5
How information asymmetry between investors and issuers can affect
reputation of underwriters?
Author: Lixuan Long
Student number: 11086947
Thesis supervisor: Dr. Jan Lemmen
Finish date: August, 2016
UNIVERSITY OF AMSTERDAM ECONOMICS AND BUSINESS MSc Business Economics
2
PREFACE AND ACKNOWLEDGEMENTS
I have put in a lot of effort on this project. However, it would not have been possible without the kind support and help from my supervisor Jan Lemmen. I also would like to extend my sincere thanks to all of my other colleagues and friends who supported me in many ways.
I am highly indebted to Jan Lemmen and my friends for their guidance and constant supervision as well as for providing necessary information on the collection of IPO data and useful suggestions on the construction of my master thesis.
I would like to express my gratitude towards my parents for their encouragement and support which has helped me in completion of this master thesis.
My thanks and appreciations also go to my director in Huawei where I am working in an internship during my master thesis phase for his understanding and support to help me have time to complete the thesis.
ABSTRACT
Statement of Originality
This document is written by Student [fill out your Given name and your Surname] who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
5
ABSTRACT
Underwriters’ reputation plays an increasingly important role in the initial public offering (IPO) activities in the primary market. As a key distinction from previous models of underpricing in initial public offering, I use IPO underpricing as a sign of information asymmetry and develop a model to investigate the determinants of the underwriters’ decisions regarding their long-run reputation in the IPO market. Under both short-term and long-term information-asymmetry situations, I find that information asymmetry increase the incentive of underwriters to build up a positive reputation. This suggests that out of reputation concerns, underwriters tend to choose to protect their long-run reputation rather than pursuing for an immediate profit. Underwriters tend to build their reputation when investors, rather than issuers, face more adverse market environment if they have information advantage on the IPOs that they underwrite. And I also find that reputable underwriters are more associated with lower offer price, which indicates that they tend to sufer the risk of “leave the money on the table”.
Keywords:
Initial public offering; Under-pricing; Reputation; Information asymmetry; Self-selection,
2 TABLE OF CONTENTS
PREFACE AND ACKNOWLEDGEMENTS ...2
ABSTRACT ...2
LIST OF TABLES ...3
LIST OF FIGURES ...4
CHAPTER 1 Introduction ...5
CHAPTER 2 Literature review ...8
CHAPTER 3 Hypothesis and Methodology...11
CHAPTER 4 Data and Descriptive statistics ...15
4.1 Sample selection ...15
4.2 IPO characteristics: Summary statistics for different types of underwriters ...19
4.3 Short-run and long-run returns from 1990 to 2009...23
4.4 IPO “heat” ...27
4.5 Firm-level risk Beta ...29
4.6 Correlation between control variables in the model ...31
CHAPTER 5 Regression Result ...33
CHAPTER 6 Robustness Checks ...38
6.1 Robustness check using standard error specification...38
6.2 Robustness check using alternative definition of reputation ...39
CHAPTER 7 Conclusion and Limitation ...41
7.1 Conclusion ...41
7.2 Limitation ...41
7.3 Suggestions for future research ...42
REFERENCES ...43
3
LIST OF TABLES
Table 1 Summary statistics of the change of stocks price and market index over different time
horizons ...17
Table 2 League table for TOP 25 underwriters in Thomson One database ...20
Table 3 Number of IPOs, First-day Returns, and Long Run Performance, 1990 to 2009 ...24
Table 4 Number of IPOs, First-day Returns, and Long Run Performance by Underwriter Reputation ...25
Table 5 Estimate results of short-term regression ...34
Table 6 Estimate results of long-term regression ...36
Table 7 Robustness check for heteroscedasticity and cluster within groups ...38
4
LIST OF FIGURES
Figure 1 Important issue charateristics by underwriter reputation ...21
Figure 2 Summary statistics of independent variable by different types of underwriters ...22
Figure 3 The Number of foreign and domestic companies listed on CRSP from 1980-2014. ...28
Figure 4 Summary statistics of IPO heat by underwriter reputation ...28
Figure 5 Summary statistics of firm-level risk of issues ...30
Figure 6 Correlation between control variables ...31
Figure 7 The line graph of the annual number of IPOs, 1990-2009 .. ...47
Figure 8 The scatter graphs of IPO heat and industry IPO heat against the time serial 1990-2009 by underwriter reputation...48
5
CHAPTER 1 Introduction
The impact of underwriter reputation has been well established in the literature (Fang, 2005), and this impact concerns mainly the relationships between investment bank reputation and their price and quality of underwriting services. Since insider’s information sets an incentive to the intermediary’s underwriting decision, this kind of information asymmetry has a significant implication on underwriters’ existence in the primary market. So I put forward the following research question: How can information asymmetry between investors and issuers affect the reputation of underwriters or intermediaries when the underwriters are faced with both immediate profit and long-term objectives? Immediate profit could indicate the use of insider knowledge to profit in the present and long-term objectives could indicate the build-up of a good reputation for future potential income. And underwriters, as information producer s, face the choice between issuers and investors to build their reputation among.
Investment banks raise investment capital from investors on behalf of underlying corporations and governments that are issuing these securities (both equity and debt capital). In practice, an initial public offering (IPO) may simultaneously accomplish both information extraction from institutional investors and the information dissemination about the IPO firms from the underwriters to investors in the primary market. One of the fundamental problems with issuing equity or risky debt is the potential opportunism by insiders/shareholders with superior knowledge. Intermediaries like investment banks are found to be important in playing a certification role between outsiders (investors) and insiders (issuing companies) and maintaining the equity offering function in the IPO market (Puri, 1996). And investment banks’ specialization in the marketing and sales of securities often help to lower the issuers’ transactional costs of borrowing (Fang, 2005). The role of underwriters arises from the information asymmetry that typically exists in the transactions decisions where one party has more information than the other. The issuers only go through IPOs for once in the primary market but investment banks’ underwriting activities are
repeated for many times in the financial markets. Thus investment banks’ survival and future income are directly related to their reputation: although dishonesty may increase short-term profit, such profit will be earned at the cost of losing reputation and future income (Fang, 2005). Therefore, reputation is an important capital investment in the investment bank industry. It is common that underwriters with prominent reputations might decide to underwrite high-quality issues that pose little risk to their
reputation. In return, knowing investment banks’ reputation concerns, investors might ask for less return to compensate for the risk of trading against the better informed party such as issuers (Carter and
Manaster, 1990). Overall, previous literature suggests that investment banks’ underwriting decisions can imply their reputation concerns, and thus are informative of issue quality.
The purpose of this study is to provide empirical evidence on how the existing information
asymmetry concerning the IPO’s value influences the underwriters’ decision about building up reputation, thus investors or issuers could make their investing or money raising decisions based on the interpretation of underwriters’ reputation. This study is especially important for a financial crisis where the most severe
6
information asymmetry might exist. If the outcome shows that underwriters generally tend to provide a credible initial price with less volatility given an increased degree of information asymmetry in the IPO market, then on the one hand investors can use the investment banks’ past performance, as measured by examining the performance of offering firms they previously took public, to assess credibility and make their transaction decision accordingly; on the other hand issuers can choose preferable underwriters to to support the initial price since underwriters may price their issues differently when facing a dynamic trade-off between their short-run cost and long-run reputation capital cost under the asymmetric information environment.
The initial price set by underwriters illustrates the information about how underwriters balance their reputation among issuers and investors. On the one hand, if underwriters value the reputation among issuers more than among investors, they tend to provide the initial price of stocks higher than the true value and raise more money on behalf of issuers. However, underwriters would face the risk of losing the trust from investors or reducing the bidding demand from investors from the very beginning. On the other hand, in the reverse situation, underwriters value the reputation among investors more than issuers and thus set an initial price close to the true value of stocks. As a result, underwriters win the trust from investors by providing a credible lower price but loss their future business from issuers. So the initial price set by underwriters reflects the asymmetric information by taking the form of the price change from initial price of stocks to the true value of the stocks (this true value can be approximately represented by long-term price which is the core interest later in my empirical study). That is why I test the reputation of underwriters under their underwriting issues’ price performance or so call “information asymmetry”. The empirical relationship I plan to test can tell us how will the underwriters price the IPOs and how will this price transfers the information about the reputation building of underwriters. In other words, underwriters have to balance their reputation among issuers and investors when pricing the initial price and this kind of balance is what I want to test in this study.
Overall, the findings in this study suggest that the initial prices of stocks reflect the underwriters’ reputation concerns. Using the underpricing or long-term price change as the information asymmetry, I test the underwriters’ reputation under such information asymmetry and find that out of reputation concerns, underwriters tend to choose to protect their long-run reputation rather than pursuing for an immediate profit. In other words, underwriters value the reputation among investors more than among issuers. As the research results, the reputation of underwriters has a positive relationship with
information asymmetry, which indicates that increasing information asymmetry will lead to the increase in the reputation of underwriters. And underwriters tend to build their reputation when investors, rather than issuers, face more adverse market environment. So in my study I focus on the reputation of underwriters with respect to the investors.
The rest of the study is organized as follows, Chapter 2 reviews literature. Chapter 3 and 4 discuss the main methodology and data. Chapter 5 presents and discusses the main empirical findings. Chapter 6
7
presents the robustness checks that deal with the endogeneity in the issuer-underwriter matching system. Chapter 7 presents conclusions.
8
CHAPTER 2 Literature review
The correlation between investment bank reputation and the information asymmetry through
underwriting services is reviewed in this chapter. However, but the evidence on this relationship is mixed. . Several empirical studies have proved that underwriters play a certification role to support the initial price based on the inside information of company’s future prospective earnings. Empirical studies on the relationship between underwriter reputation and stock’s initial pricing show that IPO underpricing should decrease in underwriter reputation because IPO underpricing arises from information asymmetry which underwriter reputation helps to resolve. Chemmamur and Fulghieri (1994) propose that reputation acquisition enables underwriters to act as a producer of credible information to reduce the adverse impact of information asymmetry such as the moral hazard problem. Moral hazard occurs when the party with more information about its actions takes more risks because party with less information bears the cost of those risks. Later, in contrast with the certification hypothesis of Chemmamur and Fulghieri (1994), Chemmanur and Krishnan (2012) support the market power hypothesis about the economic role of underwriters but reject the certification hypothesis, indicating that underwriters are playing a financial role to obtain the highest possible valuation of the issues rather than to price the equity close to their intrinsic value. While the certification hypothesis asserts that high-reputation underwriters price equity in IPOs closer to the intrinsic value due to their concern for preserving their reputation in the IPO market, the market power hypothesis implies that high-reputation underwriters would price the equity higher and further away from its intrinsic value.
Since few previous works have documented how information asymmetry impacts the reputation that investment banks desire to maintain, I focus on this particular angle and analyze the implication of the underwriters’ reputation under the information asymmetry situation. Both the intermediary reputation’s role in the underpricing puzzle of equity primary market and underpricing theory “asymmetry information hypothesis” can help us understand and investigate deeper on the target relationship between underwriter reputation and information asymmetry.
A. Intermediary Reputation and security pricing
Before testing the relationship outlined in this study, I posit that initial security pricing is some kind of proxy for information asymmetry. Although prior works seem to support the negative relationship between underwrite reputation and issue-date underpricing, a careful review shows there actually is mixed evidence on the relationship between underwriter reputation and security pricing.
Several previous studies by Rock (1986) and Carter and Manaster (1990) suggest that prestigious underwriters tend to take public less risky offerings. With less risk, there is less incentive for underwriters to acquire information and fewer informed investors will be out there in the primary market.
Consequently, since IPO initial returns or yields are required by uninformed investors as compensation for the risk of trading against superior information, IPOs that have lower returns are usually associated with prestigious underwriters.
9
However, on the other hand, Tinic (1988) also supports another hypothesis that underpricing serves as insurance against legal liability and against the associated damages to the reputations of investment bankers. So Tinic predicts larger initial returns of IPO for prestigious underwriters. Other earlier papers such as Beatty and Welch (1996), Cooney et al. (2001), and Logue et al. (2002) find either the opposite or no relation at all. Beatty and Welch (1996) and Cooney et al. (2001) have contradicted their previous findings in the 1980s and stated that the higher underwriters’ reputation leads to a higher degree of underpricing in the 1990s. Cooney et al. (2000) suggest that the change in the relationship between underwriter reputation and underpricing is most likely due to the difference to a significant increase in the number of high-demand IPOs in which the offer price is set above the upper estimated offer price filed in the preliminary prospectus. Loughran and Ritter (2002) provide the empirical evidence of the higher underpricing associated with prestigious underwriters in the 1990s and internet bubble period, which is inconsistent with the previous certification argument however. They explain that over time, especially in the internet bubble period, prestigious underwriters relaxed their underwriting standards and took public an increasing number of very young, unprofitable companies. So the average underpricing on their deals increases both due to the shift into riskier deals and due to the increase in the indirect fees. So concluding the reputation effect on the degree of underpricing can go either way from theory, I need empirical testing to know which theory is correct.
B. Theory of underpricing – asymmetry information hypothesis
In the theory of underpricing, there are a few hypotheses explaining the observed underpricing in an IPO. Asymmetry information is one of the important hypotheses. Baron's (1982) model assumes that the investment bankers have more information about investors' demand for the securities than the issuers possess. Moreover, the investment bankers' reputation may also help in certifying the quality of the issues and then generate demand that may not otherwise exist. Baron demonstrates that the price discount is an increasing function of the issuer's uncertainty about the market demand for its securities. His model predicts larger average underpricing for IPOs that are subject to greater uncertainty about their market clearing prices. Rock (1986) presents a model showing that the underpricing of initial public offerings depends upon the existence of a group of investors whose information is superior to that of the firm as well as that of all other investors. So the offering firm must price the shares at the discount in order to guarantee that the uninformed investors purchase the issue. Beatty and Ritter (1986) extend Rock's (1986) model and shows that the expected underpricing is an increasing function of the uncertainty about the market-clearing price of an IPO, which is consistent with Baron's model as well.
Issuers can also convey inside information through underpricing. Allen and Faulhaber (1989) argue that the company with better profit expectation may try to draw attention to their better quality by fixing a lower placing price since only the “good” companies can recuperate the cost of underpricing through successive placement at more favorable prices, for instance through a secondary or seasoned equity offering (SEO), an issue of additional securities from an established company whose securities already trade in the secondary market.
10 C. Important independent variable - premium fee
In the underwriting process, economic rents in the form of premium fees are earned on reputation, which is necessary to provide a continued incentive for the reputable underwriters to provide high-quality underwriting services (Fang, 2005). Klein and Leffler (1981) argue that reputational capital reasoning is important by demonstrating that the non-salvageable investment is preserved by investors as a
commitment to produce quality. Extension of the Klein and Leffler “reputational signaling” to financial markets includes: Easterrook (1984) on the role of dividends in guaranteeing future financial performance, DeAngelo (1981) on the role of an auditor reputation in accounting statement certification, Wakeman (1981) on the reputation of bond rating agencies, and Mayers and Smiths (1982) on theuse of proportionate insurance to guarantee contractual performance.
Under the key assumption that the true stock quality is ex-ante unobservable, a spread is needed as a means of quality assurance. Such a spread ensures that the present value of future income is greater than the short-term profit from selling low-quality goods at high-quality prices. So the premium fee is needed to increase high-quality information production.
Empirical evidence on the relationship between underwriter reputation and compensation is sparse. Fang (2005) finds that reputable underwriters obtain lower initial yields and charge higher fees. He also suggests that underwriting fees are earned on the reputation, and thereby provide continued incentives for underwriters to maintain reputation. Christopher (1992) finds there is a negative relation between an underwriter reputation and fees in the IPO market. Livingston and Miller (2000) also document such a “reputation discount” in the bond underwriting market. These results are at odds with both intuition and the theoretical prediction that a fee premium is needed to induce high-quality underwriter.
Overall, based on above reviews about underwriters’ reputation and information asymmetry and also their relationship with underpricing of IPO respectively, especially the key research of Fang (2005), this study intends to empirically test whether underwriter reputation will increase in the information
asymmetry. If the answer is yes, this research’s result will be in contrast with Rock (1986) and Carter and Manaster (1990), which state that lower returns are usually associated with prestigious underwriters, but will be consistent with Tinic (1988) and the later arguments from Beatty and Welch (1996) and Cooney et al. (2001), which suggest larger initial returns of IPO for prestigious underwriters.
11
CHAPTER 3 Hypothesis and Methodology
This study builds upon and is a further extension of the model of Booth and Chua (1996) and Fang (2005) but it focuses on testing the effect of information asymmetry of IPOs on the underwriters’ most important source of future growth, their reputation. The researcher tests the following hypothesis, also previously described in the introduction:
(1) Information asymmetry has a positive influence on the underwriter reputation. Underwriters tend to utilize the advantage of inside information to build up their reputation and hold long term objectives above short term objectives in equilibrium.
(2) If insiders have more information than outsiders, then long-term interests of the intermediaries will outweigh their short-term interests.
For the unobserved dependent variable underwriter reputation (Reputation), several proxies for underwriter reputation have been developed in the IPO literature. Logue (1973) and Beatty and Ritter (1986) are among the first to develop a measure of underwriter reputation. The Carter-Manaster (1990) (CM) system uses underwriters’ relative placements in the stock offering “tombstone” announcements. The other two methods Johnson and Miller (1988) (JM) and Megginson and Weiss (1991) (MW) are also easy to construct. JM uses a modified form of CM method by classifying underwriters into one of the four categories. MW uses the relative market share of underwriters as a proxy for their reputation which has a high correlation with the CM ranking. Another reputation measurement put forward by Hayes (1971) classifies investment banks into four groups in descending order of prestige and originating experience: special bracket firms, major bracket firms, sub-majors, and others. Tinic (1988) simplifies Hayes’s model and develops a two-group classification: “ranked” and “non-ranked” investment bankers. Sub-major or better banks are designated as ‘ranked” banks and the remaining firms as “unranked” banks based on the assumption that the ranked investment bankers would have superior origination and investigation capabilities relative to their “non-ranked” competitors.
Hence, I choose to measure underwriter reputation based on the bank’s market share, because intuitively, market share captures the “brand name” and “goodwill” of an investment bank (Fang, 2005). To be specific, I expect to study the unit change of underwriters’ reputation ranking (or the quality of underwriter service) brought by a unit change of the degree of information asymmetry during the financial crisis.
Previous researches suggest that underwriter reputation signals the level of underlying risks of taking stocks public, or immediate aftermarket return. So I use initial return as the independent variable (see definition below). Besides this immediate aftermarket return on the first trading day, I will also consider the stock return over another three horizons (one week, half year and one year) since the information is relatively fully incorporated in the stock price over longer holding period (Loughran and Ritter, 1995).
Besides the immediate aftermarket return, alternative proxy for information asymmetry is price revision (though is more about private information) during the book building, because information
12
aggregation in the primary market is implemented in this process. After the enactment of the Securities Act of 1933, book building period between filing and offering dates may fully incorporate the information gap between issuers and investors. This price revision can be defined as the percentage change from the midpoint of the original filing range to the final IPO offer price. Wang and Chris (2007) suggest that reputable banks are more active in primary market information aggregation than non-reputable banks.
Reputation
1
2*underpricing issuecharacteristics IPOheat AssetTotal error The dependent variable is the reputation of an investment bank, and I use market share as acontinuous measure of reputation. Because prominent banks with larger profitability is more competitive and prestigious in the primary market, in my sample I assume that market share can capture the content of reputation with precision and it is constantly affected by the variable of interest. Underwriter’s earnings before interest and tax (EBIT) is used to represent the market share. In the robustness checks, I will use a similar reputation definition as used by Carter-Manaster (1990) to test the robustness of the main
regression result.
Underpricing (Initial return) is the percentage stock price change from offer price to the closing price at the end of the first day of trading. It is defined as [(MP - OP)/OP] x 100, where MP is the closing price of the first trading day and OP is the offer price. Long-run returns over one-week, half-year and one-year holding periods are used as an addition to underpricing to represent the information asymmetry. To adjust all the returns of the IPOs for overall market movements, Standard and Poor's Composite Index (S&P index) is selected as the proxy for the market.
In the above model setting, the endogeneity caused by an internal matching system of issuers and underwriters can be also reduced by controlling for issue and underwriters’ characteristics such as the total proceeds of the new issues, firm-level risks compared to the market, underwriters’ premium fees and underwriters’ total asset, etc. Important control variables’ definitions are shown below:
Ln_proceeds = Natural log of the total proceeds amount of the initial public offering (incl. overallotment sold) in the primary market
File amount =Total amount filed (USD Mil$) of the new issue in the primary market Best =Binary variable that takes a value of one for best-efforts IPO, zero for
firm-commitment IPOs
Beta =Beta is the firm’s market beta, indicating the overall volatility of a security’s return against the return of a relevant benchmark
Offer price =Offer price is the price at which publicly issued securities is purchased by investors, used as a proxy for uncertainty about stocks value (Booth and Chua, 1996)
13
Premium fees =Amount of underwriting fees (USD Mil$) as a percentage of the offering amount
Asset Total =Annual total asset of underwriters over the sample period 1990-2009 IPO heat =Total number of IPOs in the 12 calendar months immediately preceding the
new issue
Industry_IPO heat =The total number of IPOs in the same industry in the 12 calendar months preceding the new issue
In my model, issue characteristics include issue amount filed in the market, beta, premium fees, proceeds and offer price. File amount is the total amount filed (USD Mil$) of the new issue in the primary market. It can be used as a preferable proxy for the accurate issue size of the IPO. Ln_proceeds is the natural log of total proceeds of the new issue and can be used as an important control variable. Proceeds is the total amount of money that issuers raise from the initial public offering. Premium fees is the underwriting fees (USD Mil$) as a percentage of the offering amount and it is only a part of the total proceeds of the new issue. I expect positive signs for all these three issue characteristics. The relationship between these three issue characteristics could be tested below in the correlation chart of all the control variables.
The firm’s beta is the overall volatility of a security’s return against the return of a relevant
benchmark. As previously mentioned, here in the study beta is calculated by (return of equity – risk-free rate)/ (return of market index – risk-free rate). Because the underlying firms with higher risk may cause more information cost for banks and request for more certification role playing for underwriters, the reputation of underwriters is damaged. So I expect the coefficient to be negative.
The IPO heat and industry IPO heat measure the total amount of new issues each year from 1990 to2009 at the whole IPO market level and industry level respectively. And since “hotness” of IPOs would bring more competition in the underwriting market, I expect the signs are both negative in the estimation results
Previous studies suggest that offer price is an important proxy for uncertainty about stocks value (Booth and Chua, 1996). In my framework, offer price generally captures each stock’s information cost. Though the offer price is an important control variable, it contains too complicated and conflicting information such as economies of scale and scope, the issuer-underwriter relationships, etc. and has high correlation with other control variables (shown in the correlation chart in Fig.7) . I don’t include it in my regression but in the general statistic description session. Another proxy for the information cost is premium fees, which is the amount of underwriting fees (USD Mil$) as a percentage of the total offering amount.
I also want to consider two main endogenous problems that my model can incur. First, the issuer– underwriter matching is likely endogenous, because banks may have chosen (self-selected) to underwrite higher quality issues mainly out of reputation concerns. Thus, failing to control for this type of self-selection could lead to incorrect conclusions. This endogeneity has also been discussed in other literature.
14
For instance, Carter and Manaster (1990) note that reputable underwriters are associated with offerings with lower uncertainty. Beatty and Welch (1996) also observe that underwriters earn a return on their reputation capital by taking larger, less risky firms public. Second, since the financial intermediaries’ ability to acquire reputation potentially mitigates the moral hazard problem in information production, there will be reverse causality between underwriters’ reputation and information asymmetry.
15
CHAPTER 4 Data and Descriptive statistics
4.1 Sample selection
The data for this study consists of initial public offerings of equity which are almost drawn from the Thomson SDC database for the period from 1990 to 2009. Related issue information such as the issue date, the offer price, stock returns, the initial filing range, the number of shares sold, the identity of underwriters, underwriter’s compensation, the offering technique (best efforts or firm commitment), and whether the firm was backed by venture capital or not, etc. can be collected from Thompson SDC database and Ritter’s data. Underwriters’ financial attributes such as total asset, earnings before interest and tax can be collected from Compustat and CRSP databases. I used an information filter as described below:
(1) Restrict to the offerings taken public through best effort or firm commitment underwriting techniques (2) Exclude IPOs of financial institutions (SIC code 600-699), the real estate industry and American Depository Receipts (ADRs)
(3) Exclude private placement or “Direct” offerings in which no investment bank was employed (4) Exclude unit offerings or underwritings associated with rights issues and
The time frame of this study is from 1990 to 2009. During this testing period, there was an acknowledged technology bubble between 1997–2000 in the Internet and related sectors and then there were other anomalies including the real estate financial crisis around 2007-2008 , all of which provide a preferable market environment with high degree of asymmetric information to study. My sample is further restricted to companies taken public within North America (Canada, St. Pierre/Miquelon (France) and the United States of America. The sample IPO firms are listed in the New York Stock Exchange (NYSE), the American Stock Exchange (AMEX), or the National Association of Securities Dealers Automated Quotation System (NASDAQ). The final sample contains 776 IPOs which are drawn from 258 different (three-digit SIC code) industries.
When IPO is underwritten by multiple lead underwriters, unlike other literature who consider the average of the reputation of all involved banks, I focus on the first involved lead investment banks in the syndicate since there is little difference concerning the rank of underwriters using both methods (Wei and Chris, 2007). I then merge the issue information with underwriters’ characteristics (including
underwriters’ identification) extracted from CRSP database. This data merge is particularly important in my study because I focus on the matching between underwriter reputation and the information asymmetry associated with the issues they choose to undertake. Monthly market index (S&P 500) and its return, which is extracted from CRSP database, are included as a benchmark in the study. I assume implicitly that the market index serves as a reasonable measure of expected return for stocks in my sample. In my model, I divide my sample of underwriters into two types. Based on the underwriter’s average fees income (US$ Mil) during my sample period, banks which earn fees income above the average level
16
of the whole underwriters are grouped into “reputable underwriters”, other banks below the average are grouped into “non-reputable underwriters” (Hayles, 1971). When there is no fees income record for underwriters, the total proceeds amount (incl. overallotment sold) of underwriters is used as a classification criterion to categorize the remaining underwriters. As a result, there are 77,903
non-reputable underwriters and 25,865 non-reputable underwriters accounting for approximately 75% and 15% of the whole underwriters respectively in my sample.
17
Table 1 Summary statistics of the change of stocks price and market index over different time horizons
This table presents systematic differences in the two underwriter groups’ underwriting features concerning issues’ performance. It gives us the summary statistics of the % changes of issue price (market index as reference) underwritten by different types of underwriters: 1 day before the issue, 1 week after the issue, 1/2 year and 1 year. Since stocks’ price and their variation are a good illustration of information incorporation, this table shows the information asymmetry faced by reputable underwriters and non-reputable underwriters respectively over time. Underwriters are divided into two groups: reputable and non-reputable underwriters. The group of reputable underwriters include underwriters with fees income lager than the average level of the whole underwriter sample.
Reputable Non-reputable
Variable Obs Mean Std.Dev. Min Max Obs Mean Std.Dev. Min Max
Change_1WeekAft 307 56.17 86.85 1 245 151 49.99 68.83 1 242 ChangeS&P500_1WeekAft 307 30.94 55.55 1 282 151 56.01 83.08 1 285 Change_180dayAft 307 76.84 106.9 1 291 151 66.01 84.13 1 292 ChangeS&P500_180dayAft 307 53.65 86.78 1 280 151 56.63 85.32 1 283 Change_1yearAft 307 73.67 104.3 1 299 151 68.04 91.76 1 281 ChangeS&P500_1yearAft 307 48.18 76.08 1 288 151 62.05 91.78 1 281
Change _stock yesterday 307 1.163 1.758 1 24 151 2.497 6.110 1 44
ChangeS&P500 yesterday 307 51.82 92.97 1 356 151 111.3 122.6 1 350
Source: SDC and CRSP database
Change_1WeekAft stands for average stocks price %change one week after offer date. ChangeS&P500_1WeekAft stands for S&P500 % change one week after offer date.
Change_180dayAft stands for average stocks price %change half year after offer date. ChangeS&P500_180dayAft stands for S&P500 % change half year after offer date.
Change_1yearAft stands for average stocks price %change one year after offer date. ChangeS&P500_1yearAft stands for S&P500 % change one year after offer date.
Change _stock yesterday stands for average stocks price %change one day before offer date. ChangeS&P500 yesterday stands for S&P500 % change one day before offer date.
18
Table 1 presents descriptive statistics for the change of stocks price and market index over different time horizons for different types of underwriters respectively. Here I discuss % changes of issue price for four time horizons: 1 day before the issue, 1 week after the issue, 1/2 year after the issue, 1 year after the issue. And I define one day and one week to be short time horizons and half year and one year to be long time horizons to measure the price performance here. The asymmetric information is higher (lower) if the price change is larger (smaller).
Here I find that the asymmetric information with respect to reputable underwriters is higher for all horizons, which is consistent with the theory of Tinic (1988), Beatty and Welch (1996), Cooney et al. (2001) and Loughran and Ritter (2002). The table preliminarily shows dynamic information incorporation in the price of IPOs after going public or short term before going public. For stocks underwritten by reputable banks, their price change on average is approximately 25 % higher than the benchmark market index for all horizons. For instance, for one-week time horizon, stocks prices on average change 56.17% but S&P500 only changes 30.94%. However, the price change on average of stocks underwritten by non-reputable banks is higher than the benchmark market index though at a lower level compared to non-reputable underwriters for all horizons. The change of stocks price over one week is even less than the benchmark market index. For instance, for 180-days horizon, stocks taken public by non-reputable underwriters change on average 66.01% and S&P500 changes 56.63% while stocks taken public by the reputable counterparties change 76.84% and S&P500 changes 53.65%.
Overall, on average stocks underwritten by reputable underwriters change more than those underwritten by non-reputable underwriters for all horizons, meaning that reputable underwriters are associated with larger underpricing and issues which perform better. It is also the case in the long run that stocks underwritten by reputable underwriters perform better than stocks of non-reputable underwriters, which can be supported by the average price change over one year 73.67 % and 68.04% for these two types of underwriters respectively. The conclusion above is also true for both absolute price change and relative price change compared to the market index.
The average standard deviations of IPOs underwritten by non-reputable banks are almost the same with those of the benchmark market index for long time horizons. However, stocks underwritten by reputable banks are more volatile than the benchmark market index. This is consistent with the notion that stocks taken public by reputable underwriters incorporate information more fully than stocks taken public by non-reputable underwriters for long time horizons. Therefore, reputable underwriters are associated with more volatile issues, which is consistent with Loughran and Ritter (2002).
In sum, IPOs underwritten by reputable banks usually have higher risk and hence more volatility, and thus have higher risk premium relative to the market premium. By taking the above observations into consideration, the conclusion is consistent with the understanding that the stocks underwritten by
prestigious banks tend to perform better than the market compared to those underwritten by less prestigious banks.
19
Because IPO stocks underwritten by reputable underwriters have already incorporated relatively enough information from investors, which is supported by underwriters’ investigation and road show phase before officially going public, the average change of stocks price one day before offer day is smaller than the respective market index for both types of underwriters. And what is interesting here is that IPOs underwritten by reputable banks are more stable than those underwritten by non-reputable banks one day before going public using the benchmark market index as the reference. This means that reputable underwriters won’t change the price as much as non-reputable underwriters might in a short time before going public. This is in contrast with the phenomenon after going public where IPOs underwritten by reputable banks are more volatile than those underwritten by non-reputable banks using the benchmark market index as the reference. This is consistent with the theory of partial price
adjustment (Hanley 1993, Booth and Chua 1996) which states that in the pre-issue information-gathering process, truthful revelation of information through demand by regular investors is rewarded by increase in underpricing after going public.
4.2 IPO characteristics: Summary statistics for different types of underwriters Table 2 presents the league table of top 25 reputable underwriters during the period 1990 - 2009 from Thomson One database. This underwriter ranking table is in the order of three matrixes concerning firm’s performance in the primary market: total issue proceeds (incl. overallotment sold) in the primary market in millions of dollars , market share and the total number of issues they underwrite during the sample period. The top 10 underwriters take up roughly 72% of the total market share, indicating that the equity underwriting market is highly concentrated. This result is consistent with Fang (2005).
20
Table 2 League table for TOP 25 underwriters in Thomson One database
Table 2 presents the league table of top 25 reputable underwriters during the period 1990 – 2009 which is retrived from Thomson One database. Underwriters in my sample are ranked by their total issue proceeds (incl. overallotment sold) in the primary market in millions of dollars , market share and the total number of issues they underwrite during the sample period.
Rank Book Runner
Proceeds Amount + Overallotment Sold This Market (US$ Mil) Mkt. Share Number of Issues
1 Bank of America Merrill Lynch 6,594.15 15.9 43
2 Goldman Sachs & Co 5,565.80 13.4 30
3 Credit Suisse 4,197.46 10.1 36
4 Morgan Stanley 3,765.17 9.1 22
5 Citi 2,735.66 6.6 26
6 JP Morgan 1,951.91 4.7 32
7 CIBC World Markets Inc 1,536.84 3.7 16
8 Barclays 1,493.95 3.6 14
9 RBC Capital Markets 1,057.86 2.5 18
10 UBS 1,034.86 2.5 18
11 Deutsche Bank 937.78 2.3 11
12 BMO Capital Markets 393.91 0.9 5
13 Scotiabank 241.62 0.6 5
14 Piper Jaffray Cos 228.91 0.6 7
15 TD Securities Inc 224.07 0.5 2
16 William Blair & Co 210.03 0.5 7
17 Stifel/KBW 186.65 0.4 5
18 SunTrust Banks 183.81 0.4 3
19 Wells Fargo & Co 122.27 0.3 6
20 PGIM Inc 103.38 0.2 4
21 Cowen & Co 62.78 0.2 2
22 Peters & Co Ltd 60.58 0.1 1
23 Raymond James Financial Inc 58.00 0.1 11
24 Canaccord Genuity 57.41 0.1 16
25 Needham & Co LLC 51.84 0.1 2
Subtotal with Book Runner 41,576.68 100.0 749
Subtotal without Book Runner - 0.0 0
Industry Total 41,576.68 100.0 749
Source: Thomson SDC database
Figure 1 presents the descriptive statistics of issue characteristics by different types of underwriters. Reputable underwriters ,though as aforementioned take up only 15% of the total underwriter sample, have total issue proceeds (incl. overallotment sold) almost 7 times the amount of non-reputable underwriters and have fees income 1.5 times the amount of the non-reputable underwriters. However, the average offer price of the stocks underwritten by reputable underwriters is only half of the average offer price of those underwritten by non-reputable underwriters. Since offer price can be used as a proxy for uncertainty
21
about the equity value (Booth and Chua, 1996), the lower offer price for reputable underwriters is consistent with the notion that prestigious underwriters care more about the demand of investors than the demand of issuers to raise as much money as possible (procceeds of the issue) and are less worried about leaving money on the table. In other words, prestigious underwriters are more likely to play an
information certification role.in the primary market. This is also consistent with the proposition that reputable underwriters are more likely to choose issuers with less stock return volatility and thus avoid putting their reputation at risk.(Chitru, Gatchev, and Spindt, 2002).
Figure 1 Important issue charateristics by underwriter reputation Figure 1 presents the descriptive statistic of issue characteristics underwritten by different types of underwriters.It reveals several systematic differences in the two underwriter groups’ underwriting decisions. Underwriters are grouped into two groups: reputable and non-reputable underwriters. The group of reputable underwriters include underwriters with fees income lager than the average level of the whole underwriter sample. Proceeds Amt is the total proceeds of the issue from the IPO in millions of dollars; Fees is the underwriting premium fees (USD Mil$) as a percentage of the offering amount, used as the compensation for underwriters; Offer price is the price at which publicly issued securities is purchased by investors, used as a proxy for uncertainty about stock value; Number of issue is the total shares of underlying firms offered to the public in millions in the primary market.
Reputable
Variable Obs Mean Std.Dev. Min Max
Proceeds Amt 312 78.23 39.66 14.14 173.9 Fees 312 53.71 10.68 1 83 Offer price 307 17.78 22.15 1 101 Number of Issue 312 30.26 5.288 1 44 Non-reputable
Variable Obs Mean Std.Dev. Min Max
Proceeds Amt 186 11.9 14.66 0.08 112
Fees 186 34.24 23.91 1 82
Offer price 151 36.35 37.91 1 102
Number of Issue 186 2.597 2.147 1 14
Figure 2 presents the rough summary statistics of information asymmetry by underwriter reputation. Dummy variable Best is a binary variable that takes a value of one for best-efforts IPO, zero for firm-commitment IPOs. Since best-efforts issues are considered to be less well certified by underwriters, thus
22
containing higher risk than firm-commitment issues, information asymmetry is more severe using this offering technique than the firm-commitment technique. Overall, in the summary statistics of dummy variable Best, I find that for non-reputable underwriters, the mean of Best, though is almost the same with that of reputable underwriters, has higher standard deviation. This means that less reputatable
underwriters have larger variance and are less identical in choosing between these two offering
techniques within the group. The mean of Best* Underpricing is larger in the reputable group, suggesing that underpricing is more severe for best-efforts IPOs underwritten by reputable underwriters than the best-efforts IPOs underwritten by the non-reputable underwriters, which is consistent with Booth and Chua (1996).
I generate another variable PREV to represent the information asymmetry by calculating the deviation of offer price from the midpoint of the filing range. PREV = (Offer price – Midfile)/ Midfile, where Midfile represents the midpoint of the initial filing range. I use this variable to stand for pre-IPO price revisions (Hanleys, 1993) which reveal the information aggregation by underwriters during the book building process. Reputable underwriters have average PREV 14 but non-reputable underwriters only have PREV 6, which indicates that there is likely more active price discovery in the pre-market for reputable underwriters.
Figure 2 shows that reputable underwriters are associated with more severe information asymmetry since the mean of underpricing is larger for reputable underwriters (in the first row). The mean of VC* Underpricing is higher for reputable underwriters which can be explained by two elements. One element is that high-quality underwriters are more likely to take public issues backed by venture capital since reputable underwriters have more resource and financial ability to undertake the stock investigation and risk in undertaking the information certification role. The other is that underpricing itself is already more severe for reputable underwriters as mentioned above.
In summary, after considering all the factors above, this section concludes that reputable
underwriters, who rank in the top 10 in the league table, enjoy 72% of the total market share. Reputable underwriters are associated with lower offer price and larger initial return on the first trading day , which indicates that they tend to reduce the risk or uncertainty of initial public offering and issuing firms may leave much money on the table as a result. Information asymmetry is more severe for reputable
underwriters since the variable “underpricing” and price revision PREV are larger for them in the figure 2 , which is consistant with Beatty and Welch (1996),Kumar McGee and Womack (1998), Habib and Ljungqvist (2002), Cooney et al. (1999) and Wang and Yung (2007).
Figure 2 Summary statistics of independent variable by different types of underwriters Figure 2 presents the summary statistics of independent variable information asymmetry for different types of underwriters. This is a rough correlation between independent variable (information asymmetry ) and dependent variable (reputation of underwiters) rather than causal relation which will be tested strictly in this study using a well-designed model. Underpricing is measured as [(MP-OP)/OP]*100, where MP
23
is the aftermarket closing price of the first trading day recorded by Thomson SDC financial database and OP is the offer price of the issue; Best is a binary variable that takes a value of one for best-efforts IPO, zero for firm-commitment IPOs; VC is binary variable that is equaled to 1 if the issue is venture capital backed and zero otherwise; PREV is the offer price’s deviation from the midpoint of the filing range calculated as (Offer price – Midfile)/ Midfile, where Midfile represents the midpoint of the initial filing range. So the interaction variable Best*Underpricing indicates the underpricing degree for best-efforts IPOs and similarly VC*Underpricing indicates the underpricing degree for IPOs that are venture capital backed.
Reputable
Variable Obs Mean Std.Dev. Min Max
Underpricing 307 0.269 0.634 -0.99 2.355 Best 4867 0.998 0.0496 0 1 Best*Underpricing 4867 0.283 0.528 -0.99 2.355 VC* Underpricing 4867 0.151 0.499 -0.594 2.355 PREV 307 14.08 20.67 -0.987 100 Non-reputable
Variable Obs Mean Std.Dev. Min Max
Underpricing 151 0.150 1.144 -0.99 7.318
Best 151 0.967 0.180 0 1
Best*Underpricing 151 0.181 1.127 -0.99 7.318
VC*Underpricing 151 0.0192 0.236 -0.989 1.319
PREV 151 6.098 17.89 -0.989 101
4.3 Short-run and long-run returns from 1990 to 2009
Table 3 shows number of IPOs, first-day returns, and long run performance from 1990 to 2009. In the Table 3, period 1999-2000 has relatively highly positive average first-day return (64.5% ) and highly negative average 3-year buy-and –hold return (-53.2% ,-31.9%, -59.1%) for all long-term return styles (incl. IPOs, market adjusted and style-adjusted) in the full sample while in the other sub-periods 1990-1998 and 2001-2009, average first-day return and average 3-year buy-and- hold return are totally different from period 1999-2000 , where on average the short-term return only reaches 1/3 of the period 1999-2000 and the long-term returns are highly positive for IPOs unadjusted subsample ( or very small negative for market- or style-adjusted subsample) contrast to the negative sign of period 1999-2000.
24
Table 3 Number of IPOs, First-day Returns, and Long Run Performance, 1990 to 2009 Table 3 provides the number of IPOs, first-day returns, and long run performance, IPOs from 1990-2009. The first-day return is equally weighted average first-day return, which is measured from offer price to the first CRSP-listed closing price. EW average three-year buy-and-hold percentage returns (capital gains plus dividends) are calculated from the first closing market price to the earlier of the three-year
anniversary price, the delisting price, or December 31, 2015. Market-adjusted returns are calculated as the buy-and-hold return on an IPO minus the compounded daily return on the CRSP value-weighted index of Amex, Nasdaq, and NYSE firms. Style-adjusted buy-and-hold returns are calculated as the difference between the return on an IPO and a style-matched firm. For each IPO, a non-IPO matching firm that has been CRSP-listed for at least five years with the closest market capitalization (size) and book-to-market ratio as the IPO is used.
Average 3-year Buy-and-hold Return
Year Number of IPOs Average First-day Return IPOs Market-adjusted Style-adjusted 1990 110 10.8% 9.7% -35.9% -38.4% 1991 286 11.9% 31.2% -1.8% 5.8% 1992 412 10.3% 37.4% -0.2% 11.1% 1993 509 12.7% 44.5% -8.3% -8.8% 1994 403 9.8% 78.1% -5.7% -1.2% 1995 461 21.2% 28.9% -57.8% -24.6% 1996 677 17.2% 25.2% -56.8% 7.0% 1997 474 14.1% 58.3% -1.9% 21.9% 1998 281 21.9% 24.1% 6.4% -4.3% 1999 477 71.1% -47.8% -32.7% -60.8% 2000 381 56.3% -60.2% -30.9% -56.8% 2001 79 14.2% 17.8% 14.4% -28.1% 2002 66 9.1% 68.6% 39.0% -0.4% 2003 63 11.7% 34.0% -7.7% -11.2% 2004 173 12.3% 51.4% 6.9% -7.0% 2005 159 10.3% 14.6% 3.1% -2.5% 2006 157 12.1% -28.8% -11.1% -4.5% 2007 159 14.0% -16.5% -0.5% 0.5% 2008 21 5.7% 11.4% 8.1% 5.1%
25 2009 41 9.8% 37.0% -5.1% -18.3% 1990-1994 1,720 11.2% 46.2% -6.4% -1.7% 1995-1998 1,893 18.1% 34.1% -34.1% 1.3% 1999-2000 858 64.5% -53.2% -31.9% -59.1% 2001-2009 918 11.02% 21.06% 5.23% -7.38%
Source: Jay R. Ritter’s Table 19, updated table I of Ritter and Welch 2002 Journal of Finance.
Table 4 Number of IPOs, First-day Returns, and Long Run Performance by Underwriter Reputation
Table 4 presents the number of IPOs, first-day returns, and long run performance by different types of underwriters, IPOs from 1990-2009. Underwriters are divided into two groups: reputable and
non-reputable underwriters. The group of non-reputable underwriters includes underwriters with fees income lager than the average level of the whole underwriter sample. Further descriptions of how number of IPOs, first-day returns, long run performance, IPOs returns, market-adjusted returns and style-adjusted returns are defined can be found in the Table 3. Data are retrieved from Thomson Financial Securities Data and other sources, with corrections by Jay R. Ritter.
1990-1994
Segmented by Number of IPOs
Average First-day
Return IPOs Market-adjusted style-adjusted
Non-reputable 504 10.20% 34.03% -7.20% 0.30% Reputable 843 13.70% 48.90% -5.08% -2.70% 1995-1998 Number of IPOs Average First-day
Return IPOs Market-adjusted style-adjusted
Non-reputable 673 15.40% 28.02% -35.01% -14.70% Reputable 980 20.50% 35.00% -30.74% 5.60% 1999-2000 Number of IPOs Average First-day
Return IPOs Market-adjusted style-adjusted
Non-reputable 151 45.10% -31.05% -25.60% -40.02% Reputable 652 72.90% -68.34% -53.50% -59.10% 2001-2009 Number of IPOs Average First-day
Return IPOs Market-adjusted style-adjusted
Non-reputable 385 11.73% 18.70% 6.23% -9.60%
26
Period 1999-2000 is a special time subset because there is a larger volume of IPOs exchanging in the primary market (so-called “IPO heat”). Investors are active in the IPO market, however, the reverse relationship between first-day return and 3-year-buy-and -hold return indicates that this active participation in the IPO market comprises mostly of financial speculation. In other words, long term prices of stocks nearly get back to initial offer price, meaning that actually IPO is just an arbitrage tool for earning short term profit. The investors employ IPOs to speculate on the first-day trading, so long-term performance tends to rebound back eventually.
For IPOs subsets from the period 1990 -1994 and the period 1995-1998, an IPO’s average long-term return is larger than their average first-day return, meaning that the initial offer price or the product cost in the primary market is generally lower than the true value of the stocks. However, the negative average market-adjusted long-term return for the period 1990-1998 suggests that IPOs actually didn’t outperform the whole market in the long run. In order to consider the results more objectively, some numbers need to be interpreted carefully. In fact, long horizons would introduce their own problems, for example, Kothari and Warner (1997) brings into caution that long-term price may be extremely noisy and biased. So to overcome this problem, I focus the interest of this study on the trading within a 12 calendar months horizon after the initial offering. In addition, my model includes the total number of IPOs in 12 calendar months preceding the offer date as a proxy for the “heat” of the IPO market. Later subchapter 4.4 contains in-depth review and discussion about this IPO heat concept.
Table 4 presents the number of IPOs, first-day returns, and long run performance by different types of underwriters, IPOs from 1990-2009. This table illustrates that stocks underwritten by reputable underwriters outperform those underwritten by non-reputable underwriters no matter for first-day return or long-term holding returns, which is consistent with the conclusion in session 4.2. And sub-period 1999-2000 also stands out by showing larger positive average first-day returns and larger negative long-term returns, which is consistent with the notion mentioned above of the financial speculation. In addition, for both types of underwriters, in the long-term performance the market-adjusted and style-adjusted returns generally perform worse than unadjusted IPO returns for all periods except for 1999-2000. This means that in fact, initial public offerings don’t necessarily mean a better performance than the whole market or the same-styled firms in the market in the long run.
However, there are also other factors that may influence the efficiency of immediate aftermarket prices. First, underwriters engage in price support and the after-IPO analyst coverage conducted by influential (all stars) analysts during the first month or so (Aggarwal, 2000). This may cause the
truncation of the distribution of the returns and thus reduce the information incorporated in prices. Second, because of the frozen period about six months after IPO, IPOs are difficult to short sell in the immediate aftermarket (Loughran and Ritter, 1995) which further hampers the price discovery aftermarket. So if I want to seek a proper proxy for the true value of stocks containing relatively full information, it is important to use the price data for at least one month after the initial offering. In addition, as mentioned
27
before I supplement short-term changes of stock price with long-term changes of stock price as the information asymmetry to estimate the regression result.
As a conclusion, the time horizon is a very important factor since different durations can have different information bias. To balance these tradeoffs, I focus on initial returns in the first trading day as well as stock returns at the 12-months horizons, which in part serves as a robustness check. The number of IPOs may have an important implication for the performance of stock returns since the frequency of IPOs can approximately capture the factor of macro economics in the stocks primary market as indicated in previous literature such as Wang and Yung (2007). For example the period 1999-2000 performs differently from other periods because it is during the technology bubble, which makes the average first-day return and 3-year buy-and-hold much higher than normally observed during longer periods.
4.4 IPO “heat”
Previous studies of IPO such as Booth and Chua (1996) have demonstrated the important effect of clustering of issues in the primary market on the information cost and thus the underpricing of the initial offering. Because clustering of IPOs can lead to information spill-over, thus lower the cost of searching information about the new issue, it is very important to control for this element for the independent variable information asymmetry. Therefore, I include the intensity measure of IPO to represent the rate at which new issues come to the market 12 calendar months prior to a particular issue. The coefficient of this variable reflects the fact that underwriters’ market share (reputation) as dependent variable can be somehow affected by the average market “heat” of IPOs in the primary market. In my study, I define the intensity of IPO market from two perspectives: the number of listed companies and the number of initial public offering.
28
Figure 3 The Number of foreign and domestic companies listed on CRSP from 1980-2014.
This graph plots the number of listed companies (both foreign and domestic) on CRSP over time period 1980-2014. Note: Operating companies only (i.e., no limited partnerships, closed-end funds, REITs, ETFs) listed on Nasdaq, NYSE, and Amex (now NYSE MKT). CRSP is the University of Chicago’s Center for Research in Security Prices. Quarter-end numbers are plotted, 1980-2015. Data are from Jay R. Ritter’s IPO data collection.
Figure 3 shows how the number of listed companies on CRSP spreads over 1980-2014. CRSP is the University of Chicago’s Center for Research in Security Prices. The number of foreign and domestic companies listed on CRSP initially reaches the peak of 8500 during the time period1997-1998 and then gradually descends. The number of listed companies reaches the lowest level of 4000 around 2013. If there are more listed companies in the market, there is a larger amount of stocks traded on the open market, which would lead to a more active and liquid equity exchange. So this measure can be used as a proxy for the macroeconomic sentiment in the stock primary market.
Figure 4 Summary statistics of IPO heat by underwriter reputation
Figure 4 presents the summary statistics of IPO heat and industry IPO heat by underwriter reputation. IPO heat is the total number of IPOs in the 12 calendar months immediately preceding the new issue. Industry IPO heat is the total number of IPOs within the same industry in the 12 calendar months preceding the new issue. Underwriters are divided into two groups: reputable and non-reputable underwriters. The
29
group of reputable underwriters includes underwriters with fees income lager than the average level of the whole underwriter sample.
IPO heat
Mean Std.Dev. Freq.
Non-reputable 381.9 214.0 151
Reputable 271.1 136.5 4867
Total 274.4 140.7 5018
Industry IPO heat
Mean Std.Dev. Freq.
Non-reputable 374.8 139.7 151
Reputable 289.5 111.9 4867
Total 292.1 113.8 5018
The total number of IPOs is another measurement for the intensity of IPOs which I include in the model as a control variable since it can capture temporal “hotness” of IPO market. Figure 4 presents the summary statistics of IPO heat and industry IPO heat by underwriter reputation in my sample period 1990-2009. In the appendix I further present the scatter graphs of IPO heat and industry IPO heat against the time serial 1990-2009 by underwriter reputation.These two figures together illustrate that for IPO heat, reputable underwriters experience on average a lower level of IPO heat in the primary market than non-reputable underwriters during my sample period 1990-2009. For industry IPO heat, non-reputable
underwriters also on average are less affected by the “hotness” of IPOs at the industry level.
The line graph in the appendix shows the frequency of IPOs on CRSP every year from 1990 to 2009. This line graph of the annual total number of IPOs has a low level in the second half of my sample time period. The number of IPOs drops steeply in 2002 which may likely be caused by the internet bubble and subsequent crash of the US stock market. Based on the important implication of the number of IPOs, I define the control variable “IPOheat” as the number of IPOs during the 12 calendar months prior to the new issue to account for the IPO clustering effect. In order to analyse the industry ( four-digit SIC code) levelinformational externalities, I also define an industry IPO “heat” to capture the within-industry potential information spill-over ( Booth and Chua, 1996).
4.5 Firm-level risk Beta
Firm-level risk beta is the equity ‘s market risk, an important issue characteristic. In order to control for the systematic differences of issue charateristics when different underwriter group is used, I include beta in my model equation (Fang, 2005). I define Beta = ( return on equity –risk-free rate)/(return on market index – risk-free rate), where return on equity is the monthly return on each IPO stocks in my sample and return on index is the monthly return on S&P500 index which is retrieved from CRSP
database . The Risk-free return rate is one-month treasury bill rate, implying for the required return asked by no-risk-seeking investors. As they are all ratios, outliers should be of concern. So I winsorize both return variables.
30
In the appendix, Figure 9 exhibits the distribution of monthly return on equity and monthly return on market index respectively using my equity sample. Returns on equity are positive and concentrates between 0 and 2% while returns on market index mostly allocate equally around zero. The preliminary evidence in Figure 5 shows that the mean beta is negative, suggesting that sample issues are inversely correlated to the market systematic risk. Normally, individual stock in the market has a positive
relationship with the whole market because of the systematic risk. Securities in the market have common movement responding to the swings of the whole market. However, this seemingly surprising inverse movement of issues in the primary market could occur even when both the benchmark index and the stocks under consideration have positive returns. Because It is possible that lower positive returns on the market index could coincide with higher positive returns on the stocks, or vice versa, the relationship between the stocks and market index in such a case will be negative.It can be explained at first by the summary statistics of the monthly return on S&P index in Figure5. Monthly returns on S&P500 index have almost half of its distribution blow zero and the mean return is nearly zero, but the returns on equity all lie above zero averaging approximately 1.47% which is far bigger than the mean of corresponding return on S&P500 index, 0.0262%.
By definition, beta is the part of the asset’s statistical variance that cannot be removed by
diversification and usually is used to measure the individual risk relative to the benchmark index. The statistical description of beta suggests that beta is approximately symmetrically distributed around zero over the full sample with a larger standard deviation, 4.55, than that of the returns on S&P index 0.184. This indicates that, compared to the market systematic movement, new issues in my sample reveal a more volitile return risk during the sample period 1990-2009.
Figure 5 Summary statistics of firm-level risk of issues
Figure 5 presents the summary statistics of firm-level risk of the issues in my study sample. Beta, as the market risk of issues, should be included as an important control variable since it associates with independent variable information asymmetry in my model. The return on index is monthly return on S&P500 index in the sample period 1990-2009. The return on equity is monthly return on the new issues in the sample period 1990-2009. Risk-free rate is one-month treasury bill rate, standing for the required return asked by no-risk-seeking investors. Beta is the firm-level risk of issuing firms, indicating the relative risk of issues to the whole market risk.
Market risk characteristics of sample issues
Variable Obs Mean Std.Dev. Min Max
Return on index 1172 0.0262% 0.184 -0.728% 0.509%
Return on equity 1135 1.47% 3.618 0.1% 15.6%
Risk free 1172 0.356% 0.134 0 0.69%
