rebalancing, and skewness biases by using the BHAR technique and the
matched-firm approach to benchmarking. The matched-firm approach corrects for these
biases in the following ways (a) researchers can eliminate new issues from the control
group, (b) there is no rebalancing, and (c) since the researcher uses a single-firm
benchmark, it is as likely to have a skewed distribution as the event firm is (Barber et al.,
1997, p. 354). Simulation techniques used to measure the independent variables ability to
identify abnormal performance have shown that this seemingly simple method is,
statistically, the best-specified and most powerful method used to measure abnormal
performance. Interestingly enough, researchers have applied the event study methodology
measure abnormal performance actively since approximately 1969, and the use of the
matched-firm approach seems to be in its infancy. The essential question left for me to
answer is, which variables best estimate normal performance, and are most accurate.
Measuring Abnormal Performance
After deciding on a benchmark to simulate normal performance, researchers have
to consider how to aggregate returns to determine if an event firm has experienced an
abnormal event becomes critical. Currently, there exists substantial debate on how
41
researchers should approach measuring this abnormal performance. Is the CAR or the
BHAR method the best-suited model to use to identify abnormal performance? In this
section, I have explained the differences between the BHAR and CAR methods of
aggregation and why I chose to use the BHAR method in my analyses of IPO
performance.
BHAR vs. CAR
When researchers attempt to aggregate abnormal returns they generally use one of
two popular methods, the BHAR or CAR technique. Both of these methods have their
advantages and disadvantages. I can illustrate the differences between these two methods
effectively by citing specification analyses. In one specification analysis, Barber and
Lyon (1997) found that when comparing the results of the CARs and BHARs researchers
arrive at conflicting conclusions in 3.7% of their analyses, and if samples consisted of
primarily small firms this rate increases to 4.7% (p. 368).
To aggregate and analyze abnormal returns using the CAR method, according to
Barber and Lyon (1997), researchers use the following calculation:
AR it = R it − E ( R it ) , where (4)
τ
CARiτ = ∑ ARit , and (5)
t =1
Rit: is the return on the on asset i in period t

E(Rit): is the estimate return on asset i in period t
ARit: is the abnormal return produced by security i in period t
42
To understand the potential differences of the two aggregation techniques I will now
introduce the BHAR calculation (Barber et al., 1997, p. 344):
τ τ
BHARit = ∏[1 + Rit ] − ∏[1 + E( Rit )] (6)
t =1 t =1
According to Sharifzadeh (personal communication; December 30, 2006), the calculation
of the BHAR can be thought of as the return obtained by a day trader that moves in and
out of a security at the opening of and close of the trading day every day. Analysts use
the geometric method of aggregation to calculate the BHAR, where as researchers use the
arithmetic method to calculate the CAR. The difference is that the BHAR calculation is a
more precise measure of security’s returns.
Sharifzadeh (personal communication; December 30, 2006), also stated that for
any holding period the average periodic return calculated using the geometric method is
smaller than the average periodic return calculated using the arithmetic method. For
example, Schaeffer (2005) claimed “assuming market returns are flat, a 50% loss in one
month followed by a 100% return the following month results in a CAR of 25%, despite
the fact that the stock is now trading at its initial price” (p. 5).
As Barber and Lyon (1997)) illustrate, the BHAR more accurately reflects an
investor’s investment experience, because the returns are compounded. Mitchell and
Stafford (2000) contend that this procedure is faulty, because not all investors are
interested in comparing their returns against investors who bases their strategy for
investing on a buy-and-hold investment strategy (p. 296). This argument is

43
understandable, although in this research project I am only concerned with the
performance of IPOs measured against the buy and hold investor’s return experience.
Therefore, henceforth I will assume the researchers can obtain the best measure of an
investor’s experience of investment performance using BHAR methodology.
Barber and Lyon (1997) mentioned that CARs are just biased predictors of
BHARs. Barber et al. (1997) illustrate this finding using a 12-month CAR and running a
regression against the 12-month BHAR calculation for a random sample of 200,000
observations. Barber et al. (1997) provided regression results that produce an intercept
and slope coefficients of -.013 and 1.041, with an adjusted R-Squared value of 77.6% (p.
346). According to Barber et al. (1997) if the CAR methodology were unbiased the slope
and intercept coefficients would not be significantly different from 0 and 1, respectively.
Additionally, according to Sharifzadeh, M. (personal communication, December 30,
2006), The Institute of Charted Financial Analyst (CFA institute) recommends that
investment professionals use the geometric (BHAR) methodology to calculate returns, in
accordance with this recommendation the SEC has issued guidance to mutual funds and
portfolio managers advocating the use of the geometric approach to calculate returns
when reporting their performance.
Summary
In this section, I have examined the issuance processes of IPOs, defined event
horizon for events that may produce abnormal performance, provided a structure for
analyzing IPO performance, and methodologies used to measure abnormal performance.
Possibly the most important decision I have reached in this section is to use the BHAR
44
method over the CAR. I based my decision primarily on the theory that CARs are biased
predictors of BHARs. I supported this decision by reasoning that BHARs measure
investor returns, which more accurately reflects an investor’s buy and hold performance.
CHAPTER 3:
RESEARCH METHOD
Introduction
In this section, I have focused upon elements of the research methodology that:
(a) are used to conduct event studies, (b) test the hypotheses contained in the IPO
literature, and (c) present a methodological approach for evaluating abnormal
performances related to IPO issuance. In the beginning of this chapter, I will explain how
I will evaluate the hypotheses tested in my dissertation using the event study design.
Next, I will describe the sample size and frame used in this study. In addition, I
developed the rationale for canvassing a broad population. Then, I will introduce the
reader to the database from which I obtained the data needed for testing of the majority of
the hypotheses. Finally, I fully developed all of the hypotheses tested in this dissertation.
At the conclusion of this chapter, I will have provided the reader with a comprehensive
review of the elements of the research method, and defend the use of the selected
research method.
Research Design and Methodology
Justification for Using the Event Study Methodology
I have chosen to use the event study methodology in the current research project. I
decided on this method because in this dissertation I have analyzed the change in the
performance of securities’ price formation as systematic events occurrence. I described
the event study design as a subset of the existing data research methodologies.
Historically, researchers have applied the event study methodology to measure

46
performance reactions to a significant number of corporate events (e.g. stock splits,
earnings releases, mergers and acquisitions, dividend changes, etc). Campbell, Lo, and
MacKinlay (1997) stated that Dolley presented the first published event study in financial
literature in 1933. Dolley studied the effects of stock splits on an organization’s common
stock. Two papers, in particular, have provided a more recent rationale for the wide
acceptance of this method. Ball and Brown (1968) used this method to analyze how a
change in income affects the performance of a security, and how quickly the market
adjusts to this change. Similarly, Fama, Fisher, Jensen, and Roll (1969) used this method
to document how the securities’ markets adjust to securities’ pricing splits and a
security’s dividend history. The preceding pioneers of this method created the event
study methodology to measure the impact of information on the performance of a
security.
The reasons for choosing the event study approach to examine abnormal
performance changes in response to IPO issuance are as follows. First, I chose the event
study methodology to examine some specific events that had already occurred, that is,
changes in the performance of a security presumably resulting from information
pertaining to a specific event or occurrence. Second, judging by the number of research
projects that have applied the methodology, the event study methodology is the preferred
method employed by researchers attempting to examine questions related to market
efficiency. Moreover, because I am questioning whether the markets efficiently price and
adjust for information embedded in the issuance process, it is appropriate to use the event
study methodology to conduct my analysis of market efficiency with regard to IPO

47
performance. Third, the event study methodology has been around for more than 75
years, which would suggest that the method is not in its infancy and is a reasonable tool
of measurement. Finally, as I have argued in the next three paragraphs, the event study is
the most suitable methodology for my research as compared to survey methodologies
(longitudinal, cross-sectional, and time series), correlational or casual comparative
methodology, case study approach, and experimental designs methodology.
Survey methods seemed to be a reasonable approach when I began thinking about
how to analyze potential abnormal performance occurring throughout the IPO maturation
process. However, when I drilled down into the details of how researchers should apply
each method it became apparent that the event study framework would be better suited to
examine abnormal performance occurring throughout the IPO process. For cross-
sectional approaches researchers look at occurrences at one point in time, in this research
project, I used historical data—I took observations over a substantial time horizon. If I
choose to use the longitudinal approach, I would have to use the same sample throughout
the entire study. Because my sample spanned a 22-year period, during which new
companies went public every year, and given that IPO performance has the potential to
change over time—the longitudinal approach would not be suitable for this type of
analysis.
The event study approach is also more suitable than the causal comparative
methodology for this research. According to Rumrill et al. (2004), when researchers use
the causal comparative research method, they are interested in comparing, “differences
between derived or intact groups on theory-driven dependent measures” (p. 258).

48
Rumrill et al. further argues that causal comparative approach is suitable for the study of
naturally occurring groupings, such as, race, gender, and socioeconomic strata (p. 258).
However, my research is about measuring the abnormal performances of IPOs resulting
from information transmitted into the market and as such, the event study approach is a
more suitable approach than the causal comparative approach
I also considered other research methodologies before deciding on the event study
methodology, they were: (a) case study, (b) correlational, and (c) experimental. After
reviewing the description of the case study approach found in Leedy and Ormrod (2005),
I felt that one of the overarching focuses of the event study is to pay particular attention
to the contextual clues that influence the events occurrence. In my opinion, by using a
case study, I would be focusing more on describing why IPOs are performing abnormally
and less emphasis on the question of whether IPOs produce abnormal performance, and it
is the latter question I was interested in answering. In correlational studies, researchers
are interested in describing a relationship between two or more variables, which is the
focus of this research project; however, I am interested in describing how different the
performances are or how much the return on IPOs differ from expectations. The
identification of a correlation between two of my variables, would not enable me to state
that there is a potential anomaly, only that I have found a correlation between these
issues. Therefore, the event study methodology is preferred to correlational methodology.
Finally, when researchers conduct experiments, they need to randomly assign participants
into the control and experiment groups and I would need to manipulate some factors in
one group and not in the other. Because I am using historical data and cannot manipulate
49
events that have already occurred, experimental design methodology is not suitable for
this study.
Approach to Implementing the Event Study Methodology
When conducting event studies researchers have to focus on numerous elements
of the methodology. According to Campbell, Lo, and MacKinley (1997) the outline of an
event study is as follows: (a) event definition, (b) selection criteria, (c) identifying normal
and abnormal performance, (d) estimation procedures, (e) testing procedures, (f)
empirical results, and (g) interpretations and conclusions (pp. 151-2). In the hypothesis
subsection of the data analysis section of this chapter, I have addressed steps a-e, in
addition, I have evaluated steps f and g in subsequent chapters. Because IPOs have no
pre-event returns, I am unable to estimate the IPO’s correlation with the benchmark.
Therefore, I replaced my estimation period with a simulation and sensitivity analysis (see
Hypothesis 1).
For the testing procedure, I used the BHAR in favor of the CAR method to
identify abnormal performance. To estimate the return on the benchmark, I have,
executed a specification and power analysis and decided what combination of factors
produce the best-specified estimates of abnormal performance. With the proceeding
elements of my design of this event study specified, I will turn to a discussion of the
setting and sample from which I will make inferences about abnormal performance.
Setting and Sample
I could not obtain data throughout the entire sample to analyze IPO performance
occurring prior to public trade and on the first day of public trading. My initial sample
50
was comprised of all securities traded on major U.S. exchanges that went public from
1985 to 2002. I used this sample for the analyses conducted on: (a) the specification and
power analyses, (b) the longer-term performance of IPOs, and (c) the analyses conducted
at the conclusions of their quiet and lockup periods. However, data from the CRSP
database, used for the aforementioned analyses are not readily available for the two short-
term analyses. The sample shrank from approximately 5583 observations for longer-term
analyses, to 5529 observations used to evaluate performance around the expiration of the
quiet and lockup periods, to 2143 observations for IPOs first day of trade performance,
and to 1876 observations for the analysis conducted on pre-trade performances. In the
analyses associated with the long-term, quiet, and lockup analyses I used all available
data on IPOs that went public from 1985-2002. This time horizon shrank considerable to
January 1, 1997 to December 22, 2005 for tests conducted on the initial day of public
trading and to April 12, 1996 to January 28, 2008 for tests conducted on pre-trade
performance. I am accepting this shrinkage in sample size because I can support the
results of the tests whose sample size shrunk by complimentary findings, for the results
that are somewhat controversial, the sample size remained large.
Canvassing a Significant Target Population
The execution of my research design included canvassing the entire population of
unseasoned equity securities issued from 1985-2002. I based my rationale for canvassing
the entire population on the fragmentation of research projects in which researchers
analyzed different events related to IPO performance. This concept is critically important
to the significance of this research project--I have tested the hypotheses through both
51
time and used a significant number observations. To illuminate the problem, Bradley,
Jordan, and Ritter (2003) found that from 1996 to 2000 IPOs experienced a 3.1% positive
abnormal performance event in the 5-day window of time surrounding the expiration of
the quiet period. Bradley et al. (2003) quickly reverse their position; Bradley, Jordan,
Ritter, and Wolf (2004) found that between January of 2001 and July of 2002 the
abnormality disappears. The second research project examines just 37.5% of the
observations analyzed in the original study. In addition, the period analyzed in the second
study was only 1.5 years, the original period was 4 years, I believe that researchers need
to expand both of these studies and that the finding presented in the second study may be
a result of a small sample bias.
Data Collection
The data collection process also may prove to bias the research project. In this
research project, since I am interested in evaluating existing data, I have pulled and
pooled data from a variety of sources to produce the most accurate reflection of the
population as possible. The potential sources that I have used to obtain data are the
Center for Research on Securities Prices (CSRP) database, Standard & Poor’s Compustat
database, Thomson Financial, The Wall Street Journal securities’ pricing databases,
Google Finance, Hoovers IPO database, the IPO Reporter, Edgar Online IPO, and various
governmental and financial web sites. The first goal of this research project was to
identify all IPOs that have gone public in the 18-year period from 1985-2002.
For the 1985 to 1996 period, I will use the Field-Ritter dataset of company
founding dates, as used in Loughran and Ritter (2004; as noted on

52
http://bear.cba.ufl.edu/ritter/foundingdates.htm). This data set provides me with a
substantial listing of IPOs, but as Ritter notes on his website, it is not a comprehensive
IPO list, I excluded those IPOs that do not have a reliable founding dates from the
analyses. I used the firm name to query the CSRP/COMPUSTAT merged database to
find historical pricing data for IPOs that issued shares from 1985 to 2002. If after
searching through the preceding resources for pricing and other data, if I do not obtain the
data using the identified resources I have decided to drop the IPOs from the analysis.
Data Analysis
The Variables
RIPOAverage(Offering→InitialTrade)
: The average return on unseasoned IPOs in the pre-public
trading period
RDJIAMonthly
: The average monthly return for the Dow Jones Composite Index
RIPOMonthly (Offering →InitialTrade ) : Pre-trade returns aggregated in month cohorts.
R IPO ( Day 1) : The return obtained by IPOs in their initial day of trade.
R IndustyMat chedFirm : The return obtained by the matched firm benchmark, utilizing industry
affiliation as the matching criteria.
BHAR ( TradingDay 2 − 750 ) : The average BHAR calculated from trading day 2 until trading day
750.
R Quiet : The average return obtained by an IPO during the 5-day period surrounding the
expiration of the quiet period.

53
RLock−up : The average return obtained by an IPO during the 5-day period surrounding the
expiration of the lock-up period.
The Hypotheses of Research
In the ensuing section, I have outlined the hypotheses of this research project.
There are four sections, which are as follows: (a) specification and power analysis, (b)
short-term performance, (c) long-term performance, and (d) events related to IPO
issuance. The results of some of these sections influence other analyses. Specifically, the
selection of a well-specified model in section 1 has influence over each of the subsequent
analyses. Therefore, the reader should evaluate each of the hypotheses individually and
then collectively.
Section 1: Specification and Power Analysis
Researchers have recently begun to debate the validity of tests that identify
abnormal performance in long-term event studies. Barber and Lyon, (1997), Lyon,
Barber, and Tsai (1999), and Purnanandam, and Swaminathan, (2004), derived
remarkably different conclusions from their analyses. After a thorough review of the
three preceding documents, it is apparent that any application of a method used to detect
abnormal performance needs to be accompanied by a specification and power analysis.
Therefore, I ran a specification analysis and simulated abnormal performance using the
Brown and Warner (1985) methodology, which calls for 250 samples of 50 simulated
events (p. 6). I drew these samples using the stratified sampling technique to ensure I
gave each year of performance an equal weighting. I used companies comprising the
54
Russell 3000 Index, in each year of the sample period, as the population securities
available for matching.
I took ten random samples of 50 securities each year. For the specification
analysis I have taken this data and calculate the average BHAR, standard deviation, and
the empirical size (ES) statistic of the of specification analysis. Also using this sample, I
have imposed abnormal performance, and analyzed whether this abnormal performance
is identified given the model of normal performance, whether theoretical rejection levels
are adhered to (power analysis), and whether the model fails to identified this
performance over from trading day 2 until trading day 750 (approximately 3 years).
The specific methodologies used to test for abnormal performance that were
analyzed in this analysis are the reference portfolio and the matched-firm approach,
which were matched based upon the market capitalization, industry affiliation, and book-
to-market ratios. When using the matched-firm approach sorting by market capitalization
and book-to-market ratios I first sorted firms based upon market capitalization,
identifying the closest 25 firms when compared against the event firm, and then identified
a firm with the closet book-to-market ratio. To sort using the firm’s industry affiliation
and market capitalization, I first sorted by industry affiliation and then by market
capitalization. Finally, I sorted the matched-firm approaches using just the market
capitalization or industry characteristic based on the given characteristics. After this sort,
using the market capitalization approach, I scanned the sample to find the closet match or
using the industry approach I grouped the sample by industry affiliation and then
conducted a random sample to obtain a match. If I could not find a 4-digit SIC Code
55
industry match I moved to a 3-digit match, and if I could not identify a 3-digit match, I
removed the firm from the subsequent analysis. Similarly, if no book-to-market match
could be identified the firm was removed for the analysis. If data from a matching
company is missing, I replaced the missing data with a new matched-firm.
For Hypothesis 1, I have taken data from 1985 to 2002 and conducted 10 random
nonrepeating samples of 50 companies—overall, I distributed the 180 samples evenly
across the 18-year period. For each benchmark and each sample, I matched the
benchmark return against a randomly sampled company or the matching portfolio and
then the BHAR was calculated.
Hypothesis 1:
H0: BHAR = 0
H1: BHAR ≠ 0
After running the preceding test on the 180 samples of 50 companies, I calculated the ES
statistic. I then compared this statistic against a confidence interval to determine the
specification of each method. I combined the results of this analysis were combined with
the ensuing power analyses to select a benchmarking method.
In the power analysis, I took the same sample that was mentioned in the
specification analysis and simulated abnormal performance at different intervals to
determine when each of the seven methods of abnormal performance detection reacted to
simulated abnormal performances of +/- 1, 5, 10, 15, 30, 50, and 75%.
56
Hypothesis 2:
H0: BHAR = 0
H1: BHAR ≠ 0
Next, I combined the results of the individual tests, and calculated the Empirical Power
(EP)—I calculated the EP by taking the number of times I rejected the null hypothesis
divided by the number of tests conducted. I then used the results of the specification and
power analyses to decide which benchmarking metric was the best specified.
Section 2: Initial and Subsequent Short-term Performance
In this round of tests I analyzed whether and when IPOs generate substantial
abnormally positive returns. The initial test of abnormal IPO performance is constrained
to the period prior to public trading, from the offering to the initial public trade. I
conjecture that, as expected in the efficient market theory, on average, speculators or
investors’ using systematic trading schemes cannot obtain abnormal profits after the issue
begins public trading. However, the results of analyses attempting to test this hypothesis
have presented evidence that investors can obtain abnormal performance either prior to
public trading or on the first day of public trade—most analyses do not differentiate
between the two trading periods. Therefore, in my hypothesis tests I have assumed after
issuance no abnormal performance occurs.
To test Hypothesis 3, I used data from various sources--Google Finance, Hoovers
IPO database, The IPO Reporter, Edgar Online IPO--to obtain the IPOs offering price,
the initial opening price on the first day of trade, and industry affiliation. To generate a
sample for this analysis, I used IPO data using the aforementioned search tools and I
57
compiled a canvassed sample of 1,876 IPOs, companies issued these shares from April
12, 1996 to January 29, 2008. First, I determined if the performances produced in this
period were significantly different from zero (Hypothesis 3), and then in the Hypothesis
4, I determined if the average monthly performance of the IPOs in the pre-public trading
period were significantly greater than the best performing benchmark.
Hypothesis 3:
H 0 : RIPOAverage(Offering→InitialTrade) ≤ 0
H1 : RIPOAverage(Offering→InitialTrade) > 0
To conduct the analysis found in the Hypothesis 4, I calculated the average
monthly return for the IPO group, and then assumed that an investor obtains an allocation
of one share of the issuing firm’s stock at its offering and sells it at the opening price on
the initial day of public trading. The individual performances were then average and
grouped by offering month. I grouped these firms so that I could compare the returns to a
standard benchmark. The conjectured relationship was as follows:
Hypothesis 4:
H 0 : R IPOMonthly (Offering →InitialTrade) ≤ R DJIAMonthly

H 1 : RIPOMonthly (Offering → InitialTrade) > RDJIAMonthly
In Hypothesis 5, I determined whether the abnormal performance that occurred
prior to the issuance of unseasoned equity offerings continued to occur when the shares
begin public trading. In this analysis, I have matched the returns of the IPOs against the
best-specified and most powerful method of benchmarking—identified in Hypothesis 1
and 2. The sample in this analysis was obtained using the list of IPOs found in the Field-
58
Ritter dataset of company founding dates previously mentioned, the sample was
canvassed and compared against the CRSP database to identify the opening price and the
closing price for the IPOs used in this sample, on the first day of trading data. The CRSP
database started tracking this data in 1997; therefore, tests were constrained the period
1997 to 2005 periods, which yielded 2143 observations.
Hypothesis 5:
H 0 : R IPO ( Day 1) ≤ R IndustyMat chedFirm

After Hypotheses 2 through 5were tested, I had determined whether initial abnormally
positive performance was generally constrained to the period preceding the issuance of
shares, and provided a detailed account of short-term IPO performance.
Section 3: Longer-term Performance
Given that the results of the hypothesized relationships in Hypotheses 3 and 4 are
correct, using either the offering price or the initial trade as the starting price would have
influenced the estimate of long-term abnormal IPO performance. Therefore, I decided not
to include the performance obtained in these two periods to analyze longer-term IPO
performance. The general questions I was interested in answering is whether IPOs
significantly underperform the chosen benchmarks, and when long-term
underperformance occurs. In order to answer the question took the list of IPOs found in
the Field-Ritter dataset focusing on the 1985-2002 period.
I then calculated the IPO’s and the selected benchmark’s compounded returns for
each holding period from trading day 2 through 750. I then subtracted the benchmark
59
returns from the IPO’s returns; the resulting value, which was the BHAR, I then averaged
the BHAR across the sample, grouped by trading day.
Hypothesis 6:
H 0 : BHAR ( TradingDay 1 − 750 ) = 0

H 1 : BHAR ( TradingDay 1 − 750 ) ≠ 0
I chose to focus on only three years of data because the power of the metrics tested in
Hypothesis 1 and 2 deteriorate significantly, as I increased the event horizon beyond
three years (see Appendices E-H).
Section 4: The Quiet Period & Lockup Expiration
In this round of the analysis, I was interested in determining whether IPOs exhibit
significant abnormal performance within the 5-day period surrounding the expiration of
the quiet and lockup periods. The quiet period is a period that occurs after an issuing
company submits their initial registration to the Securities and Exchange Commission.
This period last for 25 to 40 days after the company issues its shares to the public. In this
period, the SEC prohibits the company and the companies’ agents from issuing forward-
looking statements about the future prospects of their company. The lockup period
typically ends 180 days after the company goes public, but the underwriters can alter this
date. I used the five-day trading period surrounding the expiration of the quiet and lockup
periods as the measurement period--to determine whether abnormal performance
occurred at these expiration points.
The samples canvassed to conduct these analyses were the same as previously
mentioned in the analysis of long-term performance. However, if I identified problems

60
with any of the IPOs (e.g. the dates do not match as expected or a firm trades prior to its
issuance) they were excluded from this portion of the analysis. I calculated the
compounded holding period return (HPR) for the IPO and the benchmark, for the five-
day window surrounding the conclusion of the quiet and lockup periods. I then subtracted
the benchmark from the IPO’s HPR and averaged over the entire sample to determine the
results of the following tests.
Hypothesis 7:
H 0 : BHAR Quiet ≤ B H A R IndustyMat chedFirm

H 1 : BHAR Quiet > B H A R IndustyMat chedFirm
When I analyzed the performance around the expiration of lockup, the sample and
methodology were the same; however, instead of a predicted positive abnormal
performance as the quiet period expired, I predicted that there would be a negative
performance result as the lockup period expired. I forwarded this conjecture because as
the shares of the newly issued IPO approach the expiration of the lockup period, the
restriction of insider shares expires and underwriting syndicate allows the officers of the
company and other powerful shareholders to sell their shares on the open market. Prior to
this expiration, they were constrained from selling these shares. Because of the increase
in volume of securities available for public trade and the likelihood that powerful
shareholders will tend to diversify their holdings, it was my conjecture that the shares
will experience a significantly negative performance event.

61
Hypothesis 8:
H 0 : BHAR Lock − up ≥ B H A R IndustyMat chedFirm

H 1 : BHAR Lock − up < B H A R IndustyMat chedFirm
By forwarding, the preceding hypotheses I was interested in addressing whether the
market reacts to events related to IPO issuance, which occur systematically, and generate
abnormal performance. These tests, and the substantial period that I used to test whether
these events produce abnormal performance either counter or substantiate the claims
forwarded by researchers.
Summary
The proceeding section provided the fundamental relationships that I was
interested in evaluating in my dissertation, and how I endeavored to test them. Of
particular interest was how well the different benchmarks estimated normal performance,
and how well specified and powerful the tests that I have run to identify abnormal
performance were. Currently, researchers seem to be focusing on identifying and
examining different methodological applications used to conduct event studies; therefore,
the significance and the power tests are critically important to the academic interpretation
of the results of this study. At the conclusion of the proposed analysis, I have provided
the academic and professional communities with direct tests analyzing whether events
associated with an organization’s issuance generated abnormal performances. Given that
I have used daily data in this analysis, investors and other interested parties will be able
to visualize the performance of the average IPO from its issuance to its three-year
62
anniversary. Focusing on the entire three-year data set should enable investors to improve
their expectations in regards to long-term IPO performance.

CHAPTER 4:
RESULTS
Introduction
In this chapter of the dissertation, I have reported the results obtained from the
tests of eight hypotheses. In the first section, I have reported the results of the power and
specification analysis. I used the best performing benchmarking method, identified by
specification and power analyses, to test for abnormal IPO performance. In the next
section, I have analyzed whether unseasoned IPOs generate abnormal performances prior
to public issuance and on the first day of public trading. I followed this analysis by an
evaluation of longer-term IPO performance. Finally, I have evaluated whether IPOs
perform abnormally after the expiration of their quiet and lockup periods. I have
addressed each of these sections in their respectively order, in the following paragraphs.
Specification and Power
The purpose of this section was to determine, which method of benchmarking was
more effective testing for abnormal performance of the IPOs. Based upon the review of
literature, I employed two broad methodological strategies to conduct specification and
power analyses--the matched-portfolio and the matched-firm approaches. I performed the
portfolio-matching strategies using daily return data from Dr. Kenneth French’s web site4
based upon three characteristics: (a) market capitalization, (b) market capitalization and
book-to-market ratios, and (c) industry affiliation. In addition, I implemented the
matched-firm approach by taking the characteristics of the randomly selected firm and
4
http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/index.html
64
matching the firm with a similar firm from the pool of Russell 3000 Stock Index based
upon: (a) market capitalization, (b) market capitalization and book-to-market ratios, and
(c) industry affiliation.
I randomly drew 50 companies, without replacement, from the Russell 3000 list
of companies, each year from 1985 to 2002, and repeated this procedure 10 times each
year. I matched each company, from the draw of fifty companies, to portfolios and firms
based upon the aforementioned characteristics. I then obtained 4-years of performance
data from the CRSP database for the randomly selected firms, and their matched-firm
counterparts--relying on the previously mentioned portfolio data. I then calculated
BHARs for each of the benchmark for 1-, 2-, and 3-year time horizons. When attempting
to use a portfolio matching strategy to conduct event studies relying on the BHAR
method to calculate abnormal performance, they will often conclude that abnormal
performance is evident when no abnormal performance has occurred.
Test of the Hypotheses
In this section, I have provided the results from the tests of the hypotheses used to
detect abnormal performance. In the initial rounds of this analysis, I selected the method
used to detect abnormal performance based upon specification and power analyses. In the
subsequent hypotheses tested, I transitioned to testing whether IPOs perform abnormally
throughout their maturation process.
Section 1: Test for Specification
I conducted this analysis based upon the methodological approach documented by
Ang and Zhang (2004). The purpose of the specification analysis was to determine what
65
methodology has the least type I error. That is, to minimize the probability of concluding
that there had been an abnormal return when in reality abnormal performance had not
occurred. To test whether a method was well specified Ang et al. (2004) used the
empirical size (ES) statistic. In these tests, I conducted simulations on a sample of data
and calculated whether the ES is close to “the pre-specified significance level at which
the test is conducted” (see Ang et al., 2004, p. 261), if this occurs I can claim that the test
was well specified.
The hypothesis test for specification was as follows:
Hypothesis 1:
H0: BHAR is zero

H1: BHAR is not zero
To determine which benchmarking methodology led to a well-specified test statistics, I
randomly selected companies from the lists of the yearly Russell 3000 Index. I randomly
drew event firms from the pool of Russell 3000 companies each year and paired them
with matched firms, and matched portfolios based upon their market capitalization,
industry affiliation, and book-to-market ratios. The percentages showed the percent of
rejections or identifications of abnormal performances across the entire set of samples. In
each of the 18 years, I generated 10 samples of 50 observations, and evaluated these
samples using Barber and Lyon (1997) tBHAR test statistic. With this statistic, I calculated
whether abnormal performance was significant in each of these samples. If I identified
abnormal performance, rejected the null, this rejection was identified and given a value of
one--if not rejected, given a value of zero. I then summed the results and divided the sum
by 180, the number of samples. Therefore, the percentages found in Table 1 illustrate
66
how many times the researcher found statistically significant abnormal performance
when none was expected. The following table presents the results of specification
analyses:
Table 1
Specification Analyses I
Empirical Size - Specificaiton

Matched Firm Matched Portfolio
Years MCap Ind Ind -> MCap MCap -> BE/MC MCap MCap -> BE/MC Ind
1 5.00% 3.89% 4.44% 3.33% 43.89% 44.44% 46.67%
2 2.78% 1.67% 3.33% 2.78% 31.67% 25.56% 56.67%
3 2.22% 1.67% 2.22% 1.11% 33.33% 28.89% 65.56%
4 5.56% 3.89% 3.89% 2.78% 47.22% 36.67% 79.44%
At a 5% level of significance, the ES interval ranges from .01816 to .08184—I
calculated this confidence interval using the following formula:
.05(1 − .05)
.05 ± 1.96
180 (7)
All of the approaches using the matched-firm approach failed to reject the null, declaring
that no abnormal performance had occurred, which I had not anticipated, but this is
extremely positive result. However, each of the portfolio-matching strategies rejected the
null hypothesis; identifying abnormal performance even though, I have not simulated
abnormal performance. Given the results of the preceding analysis, it is obvious that the
matched-firm approaches produced better-specified results than the matched-portfolio
technique.
To attempt to differentiate between the different matched-firm approaches used to
identify abnormal performance, I tested for abnormal performance using an entire years’
67
sample--n = 500 firms. In Table 4, exhibits how different the ES statistic was when the
sample is increased from 50 to 500—the confidence interval, due to the increase in
observations and decrease in samples, increased in range, all samples that identified
abnormal performances between 0 and 15.06% was not significantly different from a 5%
level of significance.
Table 2
Specification Analysis II
Empirical Size - Specificaiton

n=500
Time Horizon (Yrs) MCap Ind (Rand) Ind -> MCap MCap -> BE/MC
1 0.00% 0.00% 0.00% 0.00%
2 0.00% 0.00% 0.00% 5.56%
3 11.11% 0.00% 0.00% 5.56%
4 0.00% 0.00% 5.56% 0.00%
As the sample size increases from 50 to 500, it is obvious that the observed percentage of
rejections move closer to theoretical or expected rejection levels. This occurs because,
without simulating abnormal performance, researchers would expect to detect no
abnormal performance. As the sample size increases, the ES statistic approached zero, to
verify this conclusion compare the ES statistics found in Table 1 with those in Table 2.
Given the preceding results, I found I believe it is evident that the matched-firm
approach is a better-specified method of abnormal performance detection than the
methods using portfolio-matching strategies. However, I could not find significant
statistical evidence to support my affirmation that the matched firm approach using
industry affiliation as the matching criteria produces better-specified test statistics than
the other matched firm approaches. When samples contain 50 firms the matched-firm
68
approach using market capitalization and book-to-market ratios perform a little better
than the industry benchmarks; on the other hand, when samples contain 500 firms the
matched-firm approach using industry affiliation was better- specified than the competing
methods, but again, the differences are not statistically significant.
Section 2: Tests of Power
The purpose of power analysis was to determine what test method had the least
type II error, and which methodology had the highest power. To run the power test, I
calculated the empirical power (EP) statistic by simulating abnormal performance to
determine at which level of abnormal performance various methods of detection
identified simulated abnormal performances. I based the method used to calculate the EP
statistic upon the method advanced by Ang and Zhang (2004); however, I simulated
abnormal performance at levels of +/- .01, .05, .10, .15, .20, .30, .50, and .75. In essence,
to calculate the EP statistic I have forced the average abnormal performance away from
zero, the hypothesis is as follows:
Hypothesis 2:
H0: BHAR is zero

H1: BHAR is not zero
To continue the examination of the seven potential methods, I conducted power
analyses by simulating abnormal performance using the same number of samples and
sample sizes as in the specification analysis. By taking the average performance and
either adding or subtracting a certain percentage of performance to simulate abnormal
performances I have identified significant differences between the matched-firm and

69
portfolio-matching strategies. Figure 1 illustrates how fast each of the seven methods of
abnormal performance detection responded to simulated abnormal performance.
Figure 1. One-year power curve.
In Figure 1, all of the matched-firm approaches have a defined power curve--the
traditional U or V shaped curve—the power curve are centered approximately centered
on zero, the point where no abnormal performance is simulated. In comparison, the
portfolio-matched benchmarks had no defined structure or at least not the structure
needed to make credible inferences pertaining to the power of the benchmark. Again, the
portfolio benchmarks failed to approach acceptable standards that are necessary to judge
the benchmarks ability to detect abnormal performance; in the remaining analyses, the
portfolio techniques were not included in the analyses because I did not considered them
to be meaningful alternatives to the matched-firm approach.
Figure 1 shows that even if the event firm produced were 15% of abnormal
performance, the competing matched-firm approaches only rejected the null hypothesis
70
(identifying abnormal performance) in approximately 30% of samples. This seems rather
low; researchers would reject simulated abnormal performances of greater than 30% in
70% of the tests. After comparing these numbers with the results found in Figure 2 where
I expanded the number of firms per sample to 500 from 50, the matched firm approach
rejected 80% of the samples if I imposed 10% points of abnormal performance, compared
with approximately a 17% rejection using the smaller sample. Therefore, as I increased
the sample size, the power curve narrowed making the employed methodology
appropriate.
Figure 2.One-year power curve (n=500).
There is still no statistically significant difference between the different matched-
firm approaches to benchmarking. When conducting these tests I was concerned with the
speed at which the metric deteriorates. As I increased the event period, for example when
analyzing long-term abnormal performance, the method’s ability to detect abnormal

71
performance decreases. I presented evidence in Figure 3, using a four-year time horizon
with a simulated abnormal performance of 10%; I identified of abnormal performance
6% of the time, which was a third as powerful as the one-year time horizon. However,
relative power differences between the two time horizons were not consistent throughout
the entire power curve; sometimes methods used to test the four year time horizon was
1/6th as powerful as the one-year time horizon and near the extremes +/- 75%, is the
differences decrease to 6/9ths
Figure 3. Four-year power curve (n=50).
It is apparent that, as researchers lengthen the study’s time horizon concerned for
falling prey to Type II statistical error--not rejecting the null even though abnormal
performance has occurred--increases. As the sample size is increased, this error should
decrease and the power of the statistical tests used in the analysis increased. While
reviewing the changes in the shapes of the power curves in Figures 3 and 4, I noticed the
significant changes in the power curve as the sample size was increased. However, I
could not detect any simulation of abnormal performance under 15% by the BHAR-
72
Matched firm approach to benchmarking using a four-year time horizon. It is my
subjective assessment that researchers need to be skeptical of any analysis using an event
horizon in excess of 3 years. Even a three-year event window may be too long of a
horizon to lead to a meaningful.
Figure 4. Four-year power curve (n=500)
In this power analysis, as the sample size increased (50 companies → 500
companies) the number of samples decreases immensely (180 samples → 18 samples).
The bands used to evaluate whether I could state with a 95% confidence that the metric is
powerful enough to detect abnormal performance at given levels of performance
increased as the number of observations used in the study decreases. I employed the
following formula to construct the confidence interval:
.95(1 − .95)
.95 ± 1.96
18 (8)
The following confidence bounds were constructed, on the upper end 100% and at the
lower bound 85%, this suggested that if abnormal performance in excess of positive or
73
negative 10% is experienced--using a one-year time horizon--I can be 95% confident that
abnormal performance actually occurred. As the time horizon of the event increases to
two (three) years the simulated abnormal performance necessary to induce 95%
confidence in the outcome of the analysis, exceeds 20% (30%). I have presented a more
comprehensive review of my power analyses in Appendix C through H.
The purpose of the test of Hypothesis 1 and 2 was to determine which of the
seven benchmarking methods selected in the literature review would provide the best
balance between specification and power or which method most appropriately balanced
Type I and II errors. As in most analyses I was more concerned with type I errors than
type II--more succinctly it is better to err on the side of caution, not to conclude that
abnormal performance occurs when in actuality it had occurred, than to conclude that
abnormal performance occurred when in reality it did not. In the sensitivity analysis, the
only benchmarking method that did not generate well-specified test statistics was the
matched-firm based upon market capitalization alone. After taking more time to analyze
the differences between the competing methods it is evident that, although not
statistically significant, the match-firm strategies based upon industry affiliation and the
market capitalization combined with the book-to-market ratio were the best specified of
the competing benchmarking techniques, which implies that these methods are better than
the portfolio approaches.
Even though I cannot conclude that the matched-firm approach to benchmarking
using industry affiliation was statistically better than the matched-firm approach based
upon market capitalization and book-to-market ratios, from a qualitative perspective this
74
conjecture made sense. In the preceding specification analysis using both a sample size of
50 and 500, the results of my analysis show that both metrics produce well-specified test
statistics and in the power analyses both metrics performed similarly. However, the
additional information needed to obtain data for the matching strategy using market
capitalization and book-to-market ratios make this technique less preferable. Using IPO
data from my list and querying the CRSP database for data on 82 IPOs I found that I
could match all queries based upon industry affiliation. However only 24 of these IPOs
generated all of the data needed to calculate either market capitalization or the book-to-
market ratios--this means only 29% of the IPOs would have survived the initial sort.
Continuing this argument, I attempted to obtain at least 90% of the original sample, by
matching the IPOs based upon the appropriate characteristics, if I matched only 29% of
the sample using CRSP data this would leave 61% of the sample without a match. The
only effective alternative matching procedure I identified was conducting a Google
search using the company’s ticker or name, the Google search results are less reliable,
and should be used as a last resort – therefore, I have used the matched-firm approach
using industry affiliation as the matching criteria in the following analyses.
Section 3: Initial Performance
In the following section, I have focused on the initial trading period. The main
questions posited in the following section were whether unseasoned IPOs produced
abnormal performances in the time proceeding public trading and if this abnormal
performance continued into the first day of public trading. I first reported the results of
75
the analysis carried out prior to public trade and analyzed whether IPOs produce
abnormal performance on their first day of trade.
IPO Performance (Pre-issuance)
I now turn to addressing the results of analyses conducted on IPO performances
obtained prior to public issuance, from offer to issuance. I used standard indexed
benchmarks to gauge normal performance, because the pre-issue performance period was
generally constrained to a period when the U.S. markets were not trading—the company
issues its shares on the night preceding its initial day of trade.
Hypothesis 3:
H 0 : R IPOAverage (Offering → InitialTrade ) ≤ 0

H 1 : R IPOAverage ( Offering → InitialTra de ) > 0
X −0
The t test is as follows - t = , the sample average was 11.74%, with a sample
S n
standard deviation of 31.16%, and 1876 observations taken from April 12, 1996 to
January 29, 2008. The resulting t statistic was 16.32, which is outside the critical value of
1.645 for a one-tail t test given a 5% level of significance. On average, from April 12,
1996 to January 29, 2008 unseasoned IPOs produced an 11.74% return, this result was
statistically significant from zero using a 95% level of confidence, therefore, I rejected
the null hypothesis in favor of the alternative hypothesis.
The preceding analysis illustrated the difference between the performance
obtained by IPOs pre-public trading and an expectation of zero abnormal performance,
because this is the pre-public trading period, there is no specific way to pair the
76
individual IPO performance with a benchmark. Therefore, I have aggregated the returns
into monthly IPO performances, these performances assume the investor obtains shares
of the IPO in the offering and sells them at the initial trade on the first day of public
trading. In Appendix J, I have illustrated how abnormal IPOs perform in pre-public
trading. To accomplish this goal I calculated the average monthly performance of IPOs
versus those of DJIA, Russell 3000, and the NASDAQ Composite Indices.
Table 3
Pre-Public Trading Averaged Monthly IPO Performance
Comparision of Average Monthly Performances

IPO DJIA Russell 3000 Nasdaq
Sample Average Return 8.96% 0.55% 0.46% 0.23%
Standard Deviation 0.14 0.04 0.04 0.08
Count 139.00 139.00 139.00 139.00
T-value 7.81 1.60 1.22 0.33
As the numbers in Table 3 indicate, at 5% level of significance for a one-tail t test (t
critical of 1.66), I rejected the null hypothesis for only the IPO sample, implying that the
IPO group experienced significant abnormal returns. None of the benchmark indices
produced abnormal returns. However, the t statistic obtained for DJIA was 1.60, which is
quite close to the critical value.
Given that the DJIA was the best performing benchmark, I continued my analysis
by testing for the abnormal performance of the averaged monthly IPO performance prior
to public trade cohort against the monthly performance obtained by the DJIA.
Hypothesis 4:
H 0 : RIPOMonthly(Offering→InitialTrade) ≤ RDJIAMonthly
H 1 : RIPOMonthly(Offering→InitialTrade ) > RDJIAMonthly
77
I then carried-out a paired t test to conduct this analysis. I calculated the resulting
t value using the method mentioned earlier in this section. The average difference
between the IPO group and the DJIA was 8.41%, with a sample standard deviation of
13.86%, and observations’ occurring over 139 months--the computed t statistics was
7.15. Again, with a 95% level of significance for a one-tailed test the critical value of t is
1.66; therefore, I rejected the null hypothesis and determined that significant abnormal
performance occurred during the pre-public trading period when compared against
standard indices.
Initial Day of Public Trading
In this round of the analysis, I attempted to answer whether IPOs generate
abnormal performance on the first day of public trading. To answer the question I used
IPOs issued to the public from January 1, 1997 to December 22, 2005, the sample
contains 2,143 observations, and I conducted the following analysis.
Hypothesis 5:
H 0 : R IPO ( Day 1) ≤ R IndustyMat chedFirm

Using a standard t test, I uncovered the following; the average return across the IPOs in
this analysis was 3.44%, and the average performance of the matched-firm benchmark
was 0.13%. The sample standard deviation was 16.27%; therefore, I have calculated a t
value of 9.423, which when compared to a critical value of 1.645, at a 95% level of
significance, indicated that the IPOs abnormal returns on the first day of trade are
statistically significant. I rejected the null hypothesis, in favor of the alternative

78
hypothesis, which stated that the returns of IPOs on the first day of trade are significantly
different from the returns obtained for the matched-firm benchmark.
Section 3: Long-term Abnormal Performance
In this round of the analysis, I turned to evaluating significant abnormal
performance occur after the short-term abnormal performances. To accomplish this I
canvassed the population of IPOs issued in the U.S. from January 1, 1985 to December
31, 2002. I used 5583 IPOs that in the analysis, and took each of these IPOs and matched
them to a firm using industry. I then calculated the product of the HPRs of the IPOs and
subtracted the product of the matched-firm benchmarks HPR. Finally, I used a derivation
of the BHAR formula as per the following:
BHAR = ∏ (1 + rIPO ) − ∏ (1 + rIndustryMatch )

(9)
Next, after each trading day, I averaged the individual BHARs and I calculated the
standard deviations, across the entire sample. Therefore, the output, which encompasses
trading day 2 through 750, is the averaged BHAR across the entire sample over the
specified time horizon. Next, I conducted two-tailed t test for all 749 time-horizons.
79
Figure 5. Long-term abnormal performance
Analysis of the data provided in Figure 5 shows that, during trading days 5 through 12
IPOs significantly underperformed the matched-firm benchmark, at day 17 the trend
changed positive and it was significantly positive until trading day number 120 (with one
insignificant reading on day 33)--at day 120 the BHAR is 1.934%. The averaged BHAR
continued along insignificantly, but positive, until reaching trading day 161. However,
the BHAR did not generate a significantly negative BHAR until it reached 201 trading
days of seasoning. The BHAR remained significantly negative through the rest of the
analysis.
80
BHAR
(IPO V. Industry Firm Match)
5.00%
Cumulative Average
0.00%
-5.00%
BHAR
-10.00%
-15.00%
-20.00%
-25.00%
1 82 163 244 325 406 487 568 649 730
Number of Trading Days
(Seasoning)
Figure 6. BHAR (IPO V. matched firm)
In Figure 5, I provided the t values that I calculated when comparing IPO
performance to the matched-firm returns. Figure 6 shows that the trend of the BHAR was
unmistakably negative. Overall, there is an abnormally positive performance of
approximately 3 % occurring within the first year and at the end of year three the
maximum abnormally negative performance occurred, which was -22.41%.
Section 4: Quiet and Lockup Expiration
To construct a test for abnormal performance at the expiration of the lockup and
quiet periods I canvassed the same population of IPOs that I used in the long-term
analysis. The number of observations for the quiet and lockup period analyses was 5529.
To carry out these analyses I calculated the 5-day BHAR surrounding the date in which
the quiet period ended and the lockup period expired.

81
Hypothesis 6:
H 0 : BHAR Quiet ≤ B H A R IndustyMat chedFirm
H 1 : BHAR Quiet > B H A R IndustyMat chedFirm
For the analysis of performance surrounding the expiration of the quiet period, the sample
average BHAR was 1.64% for the five-day period surrounding the event and the sample
standard deviation was 13.9%. I used this data and calculated the t statistic of 8.75. Using
a 95% level of significance the critical value was 1.645; I rejected the null hypothesis and
concluded that at the conclusion of the quiet period IPOs produce a significantly positive
abnormal performance.
Hypothesis 7:
H 0 : BHAR Lock − up ≥ B H A R IndustyMat chedFirm

H 1 : BHAR Lock − up < B H A R IndustyMat chedFirm
In the analysis of the performance resulting from the expiration of the lockup
period, I identified a significantly negative performance of 1.42%. In addition, the sample
standard deviation was 34.4%, therefore, the resulting t test produce a test statistic of -
3.08, and with a 95 % level of significance the critical t value is, again, -1.645. Therefore,
again, I rejected the null hypothesis and conclude that the -1.42% of performance is
significantly negative.
Conclusions
In the preceding section, I have (a) presented a well-specified and powerful
method of identifying abnormal performance when conducting event studies, (b) shown
that short-term abnormal IPO performance is positive, (c) illustrated that events occurring
82
throughout the IPO process instigate abnormal performances, and (d) provided a
description of IPO performance over the initial three years of seasoning. The results of
the analyses related to event specific performances--abnormal performances occurring at
the expiration of the quiet and lockup periods--generated significant, but not substantial
abnormal performance. However, initial abnormal performance of 11% in the pre-
offering period and 3% in the initial trading day, together with long-term
underperformance of IPOs in excess of 30% seem to suggest that substantial performance
abnormalities occur when IPOs issued unseasoned equity shares.

CHAPTER 5:
SUMMARY, CONCLUSION, AND RECOMMENDATIONS
Introduction
In this chapter, I have reviewed the findings of the analyses conducted in the
preceding chapters, provided an interpretation of these findings, and summarized the
conclusions reached in the preceding sections. In addition, I discussed the implications of
the analyses for social change and recommendations are advanced. The outline of this
chapter is as follows: (a) overview, (b) interpretation of findings, (c) implications for
social change, (d) recommendations, and (e) conclusions.
Overview
The overarching goal of this dissertation was to evaluate whether IPOs produce
predictable abnormal performances around specific events and over specific time
horizons. I have divided the analyses into four groups: (a) specification and power
analyses, (b) short-term IPO performance, (c) longer-term IPO performance, and (d)
event-related IPO performance. In the following paragraphs, I have provided an overview
of the results of the analyses conducted to evaluate the hypotheses presented in the four
hypothesis testing sections.
The results of the specification and power analyses quelled some methodological
concerns I had mentioned in the literary review. All of the matched-firm (matched using
market capitalization, industry and market capitalization, industry, and market
capitalization and book-to-market ratios) approaches used as benchmarks for normal
performance produced well-specified test statistics. However, I found misspecified test

84
statistics for each method of benchmarking based upon portfolio constructions (using
primarily the same matching characteristics). This trend continued in the power analyses.
For the matched-firm approaches, the power curves centered around zero and the curve
maintained a U or V shape as I simulated abnormal performance, while the power curves
of matches based upon portfolio construction did not produce power curves centered on
zero and produced no definite form. At the conclusion of the analysis, I determined that
there was little specification and power differences between the numerous matched-firm
approaches, but all of these approaches were preferable to the portfolio-based matching
strategies.
In the analysis of short-term IPO performance, I found that both prior to issuance
and in the first trading day after issuance IPOs produced significantly positive abnormal
performances. The magnitude of performances differed substantially, for the preoffering
period IPOs, on average, generated 11.74% of abnormal performance. For the first day of
public trading IPOs, when compared against the matched-firm benchmark, produced a
significant abnormally positive performance of 3.44%. I also aggregated and combined
the monthly pre-IPO returns, compared them against the average monthly return on the
DJIA, and discovered that they produced a significantly positive abnormal monthly
performance of 8.41%.
Next, I turned to analyzing long-term IPO performance, from trading day 2
through day 750 (~3 years of daily data). Initially, IPOs underperformed their matched-
firm comparisons significantly, but shortly thereafter, they trended positive and had
significantly positive abnormal performances until trading day 120. At the 161st trading
85
day the IPO’s BHARS, on average, began to produce negative performances, and on
trading day 201, the negative performance began to be significant. After this significant
abnormal performance reading, approximately at the end of the first year of trading, IPOs
continuously and significantly underperformed their matched-firm benchmarks up until
their 3-year anniversaries. The magnitude of the long-term underperformance is evident;
the 3-year BHAR was a significant -22.41%.
In the final set of analyses, I analyzed whether significant abnormal performances
occur at the expiration of the quiet and lockup periods. When testing for abnormal
performance at the conclusion of the quiet period I discovered a significantly positive
abnormal performance of 1.64%. Similarly, at the conclusion of the lockup period, I
discovered a significantly negative performance result of -1.42%. In this section, I have
summarized the major findings generated in this analysis of IPO performance; I will now
turn to discuss how the reader should interpret these results.
Interpretation of findings
In this section of the dissertation, I have attempted to integrate the findings of the
results section, with the conjectured relationships presented in the literary review. Again,
I have addressed each of the four research sections separately. Therefore, I have
evaluated the findings of each of the four sections in the following order: (a) specification
and power analyses, (b) short-term IPO performance, (c) longer-term IPO performance,
and (d) event-related IPO performance.

86
Specification and Power Analyses
The results of the specification and power analyses were very surprising. First, I
spent a great deal of time developing ideas for dealing with misspecified tests and biases
that plague many models used as models of normal performance; in hindsight this time
expenditure seems unnecessary. Second, I was struck by the severe differences in the
specification and power of the two main methods of benchmarking—the matched firm
and portfolio construction. Third, it is important to highlight that there were similar
specification and power tests produced by all of the matched firm approaches to
benchmarking. I will now address each of these issues, respectively, in the following
sections.
Biases
In my opinion, Barber and Lyon (1997) and Lyon, Barber, and Tsai (1999)
forwarded the best and most recent attempts to evaluate the specification and power of
methods used to model normal performance in event studies. In Barber et al. (1997) the
researchers spend a great deal of time discussing biases related to methods used to detect
abnormal performance, but state that all of these biases were overcome when using the
matched-firm approach combined with the BHAR method. However, in Lyon et al.
(1999) the researchers continue the investigation of methods used to test for abnormal
performance, and attempt to correct the portfolio-based metrics using bootstrapped t
statistics and creating pseudo-portfolios to calculate empirical p values.
The explanation of the bootstrapping procedures and the calculation of the
empirical p values are beyond the scope of this analysis, but all of this work seems
87
unnecessary because the matched-firm approach to benchmarking, after researchers made
these complicated adjustments to the competing portfolio methods, the matched-firm
approach still produced better-specified test statistics than either method. The
complicated procedures, implemented in Lyon, et al. (1999), do improve the results and
distribution of the power test of the portfolio methods, as much as 20% improvement in
power given a +/- 10% simulation of abnormal performance. Although the curves are still
unevenly distributed, the empirical p value approach was the more powerful method of
detecting negative abnormal performance and the bootstrapped t statistic is a more
powerful technique to use to identify abnormally positive performance. When I compared
these to the symmetric power-curves obtained using the matched-firm approach, the
complicated adjustments seemed unnecessary. Using the matched-firm approach to
benchmarking seems to be the correct choice.
Matched-Firm vs. Matched-Portfolio Approach
When discussing the event study literature I believe it is imperative to point out
the magnitude of the differences between the matched-firm and matched-portfolio
approaches to benchmarking. In an important study analyzing event study methodologies,
Brown and Warner (1985) make a convincing argument against conducting events
studies using daily returns, again the researchers focus on how the defined benchmarks
are biased and other researchers build their cases for not conducting event studies using
daily returns around the proceeding conclusions reached in Brown et al. (1985).
However, taking the Brown et al. (1985) analysis and attempting to project it on present
day analyses would not be relevant, because the researchers were concern solely on
88
analyzing the ability of portfolio-based metrics to identify abnormal performance. Brown
et al. (1985) found similar problems with the portfolio approach as Barber and Lyon
(1997) these findings seem to encourage additional skepticism in the results of any long-
term analysis of event studies, regardless of the method used to aggregate the returns.
Researchers should approach with skepticism any evaluation of event studies using basic
portfolio-matching strategies. However, this skepticism should not transcend the portfolio
population and infect the match-firm approaches without evidence of problems resulting
from the use of the matched-firm benchmarking technique. Researchers should first
choose the metric and then evaluate it.
In the hypothesis tests, I found that all of the matched-firm approaches were well
specified and seemed to be reasonably powerful. In these analysis, I was more concerned
with how well specified the given metric was, minimizing false positives, than
concluding that no abnormal performance had occurred, when in actuality it had not.
Portfolio-matching techniques generated poorly specified test statistics in all situations,
and across all samples. The reason that the portfolio-matching techniques perform poorly
is that the portfolio construction approach minimizes idiosyncratic risk and the matched-
firm approach does not. When researchers calculate the BHAR using an event firm and
the portfolio-approach to benchmarking, researchers categorize the idiosyncratic risk
associated with the event firm as abnormal performance. Contrarily, when the matched-
firm approach is used there is not an attempt, by construction, to minimize either the
matched-firm or the event firm’s idiosyncratic risk, researchers allow both metrics to
89
perform abnormally--there is no normalizing agent. Portfolio methods are, theoretically,
poor predictors of abnormal performance.
Matched Firm—Specification and Power
Many researchers have attempted to discredit the results of analyses that illustrate
that abnormal performance occurs in IPO performance. Researchers pursuing this line of
inquiry generally discredit some element of the statistical methodology used to measure
the abnormal performance. There are many approaches researchers can use to attack
methods constructed to detect abnormal performance, these researchers can discredit: (a)
the method of calculating abnormal returns (CAR v. BHAR), (b) the statistical method
used to identify abnormal performance (specification and power analyses), or the
benchmark’s construction (portfolio- v. matched-firm or quip that the abnormality is a
size/industry phenomenon). After constructing arguments against abnormal performance
based upon the aforementioned concepts, the argument then seems to regress to problems
that have plagued evaluations of securities’ performances for decades or since their
inception--the distributions of returns or BHARs seems to be skewed, peaked, and have
fat tails. These are problems that plague every analysis of securities’ performance, and if
someone were to generate a solution to this problem, this solution would have already
been integrate into statistical tests of security’s performance (whether researchers use a
different distribution when conducting analyses on securities performance – Gaussian v.
Non-Gaussian--or something totality original).
The point I am attempting to raise is that all models are not accurate reflections of
reality—they are normative models of finance used in research projects. Even if those
90
same problems that plague the majority of models plague the data examined in this
analysis, economic and statistical assumptions, the procedure of detecting abnormal
performance using the matched-firm approach has produced well-specified results
throughout this and other specification analyses. I would have liked to generate better
results from in the power analyses, however, given the shape and structure of the power
curves across each of the matched-firm approaches; I am satisfied with the outcomes of
this analysis. The matched-firm approach to benchmarking for normal performance
seems to work very well--determining what firm characteristic to base the application of
the matched-firm approach is a more difficult problem to solve.
Short-term IPO Performance
The important questions I attempted to answer in this section of the analysis were
whether short-term abnormal IPO performance is constrained to the period prior to public
trade. Researchers have previously shown that the abnormal performance is significantly
positive in the short-term (between the offer and the close of trade on the first trading
day), however some have questioned whether the abnormal performance occurs prior to
or after the IPO begins public trading. The outcome of this analysis surprised me.
Historically, researchers seem to have categorized the first-day returns as
occurring from the offer to the close of trade on the first day of trade. Recently, Cheng,
Cheung, and Po (2001) found that in a study of the Hong Kong IPO market, abnormal
returns occurred only in the period prior to public trade. This conclusion led me to
question whether conclusions reached in previous analyses of the IPO market in the U.S.
were accurate. If abnormal performance is constrained to the period prior to public trade,
91
then questions about market efficiency are not valid because this concept of market
efficiency does not apply to interactions between the underwriting syndicate and the
prospective initial purchasers of IPO shares. Generalization of the conclusion reached in
Cheng et al. (2001) to IPO performances in the U.S. would have been a critical finding.
However, the results from my analysis illustrated the opposite; that abnormal
performances occurred in both periods--prior to public trade and on the first day of public
trading. The results were both positive and statistically significant: (a) 11.74% abnormal
return for the period prior to trading, and (b) 3.31% abnormal return for the initial day of
public trading.
Looking at the trends of the performances of these two abnormalities, it seems
that the risk-adjusted anomaly has been increasingly pronounce in the post bubble period,
or at least after 1998 (see Appendix J), and there is a correlation between the volume of
IPOs issued in a month and their initial performances in the 1st day of trading. Even
though the regression ran on the 1st day of trading was significant, the researcher believes
that these statements in this paragraph are subjective, therefore, researcher need to run
further tests (expanding the sample) before producing reliable results. However, when
looking at the preissuance period, I found that, the relationship between the risk and
return seems to be suspect in the post-bubble period. Reviewing the heightened values of
the test statistics, it seems as though companies offered investors a greater amount of
return on IPOs for less risk than in the period during and prior to the bubble.
The other interesting finding is that, for the first day of trading, there is a
relationship between the number of IPOs issued in any given month and the performance
92
of IPOs issued in that month. By running a regression of the monthly initial day IPO
performance against the number of IPOs issued in that month I produced the following
statistics: (a) R2=25.5%, (b) n=107, (c) ALPHA = 6.67, with P-Value =.019, (d) Beta =
121, with P-Value = .000, (d) FCRTICAL=34.25, and (e) the significance of F 6.15E-8.
These two findings fit theories attempting to explain current trends of IPO performance.
For example, if markets produce greater returns in times of greater issuance, the
relationship between performance and speculation seems to be important--more is good--
not based upon fundamental value. Likewise, the deterioration of the relationship
between risk and return, in the period following the 2000-bubble, seems to imply that
investors did not understanding the magnitude of the risks they were taking to allow them
to obtain the returns they were obtaining. It seems like one of the two parties, the
company (underwriting syndicate) or the investor, was being either overly generous or
overly greedy. It will be interesting to see how or if, as liquidity dried up substantially in
2008, this risk/return relationship will change in 2008.
Long-term IPO Performance
The academic community has reviewed analyses of long-term IPO performances,
critically. However, the skeptics have not been unreasonably critical, methodological
problems, small sample sizes, and a lack of integrated analyses of various components of
the long-term performance quandary have been lacking. Results presented in the
specification and power analyses illustrate how effective the matched-firm approach to
detecting long-term abnormal performance is. The real question, when considering the
93
longer-term analysis of IPOs using this matched-firm technique is if the method would be
powerful enough to identify abnormal performance.
The results and the trends seem to conform with traditional theories pertaining to
longer-term IPO underperformance. Brav, Geczy, and Gompers (2000) focus on a five-
year analysis of IPO performance, and seem to conclude that IPOs really do not produce
significant long-term term underperformance. However, I found, using the matched-firm
approach by industry, a significant three-year IPO underperformance of -22.41%. This
finding is similar to the results found in Ritter (1989) of -27.39% at the end of the first
three-years of seasoning. The conjecture remains, IPOs tend to produce abnormally
negative performances in the first three years of seasoning and somewhere between year
4 and 5 these negative performances seem to normalize.
Event-related IPO Performance
In this section, I have discussed the results reached in analyses of the expiration of
the quiet and lockup periods. The results of both of these analyses add credibility to
comparison papers and stand in contrast with analyses that produce contrary results. I will
begin by discussing the results presented in analysis of the quiet period and follow this by
the results found in the lockup period.
I found, analyzing the performance around the expiration of the quiet period, a
significantly positive abnormal performance of 1.64%, using a sample of IPOs that went
public from 1985 through 2002. Bradley, Jordan, and Ritter (2003) found that from 1996
to 2000 IPOs produced a significant cumulative five-day average return of 3.1%.
However, Bradley, Jordan, Ritter, and Wolf (2004) followed this analysis up by taking a
94
sample of 96 IPOs issued in 2001 through 2002 and found that IPOs no longer generated
significant abnormal performance at the expiration of the quiet period. The results of both
of these analyses make sense, in the 1996-2000 periods I obtained an average BHAR of
3.29% and in the 2001-2002 period the average return during the expiration of the quiet
period was equivalent to -.71%. From 1985-2002, on average, IPOs produce significant
and substantial abnormal performances, however these abnormalities are not as large as
previously thought.
After conducting an analysis of the lockup expiration, I found that IPOs produce a
significant abnormally negative performance of 1.42% when compared to firms matched
by industry. Using a sample from 1988 to 1997 Field and Hanka (2001) found similar
results, a significant abnormal performance of negative 1.5%; however, Garfinkle,
Malkiel, and Bontas (2002) using a sample of IPOs issuing shares from July 1997 to
December 1999 identified a significantly negative 4.47% abnormal performance.
Garfinkle et al. (2002) compared the returns generated on an investment in the NASDAQ
market index against the event firms, and Field et al. (2001) used the CAR method of
abnormal return aggregation, benchmarking with the CRSP Value-Weighted Market
Index. After reviewing Appendix L, yearly abnormal performance data surrounding the
lockup expiration, I am not sure how to account for the vast differences between the
results presented in the Garfinkle et al. (2002) analysis and the results obtained in this
sample and those from Field et al. (2001).
There exist, substantial discrepancies between what I have found and analyses
conducted by more seasoned researchers in the field, however, my conclusions are as

95
follows. Even though IPOs do generate significant abnormal performances in the five-day
period surrounding the expiration of the quiet and lockups period, the magnitude of these
abnormal performances are minuscule, 1.64% and -1.42%. However, the investor only
exposes his or her capital to the investment for the five-day period, and if investors repeat
this investment strategy 61 to 521 times each year, the gains could justify the time and
effort it would take to employ investments using these strategies.
Implications for Social Change
In this section, I have integrated the findings related to abnormal IPO
performance and market efficiency with concepts that researchers could use as catalysts
for social change. Some of the connections are obvious, but in some instances, they may
be somewhat opaque. The main headlines of this section are: (a) short-term IPO
performance a factor that increases the income disparity in the U.S., and (b) in this
research project I have uncovered problems with the concept of market efficiency, that
is, abnormal IPO performance provides evidence that markets are inefficient.
Short-term IPO Performance and Income Disparity in the United States
In a speech given by the Chairman of the Federal Reserve Board of Governors,
Bernanke (2007) states that although Americans should not be guaranteed equality of
outcomes, we should guarantee equality of opportunity to these outcomes. Chairman
Bernanke concludes that those at the top of the income distribution have been gaining
more wealth in percentage terms than those in the middle and lower portions of the
income distribution curve, and similar relationship exists when comparing middle to
lower on the income distribution curve. The reader can link this conversation to the short-
96
term analyses of IPO performance conducted in the preceding sections. To reiterate the
findings, in the pre-public issuance period IPOs generate abnormally positive
performances of 11.74%, and on the initial day of public trade, these shares produced a
significantly positive performance of 3.31%. The underwriting syndicate, whose role in
this process is complicated, discretionally allocates these shares, while one of their major
objectives is to incite interest in the offering.
It makes more sense for the promoters of these offerings to court investors with
higher incomes or access to more capital, than it would to court the average investor.
Even if a mechanism existed, which the underwriting syndicate could use to transfer
ownership of the company’s shares these underwriters would refrain from using an
alternative means to allocate the shares, because a portion of their compensation is their
ability to generate and sustain interest in the issue. However, the Dutch auction process of
issuance is an efficient means to allocate shares and this mechanism provides the
underwriter and the company with an approximate fair-value for the company at the time
of offering without offering investors discounts on price of the issue. Nevertheless, this
method of issuance has not become the standard; the reason for this is complex.
First, reason behind the transfer of a portion of ownership in the corporation—
there could be different reasons for this transfer. Is the issuing company attempting to
extend the life of the organization, does the organization want to transfer their risks, or is
it some combination of the two objectives? If the owners are interested in the
continuation of the enterprise and have confidence about the future prospects of the
organization, then they will most likely hold onto a significant proportional share of
97
ownership in the company. Alternatively, if they are interested in a transfer they will
attempt to liquidate their stake and obtain the maximum payoff for their organization—
this process may include the owner’s attempts to deceive the shareholders. Owners that
are interested in transferring some ownership, but also believe that they can obtain
longer-term success could sacrifice the initial 15% of abnormal performance to promote
good will and then conduct an SEO once their longer-term initiatives payoff to finish the
transfer.
Returning to Chairmen Bernanke’s commentary, the opportunity for economic
opportunity should be equivalent across classes, but the actual economic profits obtained
by people need not be equivalent. It is unjust to allow a privileged class of investors’
unfettered access to profit opportunities, if at the same time you restrain other economic
agents from the opportunity to participate in this process. Again, the Dutch auction
process seems to be a good solution to this problem. Investors need to ask the following
question: What is the reason for the equity issuance—the succession planning or risk
transfer?
Chinks in the Armor of Market Efficiency
Empirical research projects focused on detecting abnormal IPO performance
attack the EMH somewhere in between the weak and semi-strong forms of the
hypothesis. Performances obtained prior to public trading have nothing to do with
discrediting market efficiency, but if researchers consider that companies issue these
newly issued shares at or above their intrinsic value, when compared against their peers
they may find question the efficiency of the issuance process. In a sample of more than
98
2,000 IPOs issued from 1980 to 1997 Purnanandam and Swaminathan (2004) found that
IPOs, at their offer, are typically overpriced when compared to their non IPO
counterparts by 15% to 50% depending on the matching criteria.
If initially the issues were on average, overpriced, then why prior to public trading
would these same overpriced issues increase in value in excess of 11.74%? Then in the
initial day of trade, why do these shares obtain another increase of ~3%. These findings
do not fit well together and do not support the EMH. Purnanandam and Swaminathan
(2004) presented further evidence, which help to build a case against market efficiency.
Purnanandam et al. (2004) found that when the strength of the initial misvaluation of
IPOs is excessively positive the ensuing initial performance is greater than when IPOs are
either appropriately valued or undervalued.
Should an efficient market understand when a systematic misvaluation occurs and
correct this inefficiency rapidly? In longer-term analyses of IPO performance, after the
initial underperformance, IPOs then underperform benchmarks by a substantial amount
until either the fourth or the fifth year. Should it take four to five years for the market to
adjust for an initial misvaluation of shares? Likewise, the abnormal performance
experienced at the expiration of the quiet and lockup periods produce significant
abnormal performances; these performances range from 1% to 2%, just minor chinks in
the armor of the concept of market efficiency. Abnormal performances and misvaluations
of substantial proportions found in short- and long-term analyses of IPO performance
may be considered more than a chink in the armor, but a significant deterioration in the
quality and effectiveness the concept of market efficiency.

99
Integration and Summary
The two arguments developed in the preceding section are very important
elements of economic justice and understanding. First, as we allow the markets to
operate, we allow an unequal distribution of economic outcomes throughout the
population; however, we have to ensure that every citizen has an equal opportunity to
participate in economic activities. There is no concept of equality of opportunity
embedded in the process of issuing unseasoned shares of equity to the public.
In the global economy, investors are now seekers of information, the underwriting
syndicate no longer needs to sell investors on a concept or idea--investors can evaluate
the information contained in the offering prospectus and determine if the issue should be
included into their investment strategy. The underwriting syndicate can still play a role in
advising corporations on how to structure their organization, their issuance processes, and
provide an initial valuation of the entity, but is their interference with market forces still
relevant given the advances in communications methods present in today’s financial
systems? Market forces will generally make a more accurate assessment of a firm’s
prospects than a group of privileged investors or, for that manner, the underwriting
syndicate. Freeing up the shares through an auction process may stifle the irrationally
positive response to an issue that is already over valued by 10% to 30%. The positive
11.74% abnormal performance between the offer and the initial trade, followed by a
3.34% abnormal performance obtained on the first day of trade seems to be an irrational
response to newly issued shares.

100
I now will explain why the actions taken by the underwriting syndicate hinder
market efficiency: (a) selective allocation of shares, (b) determining a price without
reference to market forces, and (c) compare the ensuing positive abnormal performances
to fixing a price ceiling on an asset or commodity. If the underwriters impose a ceiling on
new issues, the issuer obtains a payment that is less than it could have obtained without
the ceiling, the quantity demanded is greater than the quantity supplied, and a nonmarket
entity (the underwriting syndicate) rations the shares. As the shares hit the market, those
kept out of the bidding process in the premarket force the price of the security to its
market-clearing price. If companies allowed the markets to operate in the preoffer period,
the process would operate in a more efficient manner.
The market efficiency questions are easier to address, researchers believe either
the magnitude of an anomaly is significant enough to question market efficiency or they
need more proof. The researcher has different interpretations of the magnitude of the
anomalies found in this analysis. Abnormal performances occurring around the expiration
of the quiet and lockup periods are significant, but 1 to 2% is not substantial. However, a
significant 10% abnormal performance for an IPO that was already overpriced, followed
by a 4- to 5-year period of adjustment period to trend back towards normality, does not fit
within the concept of market efficiency. Again, a problem that affects investors and the
firm seem to be resulting from the underwriting syndicates’ interference in the price
discovery process, which is unnecessary given the rapid nature of the dissemination
process of information among all parties.

101
Revision in the IPO premarket pricing and allocation will eliminate IPOs’
abnormal performances enjoyed by a few investors and provides equal opportunities for
all investors to benefit from the IPO process. This will lead to significant positive social
change in the distribution of wealth and income in the society.
Validity and Reliability of the Research Results
In this section, I have presented my views on the validity and reliability of the
results obtained in this research study. In the following paragraphs, I have addressed the
sample sizes used in each analysis of abnormal performance, the power and specification
results, the reliability of this analysis, and the appropriateness of using the event study
methodology to conduct this analysis. I will first address the validity and then turn to
addressing the reliability.
When researchers attempt to determine whether the results of this analysis have
external validity, they should review the findings and conclusions reached in the
specification and power analyses, the sample size of each of the analyses, and evaluate
whether the event study methodology was an appropriate research methodology for me to
apply to this analysis to identify abnormal performance. All of the matched-firm
approaches to benchmarking produced well-specified test statistics and were reasonably
powerful; however, the portfolio-matching strategies produced misspecified test statistics
and produced poor power results. These specification and power analyses were tested on
companies that fall within the Russell 3000 Index of companies; Russell Investments
claims that this index currently represents “approximately 98% of the investable U.S.
equity markets” (Russell Investments, 2008)—a large portion of the U.S. financial
102
markets. In addition, I conducted these analyses over a substantial time horizon, a 22-year
horizon, using a non-repeating sample of 10 observations of 50 companies and a non-
repeating sample of 1 observation of 500 companies per year. Thus, given the simulated
specification and power results and the sample sizes the results obtained from the
matched-firm approaches have external validity.
In my power and specification analyses, using the matched-firm approach, the
methodology I used to detect abnormal performance in sample sizes in excess of 50
companies produced well specified and powerful results; however, when I increased the
sample size to 500, the results of the specification analysis improved and the slopes of the
power curves increased substantially. Therefore, to obtain results that can be generalized
researchers should use large sample sizes when using this methodology. My sample sizes
were as follows: (a) pre-trade analysis—1,876 observations, (b) 1st day of trade
analysis—2,143 observations, (c) long-term performance analysis—5,583 observations,
and (d) quiet and lockup expiration analyses—5,529 observations. The list of IPOs used
for the quiet and lockup expiration analysis had 6,525 IPOs identified for the entire time
horizon analyzed (1985-2002); therefore, 15% of the original list failed to meet the
requirements of the quiet and lockup period analysis, 14% for the long-term analysis. The
time horizon for my analysis of performance experience during the 1st day of trading
went from January 1997 to December 2005, during this period, I successfully (obtained
all of the required data) identified 2,289 IPOs from the list, 6% of these IPOs failed to
meet the requirements for inclusion in this analysis. For the preissuance trading
performance I relied on the Hoover’s IPO and Edgar IPO databases, these databases,
103
combined, produced a sample of 1,876 IPOs that had both offer and initial trade data and
went public from April 1996 to January of 2008—100% of these IPOs were included in
this analysis. The samples used to test for abnormal performance were well in excess of
the samples size of 500 observations needed to produce a well-specified and powerful
test of abnormal performance in the simulation analysis.
To complete the discussion of validity, I conclude by arguing why I chose to use
the event study methodology to conduct this analysis. In chapter 3, I wrote a few pages
explaining the reasons why competing research methodologies would have been
unsuitable for the event study methodology when researchers are attempting to identify
abnormal performance. Experimental design, case studies, correlational analysis, surveys,
and other research methodologies are unsuitable for the event studies for numerous
reasons explained in chapter 3 pages 46-49. I could have used a causal comparative
approach in this analysis; however, it seemed that researchers could better apply this
method to conduct research in educational and social settings—researchers have
consistently applied the event study methodology to examine whether information flows
or events cause abnormal performances. My choice of research methodology for this
analysis was the event study methodology and I think researchers would have a difficult
time proving that another methodology can be better suited to analyze abnormal
performances occurring in security’s price performance.
Leedy and Ormrod (2005) define reliability as “the consistency with which a
measuring instrument yields a certain result when the entity being measured has not
changed” (p. 29). If I were to recalculate the result of this analysis using the same match-
104
firm and portfolio-approaches, on the same event firms, I would obtain the same results.
As such, reliability does not apply to this study, which is an event study.
Recommendations for Action
I have one take home message or recommendation for action and that is to let the
market control the process of unseasoned equity issuance. Let the market set the initial
value for the offering, let it stabilize the price of the security itself, and let it decide where
to redeploy the valuable resources of intellectual capital that are currently used to
interfere in the IPO market. The rationale to maintain the current system is that the
market needs some force to generate demand for an issue and that it needs a force to
stabilize the price of a new issue after its issuance. I will explain here why this process is
unlikely to occur and who needs to demand the change in the system.
Allowing market forces that determine prices would initially affect issuers and
investment bankers in a negative manner. Underwriters initially overestimate the value of
the new issue as compared to similar companies that are not IPOs and they constrain
supply creating excess demand, which forces up prices in the secondary market--why
would issuers want this to change? The underwriters, in this process, conduct road shows
to drum up demand, prepare the prospectuses and other complicated documentation,
determining the value of the offering, providing stability in the aftermarket, and the
researcher is sure there are additional tasks that the underwriting syndicate is involved in
that are unknown to the researcher. However, it seems that these market professionals are
so sophisticated that they have found a way to interfere legally in the markets in order to
obtain a proportion of the value of the offering. So are there any incentives for the issuers
105
and the underwriters to change the system? If such incentives do exist, they would be
negligible in comparison to the incentive to accept the status quo.
Because the issuers and the underwriters have no serious incentive to change,
some other governing force needs to promote change in the financial markets. Legislative
solutions are very seldom the best way to provoke change, especially in a market system.
What, then would the solution be? I believe that the free market would have to determine
that the process of shares issuance is inefficient and that changes are necessary. This
implies that the average market participant needs exposure to these abnormalities—the
the wider the dispersion the better, and that the most efficient way to change the system
seems to be a collective whisper from the general populists, call it a nudge from the
average citizen. In my opinion, this process is generally the most effective method to
accomplish almost anything, although it is probably the most difficult task.
Recommendations for Further Study
There are many potential recommendations for future study that could naturally
flow from the results, or the interpretation of the results, of this analysis; however, I
would like to focus on the most important ones. Prior to this analysis, it seemed to me
that researchers questioned the results of studies conducted on IPO performance because
of the (a) methodological approach used to evaluate the anomalies, (b) fragmentation of
the analyses, or (c) the sample sizes were inadequate. In this analysis, I conducted
specification and power analyses, which generated acceptable results, used the best
method to conduct the analyses, lengthened the time horizons of studies that researchers
106
previously executed with small samples, and integrated the analyses. Therefore, in my
opinion, these anomalies are real.
The first recommendation for further analysis is that researchers should conduct
more research on the matched-firm approach to benchmarking using the BHAR method.
Many researchers avoid using this method of analysis when conducting event studies and
prefer to make complicated adjustments to portfolio –matching strategies even when
conducting studies in event-time. It would be interesting to see if researchers could use
different firm characteristics to match-firms with no pre-event performance with event
firms--maintaining the specification results derived from this analysis--and increase the
power of these methods of analyses. Increasing the power of the BHAR approach to
evaluating abnormal performance matched by some firm-specific characteristic is the
next critical step for researchers conducting analyses on the methodological approaches
to event studies.
The second recommendation is to determine whether investors initially overprice
IPOs relative to their peers. The results presented in Purnanandam, and Swaminathan
(2004) are astounding when considered in tandem with analyses of the actual
performances of IPOs subsequent to issuance. If IPOs are initially overpriced by 10% to
30%, followed by an average of 11% positive performance in the pre-offering period, and
then another 3.5% positive performance in the first day of performance, it is hard to
suggest that markets are efficiently valuing these shares. Researchers should attempt to
use different metrics to evaluate whether these shares are actually overpriced when
issued.
107
Researchers attempting to prove or disprove whether markets operate in an
efficient manner have generated a substantial base. This proposition is flawed a model of
reality and the flaws are identified in numerous analyses. However, the skeptics of
market efficiency really need to present a meaningful alternative. Of course, the new way
of thinking about market interactions will be viewed with a substantial amount of
skepticism and have to overcome substantial hurdles to dethrone the concept of market
efficiency, but the process of scientific exploration is full of intriguing challenges-
embrace them.
Conclusion
In this dissertation, I have (a) presented a well-specified and powerful method of
identifying abnormal performance when conducting event studies, (b) shown that short-
term abnormal IPO performance is positive, (c) shown that events occurring throughout
the IPO process instigate abnormal performances, and (d) provided a description of IPO
performance over the initial three-years of seasoning. These results presented in this
dissertation seem to be at odds with the concept of market efficiency. The results of the
analyses related to event specific performances, the expiration of the quiet and lockup
period, generate significant, but not substantial abnormal performance. However, initial
abnormal performance of 11% in the preoffer period and 3% in the initial day of trade,
together with long-term underperformance of IPOs in excess of 30%, combined with a
study that concluded that IPOs’ in the preoffering period are overpriced, casts serious
doubts on the concept of market efficiency. Of course, further research needs to be
conducted on some of the conclusions reached in this analysis, and an alternative

108
hypothesis needs to be offered to supplant the concept of market efficiency, but given the
results of the preceding analysis, I have significant doubts pertaining to the of market
efficiency.
REFERENCES
Affleck-Graves, J., Hegde, S., & Miller, R. E. (1996). Conditional price trends in the
aftermarket for initial public offerings. Financial Management, 25(4), 25-40.
Alexander, S. S. (1961). Price movements in speculative markets: Trends or random

walks. Industrial Management Review, 2, 7-26.
Ang, J. S., & Zhang, S. (2004). An evaluation of testing procedures for long horizon
event studies. Review of Quantitative Finance and Accounting¸23, 251-274.
Bachelier, L. (1900/1964). Theory of Speculation. In Cootner, P. H. (Eds), The Random

Character of Stock Market Prices (pp. 17-78). Cambridge, MA: MIT Press
(1964).
Ball, R., & Brown, P. (1968). An empirical evaluation of accounting income numbers.
Journal of Accounting Research, 159-178.
Barber, B. M., & Lyon, J. D. (1997). Detecting long-run abnormal stock returns: The
empirical power and specification of test statistics. Journal of Financial
Economics, 43, 341-372.
Barron, D. P. (1982). A model of the demand for investment banking advising and
distribution service for new issues. The Journal of Finance, 37(4), 955-976.
Beatty, R. P., & Ritter, J. R. (1986). Investment banking, reputation, and the underpricing
of initial public offerings. Journal of Financial Economics, 15, 213-232.
Bernanke, B. S. (2007). The Level and Distribution of Economic Well-Being. Speech

Presented on February 6, 2007 to the Greater Omaha Chamber of Commerce,
Omaha, Nebraska. URL:
http://www.federalreserve.gov/newsevents/speech/bernanke20070206a.htm
Bernstein, P. L. (1998). Against the gods: The remarkable story of risk. John Wiley &
Sons, Inc.: New York, NY.
Bhabra, H. S., & Pettway, R. H. (2003). IPO prospectus information and subsequent
performance. The Financial Review, 38, 369-397.
Booth, J. R., & Smith, R. (1986). Capital raising, underwriting and the certification
hypothesis. Journal of Financial Economics, 15, 261-281.
Bradley, D. J., Jordan, B. D., & Ritter, J R. (2003). The quiet period goes out with a bang.
The Journal of Finance, 58(1), 1-36.
110
Bradley, D., Jordan, B., Ritter, J., & Wolf, J. (2004). The quiet period revisited. Journal
of Investment Management, 2(3), 1-11.
Brav, A., Geczy, C., & Gompers, P. A. (2000). Is the abnormal return following equity
issuances anomalous? Journal of Financial Economics, 56, 209-249.
Brown, S. J., & Warner, J. B. (1980). Measuring security price performance. Journal of
Financial Economics, 8, p. 205-258.
Brown, S. J., & Warner, J. B. (1985). Using daily stock returns: The case of event
studies. Journal of Financial Economics, 14(1), 3-31.
Brown, S. J., & Weinstein, M. (1985). Derived factors in event studies. Journal of
Financial Economics, 14(3), 491-5.
Campbell, J. Y., Lo, A. W., & MacKinlay, A. C. (1997). The econometrics of financial
markets. Princeton, NJ: Princeton University Press.
Carter, R. B., Dark, F. H., & Singh, A. K. (1998). Underwriter reputation, initial returns,
and the long-run performance of IPO stocks. The Journal of Finance, 53(1), 285-
311.
Cheng, W., Cheung, Y. L., & Po, K. (2004). A note on the intraday patterns of initial
public offerings: Evidence from Hong Kong. Journal of Business Finance &
Accounting, 31 (5-6), 837-860.
Cowles, A. (1933). Can stock market forecasters forecast? Econometrica, 1(3), 309-324.
Cowles, A., & Jones, H. E. (1937). Some A Posteriori probabilities in stock market
action. Econometrica, 5(3), 280-294.
Dalton, D. R., Certo, S. T., & Daily, C. M. (2003). Initial public offerings as a web of
conflicts of interests: An empirical assessment. Business Ethics Quarterly, 13(3),
289-314.
De Bondt, W. F. M., & Thaler, R. (1985). Does the stock market overreact? The Journal
of Finance, 40(3), 793-805.
Ellis, K., Michaely, R., & O’Hara, M. (1999). A guide to the initial public offering
process. Corporate Finance Review, 3, 14-18.
Fama, E. F. (1970). Efficient capital markets: A review of theory and empirical work.
Fama, E. F. (1976). Foundations of Finance: Portfolio Decisions and Securities Prices.

New York, NY: Basic Books, Inc.
111
Fama, E. F. (1991). Efficient capital markets: II. The Journal of Finance, 46(5), 1575-
1617.
Fama, E. F. (1998). Market efficiency, long-term returns, and behavior finance. Journal
of Financial Economics, 49, 285-306.
Fama, E. F, Fisher, L., Jensen, M. C., & Roll, R. (1969). The adjustment of stock prices
to new information. International Economic Review, 10(1), 1-21.
Fama, E. F., & French, K. R. (1992). The cross-section of expected stock returns. The
Journal of Finance, 57(2), p. 427-465.
Fama, E., & French K. (1993). Common risk factors in the returns on stocks and bonds.
Journal of Financial Economics. 33, 3-56.
Fama, E. F., & French K. R. (1996). The CAPM is wanted, dead or alive. The Journal of
Finance, 51(5), 1947-1958.
Field, L. C., & Hanka, G. (2001). The expiration of IPO share lockups. The Journal of
Finance, 56(2), p. 471-500.
French, C. W. (2003). The Treynor capital asset pricing model. Journal of Investment
Management, 1(2), 60-72.
Garfinkle, N., Malkiel, B. G., & Bontas, C. (2002). Effect of underpricing and lock-up
provisions in IPOs: Implications of a classic case of supply and demand. The
Journal of Portfolio Management, 28(3), 50-58.
Gompers, P. A., & Lerner, J. (2003). The really long-run performance of Initial Public
Offerings: The pre-NASDAQ evidence. The Journal of Finance, 57(4), 1355-
1392.
Helwege, J., & Liang, N. (2004). Initial public offerings in hot and cold markets. Journal
of Financial and Quantitative Analysis, 39(3), 541-569.
Henderson, G. V. (1990). Problems and solutions in conducting event studies. Journal of

Risk and Insurance, 57(2), 282-306.
Ibbotson, R. G. (1975). Price performance of common stock new issues. Journal of

Financial Economics (2), 235-272.
Ibbotson, R. G., & Jaffe, J. F. (1975). “Hot issue” markets. The Journal of Finance,
30(4), 1027-1042.
Ibbotson, R. G., Sindelar, J. L., & Ritter, J. R. (1994). The market’s problems with the
pricing of initial public offerings. Journal of Applied Corporate Finance, 7(1),
112
66-74.
Jegadeesh, N. (2000). Long-term performance of seasoned equity offerings: Benchmark

errors and biases in expectations. Financial Management, 29(3), 5-30.
Jegadeesh, N., & Titman, S. (1993). Returns to buying winners and selling losers:
Implications for stock market efficiency. The Journal of Finance, 48(1), 65-91.
Jegadeesh, N, Weinstein, M., & Welch, I. (1993). An empirical investigation of IPO

returns and subsequent equity offerings. Journal of Financial Economics, 34,
153-175.
Krigman, L., Shaw, W. H., & Womack, K. L. (1999). The persistence of IPO mispricing
and the predictive power of flipping. The Journal of Finance, 54(3), 1015-1044.
Kothari, S. P., Shanken, J., & Sloan, R. G (1995). Another look at the cross-section of
expected stock returns. The Journal of Finance, 50, 185-224.
Kothari, S. P., & Warner, J. B. (2006). Econometrics of event studies. Working Paper,
Tuck School of Business at Dartmouth.
Krigman, L., Shaw, W. H., & Womack, K. L. (1999). The persistence of IPO mispricing
and the predictive power of flipping. The Journal of Finance, 54(8), 1015-1044.
Leedy, P., and Ormrod, J. (2005). Practical research: Planning and design, 8th edition.
Upper Saddle, NJ: Pearson Prentice Hall.
Lintner, J. (1965). The valuation of risk assets and the selection of risky investments in
stock portfolio and capital budgets. Review of Economics & Statistics, 47(1), 13-
37.
Loughran, T., & Ritter, J. R. (1995). The new issues puzzle. The Journal of Finance,
50(1), 23-51.
Loughran, T., & Ritter, J. R. (2004). Why has IPO underpricing changed over time?
Financial Management, 33(3), 5-37.
Lowry, M., & Schwert, G. W. (2002). IPO market cycles: Bubbles or sequential learning?
Lyon, J. D., Barber, B. M., & Tsai, C. L. (1999). Improved methods for tests of long-run
abnormal stock returns. The Journal of Finance, 44(1), 165-201.
McDonald, J. G., & Fisher, A. K. (1972). New-issue stock price behavior. The Journal of
Finance, 27(1), 97-102.
113
Miller, R. E., & Reilly, F. K. (1987). An examination of mispricing, returns, and

uncertainty for initial public offerings. Financial Management, 33-38.
Perfect, S. B., & Peterson, D. R. (1997). Day-of-the-week effects in the long-run

performance of initial public offerings. The Financial Review, 32(1), 49-70.
Purnanandam, A. K., & Swaminathan, B. (2004). Are IPOs really underpriced? Review of
Financial Studies, 17(3), 811-848.
Reilly, F. K. (1977). New Issues Revisited. Financial Management, 6(4), 28-42.
Reilly, F. K., & Hatfield, K. (1969). Investor experience with new stock issues. Financial
Analysts Journal, 25, 73-80.
Ritter, J. R. (2003). Differences between European and American IPO markets. European
Ritter, J. R. (1984). The “hot issue” market of 1980. Journal of Business, 57(2), 215-240.
Ritter, J. R. (1988). Initial public offerings. Contemporary Finance Digest, 2(1), 5-30.
Ritter, J. R. (1991). The long-run performance of initial public offerings. The Journal of
Finance, 45(1), 3-27.
Ritter, J. R., & Welch, I. (2002). A review of IPO activity, pricing, and allocations. The
Journal of Finance, 57(4), 1795-1828.
Rock, K. (1986). Why are new issues are underpriced. Journal of Financial Economics,
15, 187-212.
Ross, S. (1977). Risk, Return, and Arbitrage. In Friend, I. (Eds.), Risk and Return in
Finance (pp. 198-218). Cambridge, MA: Ballinger Publishing (1977).
Roll, R. (1977). A critique of the asset pricing theory’s tests. Journal of Financial
Economics, 4, 129-176.
Rumrill, P. (2004). Non-manipulation quantitative designs. Work, 22, 255-260.
Russell Investments (2008). Russell 3000 Index: Fact Sheet. URL:

http://www.russell.com/indexes/documents/Fact_Sheets/3000.pdf. Accessed on
July 8, 2008.
Samuelson, P. (2000). Honoring founding fathers of modern finance economics.

Schaeffer, J. (2005). A link between IPO underperformance and long-term performance.

114
University of Chicago, Dissertation, 1-58.
Schultz, P. (2003). Pseudo market timing and the long-run underperformance of IPOs.
Sharpe, W. F. (1964). Capital asset prices: A theory of market equilibrium under

conditions of risk. The Journal of Finance, 19(3), 425-442.
Sharpe, W. F., Alexander, G. J., & Bailey, J. V. (1995). Investments, Sixth Edition.
Englewood Cliffs, NJ: Prentice Hall.
Simon, C. (1989). The effect of the 1933 securities act on investor information and the
performance of new issues. The American Economic Review, 79(3), 295-318.
Spiess, D. K., & Afflect-Graves, J. (1995). Underperformance in long-run stock returns

following seasoned equity offerings. Journal of Financial Economics, 38, 243-
267.
Tinic, S. M. (1988). Anatomy of initial public offerings of common stock. The Journal of
Finance, 43(4), 789-822.
Treynor, J. (1962). Toward a theory of market value of risky assets. Unpublished

manuscript.
Wiggenhorn, J., & Madura, J. (2005). Impact of liquidity and information on the
mispricing of newly public firms. Journal of Economics and Finance, 29(2), 203-
220.
APPENDIX A:
Power Analysis—1-Year Event Horizon—n=50
Power Analysis
Empirical Size - 1 Year Performance - n=50
MC Ind MC & Ind MC & BtoM MC MC & BM Ind
-.75 99% 99% 98% 97% 100% 100% 99%
-0.5 94% 97% 96% 89% 97% 97% 82%
-0.3 75% 82% 82% 72% 81% 84% 61%
-.20 46% 58% 56% 47% 69% 74% 53%
-.15 30% 37% 39% 30% 65% 63% 50%
-.10 16% 22% 20% 17% 54% 56% 46%
-.05 8% 11% 9% 6% 50% 49% 44%
-.01 6% 6% 4% 3% 46% 44% 47%
0 5% 4% 4% 3% 43% 43% 47%
+.01 4% 4% 6% 3% 42% 42% 48%
+.05 8% 5% 8% 8% 40% 33% 53%
+.10 19% 19% 17% 21% 38% 34% 63%
+.15 36% 42% 42% 36% 40% 41% 72%
+.20 51% 60% 57% 59% 51% 51% 79%
+0.3 78% 79% 82% 83% 67% 70% 89%
+0.5 95% 96% 96% 96% 91% 92% 96%
+.75 97% 97% 98% 98% 98% 97% 98%
Note. Power Analysis conducted on 180 samples of 50 randomly selected firms spanning the 18-
year period from 1985 to 2002. The first column of the Table provides the amount of simulated
abnormal performance (MC=Market Capitalization; Ind=Industry; BtoM=Book-to-Market
Ratio). Columns two through five give the results from the matched-firm approaches. Columns six
through eight convey the results of the power test ran on the portfolio-matched approaches.
116
APPENDIX B
Power Analysis
-.75 100% 100% 100% 100% 100% 100% 100%
-0.5 100% 100% 100% 100% 100% 100% 92%
-0.3 100% 100% 100% 100% 100% 100% 92%
-.20 92% 100% 100% 100% 92% 92% 92%
-.15 92% 92% 100% 100% 77% 92% 85%
-.10 85% 92% 92% 69% 85% 85% 92%
-.05 23% 54% 54% 23% 100% 92% 92%
-.01 0% 0% 0% 0% 92% 85% 69%
0 0% 0% 0% 0% 77% 77% 69%
+.01 8% 0% 0% 0% 69% 69% 69%
+.05 54% 46% 23% 69% 69% 62% 69%
+.10 92% 92% 100% 100% 92% 85% 92%
+.15 100% 100% 100% 100% 92% 85% 92%
+.20 100% 100% 100% 100% 77% 77% 92%
+0.3 100% 100% 100% 100% 92% 92% 100%
+0.5 100% 100% 100% 100% 100% 100% 100%
+.75 100% 100% 100% 100% 100% 100% 100%
117
APPENDIX C
Power Analyses—2-Year Event Horizon—n= 50
Power Analysis
Size Ind MC & Ind MC & BtoM MC MC & BM Ind
-.75 92% 96% 92% 88% 88% 93% 66%
-0.5 77% 78% 79% 74% 67% 71% 48%
-0.3 41% 52% 48% 40% 51% 54% 49%
-.20 18% 28% 28% 21% 45% 44% 50%
-.15 11% 14% 18% 12% 37% 38% 55%
-.10 7% 8% 10% 7% 32% 31% 56%
-.05 5% 4% 6% 4% 31% 29% 56%
-.01 3% 2% 4% 3% 32% 28% 59%
0 3% 2% 3% 3% 32% 28% 59%
+.01 3% 2% 3% 3% 31% 28% 61%
+.05 6% 6% 4% 5% 36% 29% 63%
+.10 11% 11% 9% 10% 37% 31% 66%
+.15 18% 17% 18% 22% 44% 38% 74%
+.20 27% 32% 31% 32% 52% 48% 79%
+0.3 49% 54% 56% 54% 67% 63% 86%
+0.5 76% 82% 83% 79% 86% 79% 91%
+.75 89% 91% 91% 91% 90% 90% 93%
118
APPENDIX D
Power Analyses—2-Year Event Horizon—n= 500
Power Analysis
-.75 100% 100% 100% 100% 100% 100% 92%
-0.5 100% 100% 100% 100% 85% 92% 92%
-0.3 100% 100% 100% 92% 69% 69% 85%
-.20 92% 92% 100% 77% 77% 77% 85%
-.15 69% 92% 77% 62% 92% 85% 85%
-.10 46% 54% 62% 31% 100% 85% 77%
-.05 0% 15% 8% 0% 92% 85% 85%
-.01 0% 0% 0% 8% 77% 77% 77%
0 0% 0% 0% 8% 69% 69% 85%
+.01 0% 0% 0% 8% 69% 69% 85%
+.05 8% 0% 8% 31% 62% 69% 85%
+.10 46% 69% 54% 77% 62% 69% 100%
+.15 85% 85% 85% 92% 85% 69% 92%
+.20 92% 92% 92% 92% 77% 77% 92%
+0.3 100% 100% 100% 100% 92% 85% 92%
+0.5 100% 100% 100% 100% 100% 100% 100%
+.75 100% 100% 100% 100% 100% 100% 100%
119
APPENDIX E
Power Analysis
Size Ind MC & Ind MC & BtoM MC MC & BM Ind
-.75 84% 84% 84% 78% 72% 74% 47%
-0.5 55% 64% 63% 56% 59% 56% 48%
-0.3 31% 31% 33% 21% 50% 39% 58%
-.20 13% 15% 16% 9% 39% 37% 62%
-.15 9% 9% 10% 5% 32% 32% 62%
-.10 6% 4% 7% 4% 32% 31% 66%
-.05 4% 2% 2% 1% 33% 29% 68%
-.01 3% 2% 3% 1% 34% 29% 68%
0 2% 2% 3% 1% 33% 29% 69%
+.01 2% 2% 3% 2% 34% 31% 69%
+.05 3% 3% 3% 4% 35% 34% 73%
+.10 6% 6% 7% 6% 38% 38% 78%
+.15 12% 14% 9% 8% 45% 41% 82%
+.20 17% 21% 17% 16% 49% 46% 88%
+0.3 29% 35% 33% 31% 64% 54% 92%
+0.5 61% 68% 70% 69% 81% 78% 94%
0.75 82% 83% 87% 83% 89% 87% 97%
120
APPENDIX F
Power Analysis
-.75 100% 100% 100% 100% 77% 85% 85%
-0.5 100% 100% 100% 100% 92% 85% 77%
-0.3 77% 92% 100% 62% 85% 69% 92%
-.20 62% 77% 77% 46% 92% 85% 92%
-.15 46% 46% 38% 31% 92% 92% 85%
-.10 31% 15% 23% 8% 92% 85% 92%
-.05 0% 8% 0% 0% 85% 85% 92%
-.01 0% 0% 0% 8% 77% 85% 92%
0 8% 0% 0% 8% 69% 85% 92%
+.01 8% 0% 0% 8% 69% 85% 85%
+.05 8% 0% 8% 23% 69% 77% 85%
+.10 23% 31% 23% 46% 77% 69% 92%
+.15 46% 46% 69% 85% 69% 69% 92%
+.20 85% 77% 85% 85% 77% 69% 100%
+0.3 100% 92% 92% 92% 77% 77% 100%
+0.5 100% 100% 100% 100% 100% 100% 100%
+.75 100% 100% 100% 100% 100% 100% 100%
121
APPENDIX G
Power Analysis
-.75 68% 68% 69% 65% 64% 62% 57%
-0.5 41% 48% 47% 38% 56% 53% 67%
-0.3 16% 19% 21% 14% 46% 36% 76%
-.20 11% 11% 11% 7% 43% 30% 77%
-.15 7% 7% 9% 6% 41% 27% 78%
-.10 4% 7% 8% 3% 42% 28% 81%
-.05 4% 5% 6% 2% 43% 34% 82%
-.01 4% 6% 4% 3% 44% 37% 84%
0 4% 4% 4% 3% 47% 37% 84%
+.01 4% 4% 4% 3% 47% 37% 84%
+.05 4% 6% 3% 3% 48% 41% 86%
+.10 6% 8% 6% 7% 51% 44% 88%
+.15 9% 11% 9% 8% 54% 48% 89%
+.20 12% 13% 16% 12% 59% 52% 89%
+0.3 23% 29% 27% 22% 66% 61% 94%
+0.5 41% 52% 56% 52% 78% 77% 96%
+0.75 72% 77% 76% 73% 88% 86% 97%
122
APPENDIX H
Power Analysis
-.75 100% 100% 100% 85% 92% 77% 85%
-0.5 92% 92% 85% 77% 92% 92% 85%
-0.3 62% 54% 77% 46% 85% 92% 100%
-.20 46% 38% 23% 31% 85% 77% 92%
-.15 23% 15% 15% 8% 92% 77% 92%
-.10 0% 0% 8% 0% 85% 77% 92%
-.05 0% 0% 0% 0% 85% 69% 92%
-.01 0% 0% 0% 0% 69% 62% 92%
0 0% 0% 0% 0% 69% 62% 92%
+.01 0% 0% 0% 0% 69% 62% 92%
+.05 8% 0% 0% 8% 69% 62% 92%
+.10 8% 23% 23% 23% 69% 62% 92%
+.15 31% 38% 23% 54% 69% 62% 100%
+.20 46% 62% 54% 62% 69% 77% 100%
+0.3 77% 85% 85% 100% 85% 77% 100%
+0.5 92% 92% 100% 100% 92% 92% 100%
+.75 100% 100% 100% 100% 92% 92% 100%
123
APPENDIX I
Average Initial Day IPO Performance by Year and Month
Average 1st Day IPO Return by Month & Year

T-
Year Year/Month Average % Return StDev Count Value
1997 Year 1.23% 10.92% 446 2.38
Jan -0.20% 8.76% 25 -0.11
Feb 0.76% 8.24% 43 0.61
Mar 1.12% 7.69% 27 0.76
Apr 5.33% 17.04% 20 1.40
May 1.20% 13.85% 27 0.45
Jun 0.58% 8.69% 49 0.47
Jul 1.84% 9.49% 43 1.27
Aug 3.03% 8.97% 33 1.94
Sep 2.26% 8.47% 34 1.55
Oct 1.11% 10.28% 53 0.78
Nov -1.50% 14.99% 63 -0.79
Dec 3.46% 11.10% 28 1.65
1998 Year 1.60% 10.27% 275 2.58
Jan 2.07% 11.35% 12 0.63
Feb -0.67% 5.93% 39 -0.70
Mar -0.19% 6.35% 28 -0.16
Apr 0.59% 8.95% 32 0.37
May -0.27% 6.71% 33 -0.23
Jun 2.83% 12.60% 48 1.56
Jul 4.08% 9.17% 35 2.63
Aug 7.71% 13.17% 16 2.34
Sep -4.23% 8.09% 2 -0.74
Oct -0.27% 1.09% 2 -0.35
Nov -1.07% 13.45% 13 -0.29
Dec 3.82% 17.77% 15 0.83
1999 Year 6.07% 23.78% 464 5.49
Jan 6.31% 14.36% 11 1.46
Feb 4.84% 21.34% 28 1.20
Mar 4.69% 23.07% 18 0.86
Apr 6.61% 17.38% 32 2.15
May 3.59% 15.94% 48 1.56
Jun 7.16% 23.50% 53 2.22
Jul 2.75% 15.50% 58 1.35
Aug 17.54% 41.50% 38 2.61
Sep 3.87% 27.42% 40 0.89
Oct 3.89% 24.97% 53 1.13
Nov 9.01% 23.08% 49 2.73
Dec 3.71% 21.68% 36 1.03
2000 Year 6.05% 20.62% 369 5.64
Jan 16.70% 22.68% 11 2.44
124
Feb 4.60% 21.53% 54 1.57

Mar 3.28% 20.81% 50 1.12
Apr 11.79% 22.83% 32 2.92
May 3.68% 16.18% 21 1.04
Jun 2.05% 15.78% 32 0.74
Jul 11.52% 21.93% 41 3.36
Aug 5.00% 25.05% 59 1.53
Sep 2.67% 18.48% 24 0.71
Oct 6.97% 10.93% 18 2.71
Nov 3.16% 13.44% 19 1.03
Dec 12.44% 17.51% 8 2.01
2001 Year 2.46% 9.57% 82 2.33
Jan 8.69% 21.63% 4 0.80
Feb 1.15% 6.86% 8 0.48
Mar -0.17% 8.02% 9 -0.06
Apr -0.75% 12.24% 2 -0.09
May 3.91% 12.51% 9 0.94
Jun -0.22% 8.05% 13 -0.10
Jul 4.69% 9.01% 7 1.38
Aug 3.47% 14.20% 4 0.49
Sep
Oct 0.13% 5.17% 7 0.07
Nov 5.18% 9.31% 10 1.76
Dec 3.26% 7.45% 9 1.31
2002 Year 0.01% 8.56% 70 0.01
Jan -1.54% 2.77% 3 -0.96
Feb 4.86% 11.25% 8 1.22
Mar -0.99% 3.52% 5 -0.63
Apr -1.89% 11.46% 7 -0.44
May 1.17% 6.00% 15 0.76
Jun -3.66% 8.49% 10 -1.36
Jul -3.30% 9.66% 6 -0.84
Aug
Sep
Oct 0.82% 9.83% 5 0.19
Nov 1.08% 7.42% 9 0.44
Dec 1.06% 10.38% 5 0.23
2003 Year 1.67% 10.59% 73 1.35
Jan
Feb -4.34% 4.80% 3 -1.56
Mar
Apr
May 3.26% 4.43% 2 1.04
Jun -6.37% 0.00% 1
Jul -1.15% 10.21% 8 -0.32
Aug 4.96% 3.72% 4 2.67
Sep 3.33% 6.29% 7 1.40
125
Oct 4.63% 9.33% 10 1.57

Nov -1.47% 8.33% 16 -0.71
Dec 3.55% 14.71% 22 1.13
2004 Year 2.24% 9.06% 179 3.30
Jan 6.65% 4.83% 2 1.95
Feb 1.85% 10.21% 17 0.75
Mar 1.91% 8.41% 10 0.72
Apr 6.26% 7.60% 9 2.47
May 4.63% 6.82% 11 2.25
Jun 2.59% 11.29% 23 1.10
Jul 2.76% 6.74% 21 1.88
Aug -2.61% 7.75% 13 -1.21
Sep -1.60% 9.46% 11 -0.56
Oct 3.02% 10.78% 22 1.31
Nov 3.68% 9.36% 14 1.47
Dec 1.74% 8.19% 26 1.08
2005 Year 1.82% 10.71% 185 2.31
Jan -2.77% 9.34% 8 -0.84
Feb 1.25% 5.96% 21 0.96
Mar -2.37% 17.41% 7 -0.36
Apr 0.60% 1.87% 7 0.85
May 0.85% 8.97% 13 0.34
Jun 1.49% 5.99% 23 1.19
Jul 2.56% 6.74% 15 1.47
Aug 5.87% 18.56% 30 1.73
Sep 0.92% 11.13% 16 0.33
Oct 1.29% 7.91% 11 0.54
Nov 1.98% 9.00% 18 0.93
Dec 1.31% 7.53% 16 0.69
Note: The proceeding table shows the average yearly and monthly performance of IPOs on their
first day of trade. The sample included firms that issued unseasoned equity shares from 1997-
2002, 2,143 observations could be included in the sample.

126
APPENDIX J
Yearly Average of Pre-Trade Performance 1997-2007
Averaged Yearly IPO Performance from 1997-2007

Test Ho </= 0, H1 > 0
Average Stand T- tcritical=1.796 (df

Year Return Deviation Observations value =11)
2007 0.12 0.06 12.00 6.52 Reject
2006 0.10 0.05 12.00 6.73 Reject
2005 0.10 0.05 12.00 6.73 Reject
2004 0.09 0.04 12.00 7.77 Reject
2003 0.07 0.07 12.00 3.70 Reject
2002 0.06 0.04 12.00 4.68 Reject
2001 0.08 0.06 12.00 4.29 Reject
2000 0.35 0.33 12.00 3.68 Reject
1999 0.05 0.07 12.00 2.47 Reject
1998 0.00 0.04 12.00 -0.06 Accept
1997 0.02 0.02 12.00 2.49 Reject
Note: The proceeding table illustrates the average yearly and monthly IPO performance
obtained by IPOs in their pre-IPO trading period. The sample period range from April 12, 1996
until January 29, 2008 and there were 1,876 observations.

127
APPENDIX K
Yearly Abnormal Performance at the Expiration of the Quiet Period
Yearly Quiet Period Expiration

Standard Sample T-
Year Performance Deviation Size Value
1985 -0.55% 7.42% 230 -1.12
1986 -0.07% 8.68% 445 -0.18
1987 -0.30% 15.01% 353 -0.37
1988 -1.47% 10.45% 148 -1.71
1989 1.13% 8.48% 136 1.56
1990 1.19% 12.85% 125 1.04
1991 1.11% 14.85% 275 1.24
1992 1.10% 10.09% 375 2.11
1993 ~2.00% ~10.00% 515 4.46
1994 0.88% 9.04% 328 1.76
1995 2.27% 10.40% 405 4.39
1996 2.25% 12.68% 521 4.05
1997 1.14% 11.32% 431 2.09
1998 3.80% 14.95% 271 4.18
1999 5.84% 22.54% 435 5.40
2000 3.40% 24.79% 349 2.57
2001 -2.40% 12.50% 77 -1.68
2002 0.99% 8.78% 66 0.91
Note: The proceeding table shows the 5-day abnormal performance surrounding the conclusion
of the quiet period. The sample consisted of IPOs that were issued from 1985-2002, which
generated 5,529 observations.

128
APPENDIX L
Abnormal Performance & Lockup Expiration
Yearly Lockup Expiration

Standard Sample T-
Year Performance Deviation Size Value
1985 -0.18% 8.61% 237 -0.33
1986 -0.40% 10.45% 445 -0.80
1987 -0.24% 12.47% 353 -0.36
1988 -0.32% 7.76% 148 -0.51
1989 -0.52% 12.34% 136 -0.49
1990 0.06% 9.71% 125 0.07
1991 -1.90% 13.50% 274 -2.33
1992 -0.19% 9.88% 375 -0.36
1993 -0.48% 9.64% 515 -1.12
1994 0.99% 15.75% 328 1.13
1995 -1.34% 12.82% 405 -2.10
1996 -6.17% 103.91% 521 -1.36
1997 -0.68% 12.36% 431 -1.14
1998 -0.88% 15.74% 271 -0.92
1999 -2.70% 18.56% 435 -3.04
2000 -3.65% 21.13% 349 -3.23
2001 -2.33% 11.26% 77 -1.82
2002 -0.70% 15.89% 65 -0.35
Note: The proceeding table shows the 5-day abnormal performance surrounding the conclusion
of the lock-up period. The sample consisted of IPOs that were issued from 1985-2002, which
generated 5,529 observations.

129
APPENDIX M
Popular Benchmarks used to Represent Normal Performance
Portfolio-Based Matched-Firm
• Portfolios Matched Based Upon: • Size
• Size • Size & Book-to-Market
• Size & Book-to-Market • Size & Industry
• Size & Industry
130
APPENDIX N
Methods used in Long-term Event Studies
T able of Research Based Specifically on Measuring Long-term IPO Performance

Author(s) and Date of Publication Subject Area Chosen Benchmark
Ritter, J. (1991) Performance CSRP v-w NASDAQ
CSRP v-w AMEX-NYSE
Matched Firms - industry & size
Smallest size decile - NYSE
Loughran, T ., and Marietta, J. (2005) T iming Listing of NYSE CSRP Equally Weighted
FF - T hree Factor Model
Gompers, P., and Lerner, J. (2003) Performance CSRP v-w
Portfolios - size & book-to-market
FF - T hree Factor Model
Bhabra, H. (2003), and Pettway, R. (2003) Performance Matched firm - Size
Matched firm - Size & Industry
Matched firm - Book-to-Market
Matched firm - Book-to-Market and Industry
Schultz, P. Performance CSRP v- and e-w
Carter, R., Dark, F., and Singh, A. (1998) Performance V-W CSRP (NYSE/AMEX/Nasdaq)
Brav, A., and Gompers, P. (1997) Performance S&P 500
NASDAQ v-w
NYSE/AMEX v-w
NYSE/AMEX e-w
Portfolios - size & book-to-market
FF - Industry Portfolios
Perfect, S., and Peterson, D. (1997) Performance Matched firm - industry & size
CURRICULUM VITAE
Zachary A. Smith
507 Winchester Road
Jacksonville, NC 28546
Phone: (910) 388-0573 Email: zacharyasmith@gmail.com

Cellular: (757) 553-1692
EDUCATION
Undergraduate: Old Dominion University, Norfolk, Virginia; B.S., 2003
Graduate: Capella University, Minneapolis, Minnesota; M.B.A., 2005

Walden University, Minneapolis, Minnesota; Ph.D., 2008
Dissertation: An Empirical Investigation of Initial Public Offering (IPO)
Price Performance
WORK EXPERIENCE:
Merrill Lynch Internship 07-04 through 10-04

Wachovia Securities Analyst (Trading) 02-04 through 04-04
Primerica Financial Services Internship 08-02 through 12-03
United States Marine Corps Analyst (Operations) 08-97 through 08-01
RESEARCH INTERESTS:
List: Asset Pricing Models, Event Studies, Market Efficiency, Heuristics & Biases,
Investor Decision-Making, Financial Market Anomalies
TEACHING INTERESTS:
List: Investments, Capital Markets, Financial Decision-Making, Personal Finance,

Behavioral Finance

Walden University

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Walden University

Uploaded by

Copyright:

Available Formats

Walden University

COLLEGE OF MANAGEMENT AND TECHNOLOGY

This is to certify that the doctoral dissertation by

has been found to be complete and satisfactory in all respects,

Dr. William H. Brent, Committee Member,

Dr. Robert T. Aubey, Committee Member,

Chief Academic Officer

Denise DeZolt, Ph.D.

An Empirical Investigation of Initial Public Offering (IPO) Price Performance

Zachary Alexander Smith

M.B.A., Capella University, 2005

Dissertation Submitted in Partial Fulfillment

abnormal price performance, then the identification of predictable abnormal price

portfolios and matched-firm methods of abnormal performance detection produced the

opportunities for investors to benefit from the IPO process.

Zachary Alexander Smith

M.B.A., Capella University, 2005

Dissertation Submitted in Partial Fulfillment

I dedicate this dissertation to my children—Cooper and Harris. Without your

process. Thank you and I love you.

Mohammad Sharifzadeh, CFA (Faculty Mentor/Dissertation Committee Chair), Dr.

suffering through this and my other academic adventures.

TABLE OF CONTENTS................................................................................................... iii

LIST OF TABLES............................................................................................................ vii

LIST OF FIGURES ......................................................................................................... viii

CHAPTER 1: INTRODUCTION TO THE STUDY...........................................................1

CHAPTER 2: LITERATURE REVIEW ...........................................................................23

CHAPTER 5: SUMMARY, CONCLUSION, AND RECOMMENDATIONS ...............83

Table 1. Specification analysis I ..................................................................................... 663

Figure 1. One-year power curve. ...................................................................................... 69

INTRODUCTION TO THE STUDY

Researchers have formed a substantial empirical base to support the hypotheses

methodological approaches used to identify abnormal performance and the prevalence of

abnormal performance associated with events occurring in response to the issuance of

Therefore, I will address them individually in the following paragraphs.

Researchers choose to use benchmarks or market proxies to model normal

reliable estimates of expected return. To construct a model of normal performance that

involved in a security’s price performance. Researchers have substituted models based

upon firm-specific characteristics to gauge normal performance (referred to in this

provided examples of biases that influence the results of short-term analyses.

used accurately identifies abnormal performance is a critical component of research

projects testing market efficiency.

After I select a well-specified method that can accurately identify abnormal

include (a) short-term outperformance, (b) longer-term underperformance, and (c)

biased benchmarks, empirical disagreement regarding the significance of the abnormal

mentioned. This combination is necessary to evaluate the accuracy of the conclusions

derived from the analyses, according to Lyon et al. (1999).

underperform standard benchmarks in studies of longer-term performance. Ritter (1991)

this is equivalent to an investor underperforming the benchmark by 9.13% a year. Third,

systematically; if I substantiate the systematic nature of these anomalous events, investors

Background of the Study

(NYSE), National Association of Securities Dealers Automated Quotation (NASDAQ),

standard benchmarks, obtain longer-term negative performances. For example, Ritter

would have obtained a -27.39% BHAR.

exhibit abnormal performances in a systematic manner, trending in one direction or

another (positive or negative performance). Bradley, Jordan, and Ritter (2003)

value of the offering at the IPO’s issuance.

Researchers have attempted to document the anomalous nature of IPO

aggregate cumulative returns and different portfolios constructions to attempt to

approximate normal performance. Researchers contaminate portfolios because they fall

address briefly in chapter 2. Researchers also disagree on the methodology of aggregation

begin to correct the preceding biases and potentially damaging methodological

constructions, I propose the following.