UNDERWRITER PRESTIGE AND IPO UNDERPRICING
DURING THE FINANCIAL CRISIS
Master Thesis of Floor Pennings
January 15, 2016 Name: Floor Pennings Student number: 6065937
Supervisor: Dhr. dr. J.E. Ligterink
Master Business Economics Specialisation: Finance
Statement of Originality
This document is written by Student Floor Pennings, 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 content.
UNDERWRITER PRESTIGE AND IPO UNDERPRICING
DURING THE FINANCIAL CRISIS
Floor Pennings
Abstract
IPOs have been underpriced, on average, by more than 20% in the last few decades, with peaks larger than 50% during the bubble period of 1999-2000. In this research, I examine the relationship between underwriter prestige and IPO underpricing during financially difficult times. I argue that a positive relationship remains between IPO underpricing and prestige underwriters when the CM proxy, JM proxy, or the MW proxy is used as a measure for underwriter prestige. Average underpricing declined during the financial crisis, as did the number of IPOs recorded during the crisis.
TABLE OF CONTENTS
I. INTRODUCTION ... 5
II. LITERATURE REVIEW ... 6
IPO process ... 6
IPO underpricing ... 8
Reputation of Lead Underwriter ... 10
Measurements of Prestige ... 11
Prestigious underwriters during recession ... 13
III. METHODOLOGY AND HYPOTHESES ... 14
Hypotheses ... 14 Methodology ... 15 IV. DATA ... 18 V. DESCRIPTIVE STATISTICS ... 20 VI. RESULTS ... 30 VII. CONCLUSION ... 34 VIII. REFERENCES ... 35
I. INTRODUCTION
Initial Public Offerings (IPOs) have been underpriced, on average, by more than 20% in the last few decades (Liu & Ritter, 2011), with extremely high levels during the dot-com bubble of 73% in 1999 and 58% in 2000 (Ljungqvist & Wilhelm, 2003). Previous research has been conducted on the reason for underpricing IPOs. The problem with underpricing is that issuing firms do not obtain the amount of money related to the underpricing of the shares. When the initial return is positive, when underpricing exists, the return will go to the investors and not to the company. In the first instance, it is not beneficial for the issuing firm to have high IPO underpricing.
The underwriter does have the opportunity to allocate shares among the different investors. The agency problem explains the conflict of interest between the underwriter and the issuing company. If the underwriter wants to satisfy the investors, it is beneficial to have underpriced equity. Rock (1986) suggested that uninformed investors are compensated for the risk of trading against superior information.
In this research, I would like to investigate the relationship between IPO underpricing and the reputation of the underwriter. Existing literature is not consistent about this relationship. Literature until the 1990s examines a strong negative relationship between IPO underpricing and underwriter prestige (Carter, Dark, & Singh, 1998; Carter & Manaster, 1990; Jonhson & Miller, 1988; Megginson & Weiss, 1991). However, in more recent studies, a reverse of direction was argued (Beatty & Welch, 1996; Loughran & Ritter, 2004). This shift can be explained by the analyst lust theory. Loughran and Ritter (2004) argued that analyst coverage becomes more important over time. However, not much research has been conducted on the effect of the financial crisis or other periods of distress. It is interesting to investigate whether there is a relationship between IPO underpricing and underwriter prestige, in order to be able to predict this relationship in the future. By investigating this relationship, adds transparency to the market of underwriting IPOs.
This research examines the differences in three different measures from Carter and Manaster (1990), Jonhson and Miller (1988), and Megginson and Weiss (1991). The Megginson – Weiss proxy remains significant when evaluated simultaneously with the Jonson-Miller and Megginson-Weiss measures, and acquired the highest explanatory power. However, according to the Carter-Manaster proxy, which was used most commonly in previous literature, in 692 IPOs out of 854 IPOs was a prestige underwriter involved. The minority of the IPOs was underwritten by a non-prestige underwriter. It is also remarkable
that 65.23% of the IPO’s is underwritten by one of the seven biggest underwriters in the market, which suggests an oligopoly market structure. New data of companies that went public was manually obtained, in order to give new insights in the relationship between prestige underwriters and IPO underpricing.
The aim of the research is to investigate what the relationship is between underwriter prestige and IPO underpricing during the financial crisis. The research question, therefore, will be as follows: Is the level of IPO underpricing related to the reputation of underwriters, and what is the influence of the financial crisis? This question will be answered by empirical research on IPOs in the period from 2005 until 2013, issued in the United States.
The thesis is structured as follows. The second chapter will discuss theories and existing empirical research on underpricing and the role of the underwriters in the IPO process. The third chapter will introduce the hypotheses based on the discussed literature and the method used to empirically examine the research question. The fourth chapter will present the data and the sources used to test the hypothesis. The fifth chapter will describe and explain the outcomes of the methods used. In the last chapter, a brief summary and conclusion of the empirical research conducted will be presented.
II. LITERATURE REVIEW
In this chapter, existing literature will be discussed. First, the process of going public will be described, followed by relevant theories and empirical studies concerning underpricing of IPOs during the previous centuries. The function of the lead underwriter in the IPO process, the underwriter’s reputation, and the effect of environmental fluctuations, like booming periods and distressing periods, will be explicitly pointed out.
IPO process
At a certain point in time, a company can decide to go public in the form of an initial public offering. This is the first time that a company sells its shares to public investors and, subsequently, trades on the stock market. Such an event is always intense and has large consequence concerning reporting. Firms have to publish the information about their operations, which previously had been kept private.
There are several reasons for a company to go public. An IPO can be used to raise capital for new investments and expand the company’s current activities. However, an IPO can also improve the reputation of a company and ensure that company is known in a broader
market. Additionally, the firm gains the advantage of stock that presents its value and can be used for acquisitions or compensations to employees. Other reasons include attracting talented employees, providing liquidity for shareholders, and diversifying and reducing investor holdings. Brau and Fawcett (2006) found that the primary motivation for going public is to facilitate acquisitions. However, the main downside of going public is the obligation to disclose financial information about the firm. This disclosure gives the competition more knowledge about the company, and the firm’s competitive position might weaken. Moreover, going public is quite an expensive event, including direct costs of the IPO and all other costs of the reporting requirements.
The process of going public involves some difficulties, consisting of poor quality of public information. It is difficult for the issuer to know which investors may be interested and to what extent they are interested. On the other hand, investors also do not know much prior to the offering. The expertise and experience of an investment bank can bring these two parties together and can solve some of the information problems. The investment bank performs as the underwriter (intermediary) in conducting the IPO and assures that the IPO will be properly managed and successfully supported during the whole journey of going public.
Choosing an underwriter is one of the first serious tasks that has to be accomplished. In general, not just one investment bank is responsible for the underwriting process. One or more lead-underwriters and one or more co-managers are involved in an IPO. In a survey among 336 chief financial officers (CFOs), Brau and Fawcett (2006) reported that the top three criteria in selecting a lead IPO underwriter are the underwriter’s overall reputation and status, the quality and reputation of the research department/analyst, and the underwriter’s industry expertise and connections.
After choosing an underwriter, the preparation of the prospectus and related due diligence begins. The prospectus is a legal document, required by and filed with the Securities and Exchange Commission (SEC). It contains all the information needed for an investor regarding the IPO. After these steps are accomplished, the phase of approaching the market can start. The pre-IPO research report has to be published first. The investment bank analysts prepare the research report. When the research report is released and the IPO is disclosed, the investment bank has to investigate the degree of interest in the shares that the issuer will offer the market. The bank will simply talk with selected and trusted investors to gain a general idea of the feelings in the market about the offer. This phase is called “pilot fishing”. Based on these first soundings, a price range is set. At this point, the “roadshow” can start. The
roadshow is designed to acquire further interest in the issue. Usually, the lead underwriter takes care of hosting meetings to introduce the issuer’s management to selected institutional investors, portfolio managers, analysts, and securities sales personnel. This is an important phase in the IPO underwriting process and can be influenced by the choice of the underwriter. During the roadshow period, non-binding bids are requested, which is called “bookbuilding”. Once the book is closed, the price is set and the IPO will take place on a specified date. IPO underpricing
In this section, the existing literature on IPO underpricing will be described. Different theories explain the various interests behind underpricing of an IPO and the role of the underwriters. Furthermore, empirical evidence will be discussed and a link will be made between the existing literature and this thesis.
IPO underpricing is a common financial phenomenon, calculated as first-day return of issued shares. In particular, the shares are underpriced in relation to the market value of the company’s assets. After one day of trading, it is possible to observe how the market valuates the shares. There are several theories that try to explain underpricing. The first and most common theory is asymmetric information, which means that one party has more or better information than others, which can be observed by all parties
One asymmetric model is the Winner’s Curse, introduced by Rock (1986). This theory is based on the assumption that some investors have more information than other parties. When the underwriter is unable to attract enough informed investors, the uninformed investors have to be attracted by lowering the offering price, in order to compensate for their lack of information. If the underwriter has the ability to attract enough informed investors, it is not dependent on uninformed investors, and therefore, underpricing will be less excessive.
A second theory is the control theory, which contains the agency problem. When a firm goes public, there is a separation of ownership and control. Brennan and Franks (1997) argued that underpricing is used to ensure oversubscription in the share allocation process to allow owners to discriminate against large applicants and in favour of small applicants. The shares sold after the IPO are consistent with the pre-IPO investors who wish to avoid the costs of underpricing. These sales are substantially derived from insiders who are not directors of the company. These authors interpret this as evidence that directors derive private benefits of control that are not available to non-directors.
Another theory is the dynamic strategy theory. Underpricing is a cost for the issuing firm’s insiders to convince investors to collect information about the firm so that the true
value of the stock will be established after the IPO. The second issue can be done at the true value and, therefore, benefits from cashing in the secondary offering (Welch, 1989). The large increase in stock price at the first trading day creates the interest of analysts and media. Analysts will write more recommendations and reports for the IPO. This high attention creates higher demand and also a higher offer at the secondary public offer.
The fourth theory explains underpricing as the risk underwriters face if the entire issue does not sell out. The underwriter will be left with the remaining shares (Beatty & Ritter, 1985). These remaining shares have to sell at a loss. To control for this risk, underwriters may underprice the offer price on purpose.
Chen, Ho, and Weng (2013) investigated the relationship between the underwriter and subsequent lending. They explained how firms could derive value from the IPO underwriting relationship with the investment bank by their lending supports. The underwriting bank is more likely to provide the issuer with future compensations than banks with no further relationship. This would suggest that investment bankers with the ability to supply the issuer with additional services underprice IPOs more.
Underwriters also contribute their experience and reputation to the issuing equity. It is important to keep a good reputation in order to achieve successful IPOs. Carter and Manaster (1990) showed that the higher the reputation of the underwriter, the lower the underpricing of the IPO. Their research is consistent with Rock’s (1986) theory, which explains this negative relationship. Their theory explains that underpricing of an IPO compensates the uninformed investors in the market. Carter and Manaster (1990) extended this theory by suggesting that a greater proportion of informed underwriters cause a greater equilibrium price run-up.
Davidson et al. (2006) argued that the number of co-managers in an IPO is positively related to IPO placement risk. An IPO issuer hires more co-managers when placement risk is higher and implies that co-managers help reduce this risk. In line with Carter and Manaster (1990), the less risk, the less incentive to acquire information, which results in fewer informed investors and, therefore, more underpricing. Corwin and Schultz (2005) found evidence, in their research, to say that IPOs underwritten with many co-managers are less underpriced. Co-managers tell the issuer of the possibility of obtaining a higher price if the offering price is set to low.
Reputation of Lead Underwriter
Existing literature examines the effects of underwriter reputation on the performance of IPOs (Beatty & Ritter, 1986; Carter, Dark & Sigh, 1998; Carter & Manaster, 1990; Logue, 1993; Maksimovic & Unal, 1993; Titman & Trueman, 1986).
Carter and Manaster (1990) were among the first to examine the relationship between IPOs and prestigious underwriters. They found that prestigious underwriters are associated with IPOs that have less underpricing. Consistent with the earlier explained theory of Rock (1986), Carter and Manaster (1990) suggested that prestigious underwriters are associated with lower risk offerings. Less risk results in fewer incentives for investors to acquire information. The portion of informed investors decreases and these IPOs seem to have lower returns since more informed investor capital requires higher returns. Johnson and Miller (1988) also used Rock’s (1986) theory in order to explain the level of underpricing of IPOs. The level of underwriter prestige determines the expected level of informed investors and, therefore, the degree of underpricing.
Megginson and Weiss (1991) approached this topic in a different manor. They examined whether the presence of venture capitalists (VC) can certify the offer price. They argued VC-backed firms were able to attract higher quality underwriters than non-VC-backed IPOs. By choosing higher quality underwriters, VC-backed firms are able to reduce the asymmetric information between the issuing firm/investors and underwriters. Therefore, VC-backed IPOs are less underpriced.
Since the 1980s, movements can be seen in IPO underpricing. Loughran and Ritter (2003) argued that underpricing varies depending on environmental influences. Until the 1990s, the relationship between IPOs underwriting and prestigious underwriters had been explained by the winner’s curse problem and the dynamic information acquisition. Beatty and Welch (1996) described how this strong negative relationship was reversed in the early 1990s. Loughran and Ritter (2003) explained this reverse in relationship by assuming that underwriters receive commission in return for leaving money on the table. However, this might not have happened to the same degree as before the internet bubble period, unless there was a supply shift. This shift can be explained by the analyst lust theory. Loughran and Ritter (2003) argued that analyst coverage become more important over time. Dunbar (2000) presented evidence that underwriters subsequently increased their IPO market shares during 1984 to1994 if they were ranked highly in the Institutional Investor rankings.
Liu and Ritter’s (2011) method supplies deeper insights into the relationship between the reputation of the lead underwriter and IPO underpricing based on the analyst lust theory.
They questioned why it is still possible to leave large amounts of money on the table, even though the market of underwriters appears to be a perfect competitive market: Many competing underwriters in the market and no obvious large barriers to entry the market exist. The theory is based on the issuing firms’ desire for research coverage by influential analysts. Brau and Fawcett (2006) reported the criteria CFOs used in selecting a lead IPO underwriter. In addition to the reputation of the underwriter, the quality and reputation of the research department/analyst was also important. All-star analysts are the top three analysts in each sector. These all-star analysts do form series of local oligopolies. Empirically, Liu and Ritter (2011) found that IPOs are more underpriced when they have coverage from an all-star analyst and also when underwriters have better quality of more industry expertise.
Measurements of Prestige
Carter, Dark, and Singh (1998) investigated the relationship between underwriter reputation and underpricing in the short- and long-term. In their study, they examined three existing measures of underwriter prestige to the initial returns of Jonhson and Miller (1988), Carter and Manaster (1990), and Megginson and Weiss (1991). Consistent with prior studies, they found a negative relationship. IPOs managed by more reputable underwriters were associated with less short-term underpricing. The proxy that serves best to control for underwriter prestige is the reputation measure of Carter and Manaster (1990). Carter, Dark, and Singh (1998) provided a list of investment bankers, updated with Carter-Manaster ranks.
Carter and Manaster (1990) used underwriters’ relative placements in stock offering “tombstone” announcements. They suggested that the investment banking industry is subjected to a hierarchy. This hierarchy is reflected in the tombstone announcements, which is a listing of pending public security offerings. These tombstone announcements present the lead and lead underwriter(s) in the top section. Below this section, other sections of co-underwriters are presented. The sections are divided by most prestigious co-underwriters to next most prestigious underwriters and, and so on. Based on the sections of announcements where the investment banks appear, ranks are assigned from nine (most prestigious) to zero (least prestigious). This method resulted in a ranking where the most prestigious investment banks, ranked nine, were never dominated in the announcements, and investment banks with a rank of zero were never placed above any other investment bank. The ranking scale was established by comparing tombstone announcements listed in the Investment Dealer’s Digest.
Jay Ritter updated the rankings for the period 1992-20111, based on the measurement method of Carter and Manaster (1990) and Carter, Dark, and Singh (1998).
Johnson and Miller (1988) used two types of methodologies to define prestige underwriters: Binary measurement and a four-point ranking scale. The first method is based on the prestige classification system suggested by Hayes (1971). This classification system is similar to Carter and Manaster’s (1990) measurement system, discussed earlier. Tombstones in the financial sections of newspapers define which bankers are prestigious. However, Johnson and Miller (1988) had some concerns with this method. They mentioned that it is not clear how many tiers of investment bankers should be considered as prestigious. Carter and Manaster (1990) did not use the assigned ranking from 0 to 9. They split their sample based upon the median rank of investment banker prestige. The second measurement of prestige is the four-point ranking scale. Based upon three different cut-off points, level of prestige is defined. Each underwriter identified as a member of the “bulge bracket” is assigned to the most prestige group, ranked 3. Those considered “major bulge bracket bankers” are ranked 2. Those considered “sub-major bankers” are ranked 1, and all other underwriters are ranked 0. The bulge bracket comprises a list of the world’s largest investment banks that have occupied a leading role in high-quality security underwriting (Carter, Dark, & Singh, 1998). This list includes investment banks like Bank of America Merrill Lynch, UBS, Citigroup, and Credit Suisse. A list of the different groups assigned by Johnson and Miller (1988) is provided in their paper.
Megginson and Weiss (1991) introduced another way of measuring quality of the underwriter. They used the relative market share of the underwriters as a proxy for their reputations. When the issuing firm had more than one lead underwriter, the average of the lead underwriters market share was used as a measure of prestige. Megginson and Weiss (1991) argued that the greater the average market share of the lead underwriter, the higher the level of prestige. However, the Carter-Manaster (CM) proxy and the Megginson and Weiss (MW) proxy differ in such a way that there is no security an investment bank that takes the lead in a large number of offers also appears high in the tombstone advertisements. The CM reputation measure is more concerned with the company itself (Carter, Dark, & Singh, 1998). It is based on the assertion that, for example, Bank of America, Merrill Lunch, and J.P. Morgan will not allow their names to appear in the lower brackets of the tombstone.
Prestigious underwriters during recession
Until the 1990s, Beatty and Welch (1996), Cooney, Singh, Carter, and Dark (2001) and others documented the negative relationship between IPO underpricing and the reputation of underwriters, although after the 1990s, this relationship seemed to reverse itself (Loughran & Ritter, 2004). Higher underpricing is associated with more prestigious underwriters, which increased even more during the dot-com bubble. This relationship is inconsistent with previous findings that underwriters attempt to maximize issuer IPO proceeds. Loughran and Ritter (2004) explained this phenomenon with the changing issuer objective function hypothesis, where IPO proceeds are not the only caring factor of issuers. Loughran and Ritter (2004) also cared about proceeds from future sales and side payments from underwriters to the people who will choose the lead underwriter. In some periods, the underwriter put less weight on IPO proceeds and more weight on the proceeds from future sales and side payments. Therefore, the changing issuer objective function hypothesis provides a theory for the choice of underwriter.
Loughran and Ritter (2004) hypothesized that issuing firms choose their lead underwriter on the basis of analyst coverage. The greater visibility of analyst recommendations because of internet and television issuing, firms started to find analyst coverage more important. During times where analyst coverage is even more important, firms are willing to pay more for it. This phenomenon became more important during the dot-com bubble, and issuers accepted a low price in return for star analyst coverage. The recommendations started to reach a broader public and, therefore, became more important. In the 1980s, firms competed for IPO underwriting mandates more on the basis of implied valuations and less on the basis of analyst coverage. In the 1990s, it changed. Issuing companies started to find analyst coverage more important. Underwriters with star analysts could win a mandate without committing to high IPO valuations. Liu and Ritter (2011) subsequently used this analyst lust explanation to explain the reversed relationship by mentioning that only a limited number of underwriters can provide analyst coverage for a given firm.
The theories mentioned above give explanations for the positive relationship between underwriter prestige and underpricing after the 1990s and during the internet bubble. Not much research has been conducted on this relationship during the financial crisis or difficult financial periods in general. The analyst lust theory explains the switch from a negative to positive relationship is due to increasing importunateness of star analysts’ influence. Loughran and Ritter (2004) suggested that most of the shifts occurred via changes in
underwriters’ behaviour, rather than shifting market shares. However, during the financial crisis of 2008, it might have become less acceptable to leave that much money on the table.
III. METHODOLOGY AND HYPOTHESES
In this chapter, the hypotheses will be provided, derived from the described existing literature. Furthermore, the methodology will be specified to examine the research question. Ordinary least squares (OLS) regressions were used to empirically test the hypotheses. This analysis will be used to test the relationship between IPO underpricing and underwriter prestige. Hypotheses
Existing literature presented evidence for a negative relationship between underwriter prestige and underpricing until the 1990s. IPOs managed by more reputable underwriters are associated with less short-term underpricing (e.g., Beatty & Welch, 1996; Cooney, Singh, Carter, & Dark, 2001). However, after the 1990s, the relationship seems to be reversed. Research conducted from the 1990s until the internet bubble occurred. This thesis examines the relationship between IPO underpricing and underwriter prestige in the period around the financial crisis of 2008. Because evidence suggests that most of the shifts occurred via changes in the underwriters’ behavior, the positive relationship is assumed to remain positive during the time. Since different proxies are used as measures for underwriter prestige in existing literature, different measures are taken into account in the first hypothesis:
Hypothesis 1: IPO underpricing is higher for prestige underwriters than for non-prestige underwriters, when the CM proxy, JM proxy, or the MW proxy is used as the measure for underwriter prestige.
Carter, Dark, and Singh (1998) investigated the relationship between underwriter reputation and underpricing and examined three existing measures of underwriter prestige, and believed that the CM proxy served the best to control for underwriter prestige. This thesis also examines the differences in measures of underwriter prestige, as well as which proxy has the highest explanatory power. The CM proxy is assumed to perform best, because the measure is based on tombstone announcements instead of relative market share, leading to Hypothesis 2:
Hypothesis 2: The CM proxy, used as measure for underwriter prestige, has relatively more explanatory power than the other two proxies.
During the dot-com bubble, underpricing reached extremely high levels. Ljungqvist and Wilhelm (2003) found that first-day returns were, on average, 73% in 1999 and 58% in 2000. Not much research has been conducted on the relationship during the financial crisis or difficult financial periods. Assuming that the switch in previous literature can be explained by a change in behaviour of individual underwriters, and that it became less acceptable to leave that much money on the table, underpricing of IPOs is expected to decrease but to maintain a positive relationship with underwriter prestige during the investigated period between 2005 and 2014, which leads to Hypotheses to 3 and 4:
Hypothesis 3: Underpricing of IPOs decreased during the financial crisis.
Hypothesis 4: The relationship between IPO underpricing and underwriter prestige was positive before, during, and after the financial crisis.
Methodology
For testing the hypotheses, several OLS-regressions were carried out. These regressions estimate the relationship between underwriter prestige and IPO underpricing and include control variables as well. In order to include time fixed effects and industry fixed effects, the used data was conducted as panel data.
The dependent variable IPO underpricing is the initial-day return of the traded stock. The equation used to measure this variable is as follows:
!"#$% − !"# !"#$!% =!"#$% !"# !"#$%&' !"#$% − !""#$ !"#$% !""#$ !"#$%
The independent variable is the reputation of the underwriter (prestige). Carter, Dark, and Singh (1998) examined the relationship between underwriter reputation and underpricing of IPO based on three different measures for reputation. As the first measure, the CM proxy was used. This CM proxy is based on the position of the investment banker in tombstone announcements mentioned by Carter and Manaster (1998). The sections of announcements where the investment banker appears are ranked from zero (least prestigious) to nine (most
prestigious). For this research, underwriters are divided into two groups: Prestige underwriters and non-prestige underwriters. Investment bankers ranked eight or higher are placed in the group prestige underwriters, and investment bankers ranked below eight are assigned to non-prestige underwriters. The second measure of underwriter reputation is the JM proxy, which is a measure based on a four-point ranking scale. A list of the different groups is provided by Johnson and Miller (1988). The third measure contains the MW proxy, based on relative market share of the underwriters.
The provided regressions also incorporate different control variables. These control variables are assumed to determine the independent variable underpricing. Existing literature describes the influences of these control variables.
The first control variable included is the amount of money that is raised from the IPO (proceeds). The number of shares offered times the offering price calculates proceeds. Large-size offerings are considered riskier and, therefore, expected to have higher underpricing of IPOs.
The total assets of a company indicate the size of the company. The size of the company is also considered part of the fundamental risk (Ellul & Pagano, 2006). The underwriter needs to be compensated for the extra risk, which can be accomplished in the form of IPO underpricing. The natural logarithm of assets is used in the regressions. This is common in empirical IPO literature.
The age of the issuing company has a negative impact on the level of underpricing; there is more underpricing of young firms than of old firms (Loughran & Ritter, 2004; Megginson & Weiss, 1991). The age of the issuing company is calculated by the natural logarithm of the age of a firm, plus one year. In this case, a company that started its business in the same year as the year that the company went public is assigned to have an age of one year.
Industry dummies are also added to the regressions. The Thomson Reuters proprietary macro-level industry classifications are used. These classifications are based on SIC codes, NAIC codes, and overall company business descriptions. In total, there are 12 industries and each issuing firm is assigned to one of these macro industries. Some industries might influence the underpricing of the IPO. DuCharme, Rajgopal, and Sefcik (2001) provided evidence in which internet IPOs are relatively more underpriced than non-internet IPOs. Loughran and Ritter (2004) incorporated a separate dummy for technological and internet firms in their regression as a control variable.
Venture capitalists typically invest in young companies with high growth opportunities. Lee and Wahal (2004) found a positive effect for venture capital backed IPOs on underpricing. Consistent with Gompers (1996), they found that higher underpricing results in larger outflows of capital into venture capital funds in the future. They also attributed their findings to the existence of the bubble period of 1999-2000. However, Megginson and Weiss (1991) found a negative relationship in their dataset between 1983 and 1987.
Another control variable included is the number of lead underwriters. The number of lead underwriters has increased over the years. Loughran and Ritter (2004) argued that co-managers are included in a syndicate to provide research coverage to the issuers. Corwin and Schultz (2005) agreed and empirically found that underwriting spreads increase as the number of co-managers increases, at least for small IPOs.
The first regression was used to answer the first hypothesis and was used for different measures of proxies. This regression also provided an answer to the second hypothesis:
Regression 1: First-day return = β! + β!Prestige proxy + β!Ln(Assets)+ β!Number of lead underwriters + β!Ln(Age) + β!VC-backed dummy + β! Tech/internet dummy + ∑ d! Year dummy + ∑ c! Industry dummy
This third hypothesis was tested by the following regression. The proxy used as measure for prestige was the MW proxy. The proxy was the relative market share as indication of prestige. Different crisis-dummies were used for different periods. The pre-crises dummy was 1 for the period between 2005 and 1007, the crisis dummy was 1 for the period between 2008 and 2011 and both dummies equal 0 for the after-crisis period. The different dummies were used to test whether the different periods influences, the level of underpricing. The regression below was also used to answer the fourth hypothesis:
Regression 2: First-day return = β! + β!Prestige-CM-proxy + β!Ln(Assets)+ β!Number of lead underwriters + β!Ln(Age) + β!VC-backed dummy + β! Tech/internet dummy + β!Crisis dummy + β! After-crisis dummy
IV. DATA
All the data containing IPOs was obtained from the Thompson One database. The database gave access to all fundamental data on IPOs needed for this study. The used data contains data from all IPOs over the period 2005 to 2013 in the U.S. market. This period was chosen because it contains three interesting periods: The three years before the financial crisis (2005– 2007), the three years during the financial crisis (2008–2010), and the three years after the crisis (2011-2013). This dataset offers basic details about the companies and details about IPOs, including name, nation, industry, issue date, offer price, closing price of the first trading day, stock prices after the first trading day, proceeds, date founded, amount of shares offered at the IPO, vc-backed indication, underwriters, and number of lead-managers/co-managers.
Other fundamental data was obtained from the Compustat database in WRDS. This data contains information about total assets, net sales, shares outstanding, book value, market value, and market return.
The data of the underwriters’ ranks was obtained from the Jay Ritter Database. This ranking method was initiated by Carter and Manaster (1990) and supplemented by Ritter
(2011). The reputation of the underwriters is divided into nine groups of reputation, based on
the packing order in tombstone advertisements, where 1 is low reputation and 9 is a high reputation of the investment bank. Based on previous research of Carter and Manaster (1990) and the distribution of the ranked underwriters, I chose to assign underwriters ranked 8 or above as prestigious. Non-prestigious underwriters were ranked below 8. The database was provided by the website of Jay Ritter2. The average rank between 2005 and 2013 was used. If there was more than one lead underwriter involved in the IPO, the underwriter with the highest rank was used.
Megginson and Weiss (1991) used the relative market share as an indication of prestige. To calculate the relative market share, the original dataset of 3,472 observations was used, including all IPOs during the period 2005 to 2013. The names of all lead underwriters at an IPO were presented in the dataset, as obtained from Thomson One. However, I was interested in the relative market share per lead underwriter separately. Therefore, the relative share of every IPO was calculated per lead underwriter. Proceeds were divided by the number of lead underwriters and assigned to each of them. Summing these assigned shares gave the relative market share for every lead underwriter. Counting the amount gave the number of
IPOs on which the underwriter was active. However, this information was only recorded for 2,330 observations in the dataset.
The Johnson-Miller proxy was introduced by Johnson and Miller (1988) and based on the packing order in tombstone advertisements as well. They divided investment bankers into prestigious and non-prestigious sub-groups, using three different cut-off points. Bankers in the bulge bracket are ranked 3, bankers in the major bracket are ranked 2, bankers in the sub-major bracket are ranked 1, and all other bankers are ranked 0. A list is provided by Johnson and Miller (1988); however, this list was too dated to use in this study. This study incorporates the bracket definitions as defined by Johnson and Miller (1988). Bankers assigned 9 are assumed to be in the bulge bracket, bankers assigned 8 are assumed to be in the major bracket, bankers assigned 7 are assumed in the sub-major bracket, and all other bankers are within the groups below 7.
Appendix A contains a list of underwriters and rankings of the various measures of underwriter prestige.
Jay Ritter also provides a list of firms’ ages on his website3. This list, combined with the data on age, were available via Thomson One database and, supplemented by self-collected data, gave the age of every IPO in the sample.
In order to identify data on internet-related companies, the method of Loughran and Ritter (2004, Appendix D) was used. They merged all internet identifications of Thomson Financial Securities Data, Dealogic, and IPOMonitor.com. This dataset was manually controlled and corrected in order to merge the dataset with the data obtained from the Thomson One Database. Tech stocks were defined by Thomson One as macro level industry: High-technology. Finally, all datasets were merged together.
The dataset was reduced from 3,472 observations to only 856 observations. This strong reduction was mainly due to considerable missing data and unknown underwriters in the Thomson One’s database (reduction of 1,139), but also due to missing offer prices and first-day trading prices (reduction of 926). The industries financial (387) and real estate (77) were excluded from the dataset, just as IPOs with an offer price below $5.00 per share (97).
V. DESCRIPTIVE STATISTICS
Table I shows a summary of statistics for the variables used in the complete sample and displays the distribution of the variables. The IPOs in the sample are underpriced (first-day return) by 13.7% on average. The standard deviation is 23.4%, and the first-day return is within a range of -89.8% to 131.4%. This suggests that the first-day return varies considerably, and although the average first-day return is positive, negative values also exist in this sample. Of the sample, 28.47% were zero or negative first-day returns. The mean Carter-Manaster proxy is 8.2, while the maximum rank is 9 and the minimum rank is 1, which suggests that a great part of the issuers in the sample used, as qualified by Carter and Manaster (1990), high ranked underwriters. Panel B of Table II shows 44.7% of the issued IPOs were underwritten by an underwriter ranked with the highest possible rank of 9. The Megginson-Weiss proxy is based on relative market share. Underwriters with market shares smaller than 1% of the market account for 22.3% of the IPOs. Underwriters with a market share larger than 8% account for 43.1% of the IPOs. Appendix A shows that only four underwriters meet this criterion: Goldman Sachs, Morgan Stanley, Citi Group, and Bank of America Merrill Lynch. Underwriters with a market share of 5% or higher account for 65.23% and are represented by only seven underwriters (Morgan Stanley, Goldman Sachs, JP Morgan, Citi Group, Bank of America Merrill Lynch, Credit Suisse, and Deutsche Bank), which suggests that prestigious underwriters dominate the current market. The Johnson-Miller proxy is based on a four point-ranking scale. This ranking scale is derived from the Carter-Manaster proxy and, therefore, shows a similar result. The Johnson-Miller proxy mean of 2.167 corresponds with the Carter-Manaster proxy mean of 8.154.
The average proceeds of the IPOs were 239.03 million dollars and ranged from 2.8 million dollars to 16.01 billion dollars. The average age of the companies going public was 20.6 years and differed from one year to 166 years. Some firms existed for decades before going public. The amount of involved underwriters with an IPO goes up to 11 lead underwriters and 34 lead and co-lead underwriters and co-managers.
Table 1: Summary Statistics - All Variables
The summary statistics includes 854 IPO observations from 2005 to 2013 with an offer price of at least 5$. All
the data containing IPOs was obtained from the Thompson One database. First-day return is the first-day closing price minus the offer price divided by the offer price. The Carter-Manaster proxy equals one (zero otherwise) if the lead underwriter has a rank of 8 or more. The Megginson-Weiss proxy is the relative market share of the lead underwriter. The Johnson-Miller proxy is ranked 3 if the lead underwriter is in the bulge bracket, ranked 2 if in the major bracket, ranked 1 if in the sub-major bracket, and all other lead underwriters are ranked 0. Assets are the firm’s pre-issue book value of assets in millions of dollars. No. Lead Underwriters is the number of lead underwriters that were assigned to the IPO. Ln(Age) is the natural logarithm of the firms age plus 1. Tech dummy is one (zero otherwise) if the firm is a technological firm. Internet dummy is one (zero otherwise) if the firm is an internet related firm. Tech/internet dummy is one (zero otherwise) if the firm is a technological or internet related firm. VC-back is a dummy that equals one (zero otherwise) if the IPO was backed by venture capital.
Variable Obs. Mean Std. Dev. Min Max
First-day Return 857 13.7% 23.4% -89.8% 131.4%
Carter-Manaster proxy 846 8.154 1.423 1 9
Megginson-Weiss proxy 857 7.433 5.519 0.001 17.733
Johnson-Miller proxy 846 2.167 0.934 0 3
Proceeds (in millions) 857 $239.03 $814.34 $2.80 $16,006.88
Assets (in millions) 783 $989.28 $5446.09 $0.21 $137,238.00
Vc-backed dummy 857 0.423 0.494 0 1
Ln(Age) 857 2.451 1.087 0 5.111
Age (in years) 857 20.6 26.1 1 166
High-Tech dummy 857 0.240 0.427 0 1
Internet dummy 857 0.098 0.297 0 1
No. Lead underwriters 857 2.550 1.569 1 11
No. Lead-, Co-lead underwriters, Co-managers
857 5.879 3.622 1 34
Table II gives a clear view of the distribution of the number of lead underwriters and the different ranking measures presented in this thesis. The different panels show that, in the sample, only a few underwriters are responsible for, or at least participated in, a relatively large part of the IPOs. Prestige underwriters dominate the market. The highest number of IPOs was underwritten by two lead underwriters (41.07%) and the majority was underwritten by one, two, or three underwriters (81.33%).
Table II: Summary Statistics – Distribution of number of lead underwriters and different ranking measures of underwriters
The summary statistics includes 854 IPO observations from 2005 to 2013 with an offer price of at least 5$. All
the data containing IPOs was obtained from the Thompson One database. First-day return is the first-day closing price minus the offer price divided by the offer price. The Carter-Manaster proxy equals one (zero otherwise) if the lead underwriter has a rank of 8 or more. The Megginson-Weiss proxy is the relative market share of the lead underwriter. The Johnson-Miller proxy is ranked 3 if the lead underwriter is in the bulge bracket, ranked 2 if in the major bracket, ranked 1 if in the sub-major bracket, and all other lead underwriters are ranked 0. No. Lead Underwriters is the number of lead underwriters that were assigned to the IPO.
Panel A: The number of IPOs with the specified number of lead underwriters
All 1 2 3 4 5 6 7 8 9 10 11
Number 857 191 352 154 74 42 18 10 5 6 4 1
Percentage 100 22.29 41.07 17.97 8.63 4.90 2.10 1.17 0.58 0.70 0.47 0.12
Panel B: The number of IPOs with the specified ranked lead underwriter (Carter-Manaster)
All 1 2 3 4 5 6 6.5 7 7.5 8 8.5 9
Number 846 2 8 20 9 14 13 8 69 19 90 216 378
Percentage 100 0.24 0.95 2.36 1.06 1.65 1.54 0.95 8.16 2.25 10.64 25.53 44.68
Panel D: The number of IPOs with the specified ranked lead underwriter (Johnson-Miller)
All 3 2 1 0
Number 846 446 285 185 100
Percentage 100 44.68 36.17 10.40 8.75
The number of IPOs by industry is presented in Table III. Both financial and real estate industries were excluded from the sample, so 414 financial IPOs and 83 real estate IPOs were excluded. Currently, most IPOs take place in the following sectors: Financial (414), healthcare (286), and high technology (206). During the crisis, only 133 observations of IPOs were registered in the dataset, which is less than in the given period before the crisis and after the crisis. During the pre-crisis, the highest returns were with consumer staples (23.0%), retail (20.0%), and high technology (19.2%). During the crisis, the highest returns were with consumer staples (18.4%), Telecommunications (11.6%), and high technology (11.2%). After
Panel C: The number of IPOs with the specified ranked lead underwriter (Megginson-Weiss)
All 0-1 1-2 2-3 3-4 4-5 5-7 7-8 8-9 10-11
Number 857 177 23 35 31 32 100 89 111 259
the crisis, consumer staples (41.4%), retail (41.1%), and Consumer Products/Services (29.6%) provided the highest returns. Based on this results, there is no cogent reason to only include high technology in the research. The average return of internet IPOs during the pre-crisis and after-crisis exceeds the average total return, where the opposite accounts for the crisis period. However, the number of observations in some industries was so low that some single high observations caused the average high return. For example, only nine internet companies went public during the crisis. Demers and Lawellen (2003) agued a great difference between underpricing of internet firms over non-internet firms, especially during the dot-com bubble. In this research there are great differences between internet firms and non-internet firms as well. However, this phenomenon only occurs in the periods outside the crisis. Remarkably is the respectively lower return of internet companies during the crisis.
Table III: Number of IPOs by industry
The number of IPOs is presented by industry. The sample includes 854 IPO observations from 2005 to 2013 with
an offer price of at least 5$. All the data containing IPOs was obtained from the Thompson One database.
First-day return is the first-First-day closing price minus the offer price divided by the offer price. The industry dummies are one (zero otherwise) if the firm is active in the presented industry. The Thomson Reuters proprietary macro-level industry classifications are used. These classifications are based on SIC codes, NAIC codes, and overall company business descriptions. In total, there are 12 industries and each issuing firm is assigned to one of these macro industries. Internet dummy is one (zero otherwise) if the firm is an internet related firm.
Panel A: Industry defined by Thomson Reuters proprietary macro-level industry classifications
Pre-Crisis Crisis After Crisis
Industry No. of IPOs Percentage Return N Return N Return N
Consumer Products/Services 62 7.23% 11.8% 32 9.7% 17 29.6% 13
Consumer Staples 26 3.03% 23.0% 15 18.4% 4 41.4% 7
Energy and Power 139 16.22% 9.1% 63 4.9% 17 4.8% 59
Healthcare 286 21.70% 6.9% 95 2.6% 23 14.9% 68
High Technology 206 24.04% 19.2% 93 11.2% 29 26.3% 84
Industrials 72 8.40% 13.1% 35 9.7% 16 14.2% 21
Materials 50 5.83% 2.6% 21 4.5% 14 5.6% 15
Media and Entertainment 29 3.38% 11.8% 12 7.4% 3 16.4% 14
Retail 56 6.53% 20.0% 22 10.7% 10 41.1% 24
Telecommunications 31 3.62% 3.1% 21 11.6% 4 8.0% 6
Total 856 100% 11.98% 409 8.03% 137 18.72% 311
Panel B: Industry defined by internet or non-internet
Internet 84 9.80% 21.90% 37 6.39% 9 29.93% 38
Non-Internet 773 90.20% 10.99% 372 8.14% 128 17.17% 273
In table IV, the correlations between variables are presented in a matrix. This matrix shows the linear cohesion between variables, as well as their direction. First-day return is positively significantly related to the Carter-Manaster proxy, the Megginson-Weis proxy, the Johnson-Miller proxy, the high-tech dummy, the internet dummy, the high-tech/internet dummy, and the VC-backed dummy. Notably, there is a weak correlation between the first day return and proceeds, assets, ln(age), and number of lead underwriters. Even though these independent variables were commonly used in previous literature. The correlation matrix shows a very strong significant correlation between the various proxies to measure underwriter prestige (0.613, 0.846, and 0.635), which is consistent with the expectations. The correlation matrix also shows a strong significant correlation between the high-tech dummy and the internet-dummy (0.347). This suggests that both dummies overlay each other. Therefore the two dummies are combined in a tech/internet dummy in order to avoid multicollinearity in the OLS regression between the high-tech dummy and the internet-dummy (0.347). This suggests that both dummies overlay each other.
Therefore the two dummies were combined in a tech/internet dummy in order to avoid multicollinearity in the OLS regressions presented later on. The correlation between first-day return and the new created dummy is higher than the tech- and internet dummy separate (0.187). Proceeds of the IPO and assets before IPO correlate very strongly (0.725). The larger the company, the higher were the proceeds of the IPO. Since the variable assets does have a slightly higher correlation with first-day return, this variable is used in the regressions and the variable proceeds will be left out. The residual correlation matrix shows no very strong correlations within independent variables, which indicates that multicollinearity does not exist in the rest of the sample.
Table IV: Correlation Matrix of dependent and independent variables
The correlation matrix includes a sample of 854 IPO observations from 2005 to 2013 with an offer price of at least 5$. All the data containing IPOs was obtained from the
Thompson One database. First-day return is the first-day closing price minus the offer price divided by the offer price. The Carter-Manaster proxy equals one (zero otherwise) if the lead underwriter has a rank of 8 or more. The Megginson-Weiss proxy is the relative market share of the lead underwriter. The Johnson-Miller proxy is ranked 3 if the lead underwriter is in the bulge bracket, ranked 2 if in the major bracket, ranked 1 if in the sub-major bracket, and all other lead underwriters are ranked 0. Assets are the firm’s pre-issue book value of assets in millions of dollars. No. Lead Underwriters is the number of lead underwriters that were assigned to the IPO. Ln(Age) is the natural logarithm of the firms age plus 1. Tech dummy is one (zero otherwise) if the firm is a technological firm. Internet dummy is one (zero otherwise) if the firm is an internet related firm.
Tech/internet dummy is one (zero otherwise) if the firm is a technological or internet related firm. VC-back is a dummy that equals one (zero otherwise) if the IPO was backed by venture capital. *, **, and *** indicate significance at 10%, 5%, and 1%, respectively.
First-day Return Carter-Manaster Meggins on-Weiss Johnson-Miller
Proceeds Assets Ln(Age) Tech
dummy Internet dummy Tech/Inter net dummy VC-backed No. Lead underwriters First-day Return 1 Carter-Manaster 0.107** 1 Megginson-Weiss 0.145*** 0.613*** 1 Johnson-Miller 0.170*** 0.846*** 0.635*** 1 Proceeds -0.031 0.103** 0.115*** 0.118*** 1 Assets -0.034 0.080** 0.079** 0.080** 0.725*** 1 Ln(Age) 0.005 0.043 0.066 0.02 0.085* 0.140*** 1 Tech dummy 0.171*** 0.076 0.113*** 0.142 -0.001 -0.061 -0.03 1 Internet dummy 0.141*** 0.058 0.056 0.092** 0.049 -0.037 -0.045 0.347*** Tech/Internet dummy 0.187*** 0.086* 0.106** 0.152*** -0.009 -0.067 -0.036 0.923*** 0.541*** 1 VC-backed 0.205*** 0.027 0.064 0.089** -0.088** -0.133*** -0.188*** 0.247*** 0.217*** 0.285*** 1
Table V presents the average first-day returns on IPOs categorized by proceeds, age, industry, and number of lead underwriters. Returns vary between 12.0% before the crisis, 8.0% in the crisis, and 18.7% after the crisis. Loughran and Ritter (2004) did examine the first-day returns over the period 1980 to 2003. The results from Table V do connect their results except for the bubble period. Average underpricing for IPOs during the 1980s was 7.3%, during the 1990s 14.8%, during the internet bubble 65%, and between 2001 and 2003 11.7%. However, it is interesting to observe a drop in first-day returns during the crisis. This drop in return suggests that underwriters behave in a period of distress in an opposite manner towards underpricing of IPOs than in bubble periods. Besides, there was also a drop in the number of IPOs during the crisis. Only 137 were recorded in this sample. IPOs with large proceeds compared to small proceeds tended to have slightly higher returns. This phenomenon also applies to smaller, younger firms, and returns of IPOs with a high prestigious underwriter are clearly higher. More notable is the difference in returns between small and large firms, based on assets, is 6.1% and significant. Loughran and Ritter (2004) also reported that assets have higher underpricing averages during their sample period, 1980 -2003. Also technological and internet related firms have significant higher returns than non-technological and non-internet related firms. Returns are almost doubled, but keep in mind that the number of observations of these two groups is not that high.
Table V: Average first-day returns on IPOs categorized by proceeds, age, industry, and number of lead underwriters
The sample contains of 854 IPO observations from 2005 to 2013 with an offer price of at least 5$. All the data
containing IPOs was obtained from the Thompson One database. First-day return is the first-day closing price minus the offer price divided by the offer price. The amount of money that is raised from the IPO is the
proceeds.Assets are the firm’s pre-issue book value of assets in millions of dollars. No. Lead Underwriters is the
number of lead underwriters that were assigned to the IPO. Ln(Age) is the natural logarithm of the firms age plus 1. Tech dummy is one (zero otherwise) if the firm is a technological firm. Internet dummy is one (zero
otherwise) if the firm is an internet related firm. Tech/internet dummy is one (zero otherwise) if the firm is a technological or internet related firm. VC-back is a dummy that equals one (zero otherwise) if the IPO was backed by venture capital. The pre-crisis is the period between 2005 and 2007. The crisis is the period between 2008 and 2010. The after-crisis is the period between 2011 and 2013. *, **, and *** indicate significance at 10%, 5%, and 1%, respectively.
Pre-Crisis Crisis After-Crisis Total
Segmented by Return N Return N Return N Return N T-test
Proceeds
Large 14.4% 203 9.0% 67 18.7% 155 15.1% 425 Assets
Small 15.1% 176 9.5% 67 22.8% 150 17.1% 392 3.80***
Large 9.6% 233 6.6% 70 14.9% 161 11.0% 462
Age
Young (0-10 years old) 12.4% 210 9.6% 65 18.7% 157 14.3% 431 0.53
Old (>10 years old) 11.6% 199 6.7% 72 18.7% 154 13.4% 423
Industry Non-technology 9.9% 316 7.2% 108 15.9% 227 11.6% 648 -5.07*** Technology 19.2% 93 11.2% 29 26.3% 84 21.0% 206 Industry Non-internet 11.0% 372 8.1% 128 17.2% 273 12.7% 770 -4.16*** Internet 21.9% 37 6.4% 9 29.9% 38 23.9% 84 VC-backed No 7.6% 241 7.2% 86 13.9% 167 9.7% 492 -6.13*** Yes 18.2% 168 9.4% 51 24.4% 144 19.5% 362
No. Lead Underwriter
Few (0-2 underwriters) 12.9% 338 6.9% 76 19.5% 129 13.7% 541 -0.29 Many (> 2 underwriters 7.8% 71 9.4% 61 18.2% 182 14.1% 313 Underwriter Prestige Low-prestige 9.8% 92 5.1% 26 8.2% 44 8.6% 162 -3.17*** High-prestige 12.6% 317 8.7% 111 20.5% 267 15.1% 692 All 12.0% 409 8.0% 137 18.7% 311 13.7% 854
The number of IPOs, first-day returns, number of lead underwriters and the amount of money left on the table are presented by cohort year in Table VI. Means are presented in Panel A and medians in Panel B. The number of IPOs decreased per year during the financial crisis and returned to the pre-crisis amount after the crisis. The same seemed to happen with first-day returns. The year 2008 reports an average return of 4.6% and 2010 an average return of 7.7%, while before and after the crisis returns never dropped below 10%. It seems that there is an increasing trend in the number of lead underwriters involved in the process of going public. The panel with medians shows lower returns, but the overall trends are the same as in the table showing means.
Table VI: Number of IPOs, first-day returns, number of lead underwriters, amount of money left on the table by cohort year
The mean and median returns are computed for the 854 IPO from 2005 to 2013 with an offer price of at least 5$.
All the data containing IPOs was obtained from the Thompson One database. First-day return is the first-day closing price minus the offer price divided by the offer price. No. Lead Underwriters is the number of lead underwriters that were assigned to the IPO. Money on the table is defined as the first-day price change times the number of shares issued.
Panel A: Means
Year Number of IPOs
First-Day
Return Std. Dev.
Number of Lead
Underwriters Money on the Table
2005 130 10.5% 17.55% 1.9 $1,512.20 2006 139 10.6% 18.82% 1.9 $1,871.72 2007 140 14.6% 24.58% 1.8 $2,998.30 2008 21 4.6% 18.99% 2.2 $2,248.98 2009 35 10.7% 14.95% 3.0 $4,232.41 2010 81 7.7% 13.90% 2.6 $1,942.86 2011 75 13.8% 21.98% 3.0 $2,963.79 2012 93 17.3% 27.33% 3.2 $1,062.22 2013 143 22.1% 31.99% 3.5 $5,281.48 Total 857 13.7% 22.35% 2.5 $2,690.28 Panel B: Medians
Year Number of IPOs
Frist-Day
Return Std. Dev.
Number of Lead
Underwriters Money on the Table
2005 130 7.1% 17.55% 2 $540.06 2006 139 5.4% 18.82% 2 $490.05 2007 140 8.8% 24.58% 2 $902.13 2008 21 0.1% 18.99% 2 $8.91 2009 35 7.9% 14.95% 3 $1,518.50 2010 81 5.0% 13.90% 2 $421.82 2011 75 8.5% 21.99% 3 $1,566.72 2012 93 14.2% 27.34% 3 $1,581.32 2013 143 11.5% 32.00% 3 $1,683.20 Total 857 8.4% 22.35% 2.4 $1,000.45
Table VII gives a clear impression of the difference between highly prestigious underwriters and underwriters with low prestige. Both means and medians are higher for prestigious underwriters than for those without. Age seems to be independent of underwriter prestige. Based on the data presented in the table below, no conclusions can be derived about the
relationship between age of the firm and underwriter prestige. Proceeds are higher for IPOs with a highly prestigious underwriter and the table also suggests that highly prestigious underwriters are more involved in IPOs of high technology firms. These conclusions do not differ much for the separate periods in time. To test the difference between medians, the Wilcoxon rank-sum test is used.
Table VII: Mean and median first-day returns, median age, median proceeds, and industry representation categorized by underwriter prestige
The sample contains of 854 IPO observations from 2005 to 2013 with an offer price of at least 5$. All the data
containing IPOs was obtained from the Thompson One database. The data is categorized by underwriter prestige. First-day return is the first-day closing price minus the offer price divided by the offer price. The amount of
money that is raised from the IPO is the proceeds.Ln(Age) is the natural logarithm of the firms age plus 1. Tech
dummy is one (zero otherwise) if the firm is a technological firm. Internet dummy is one (zero otherwise) if the firm is an internet related firm. The pre-crisis is the period between 2005 and 2007. The crisis is the period between 2008 and 2010. The after-crisis is the period between 2011 and 2013. The t-test and Wilcoxon rank-sum test are tests to calculate whether the value for low prestige differs significantly from the value of high prestige. *, **, and *** indicate significance at 10%, 5%, and 1%, respectively.
Pre-Crisis Crisis After-Crisis Total
Return N Return N Return N Return N
t-stat/z-test Mean first-day returns
Low prestige 9.7% 92 5.0% 26 8.2% 44 8.6% 162 -3.17***
High prestige 12.6% 317 8.7% 111 20.5% 267 15.1% 692
Median first-day returns
Low prestige 4.8% 92 1.3% 26 1.2% 44 3.2% 162 -3.02***
High prestige 7.2% 317 5.3% 111 12.9% 267 9.2% 692
Median Age
Low prestige 10 years 92 11 years 26 8.5 years 44 10 years 162 -1.24
High prestige 10 years 317 11 years 111 11 years 267 11 years 692
Median Proceeds Low prestige $ 48.70 92 $54.50 26 $42.61 44 $49.33 162 -13.15*** High prestige $ 121.50 317 $151.45 111 $145.60 267 $132.71 692 Percentage tech-related Low prestige 18.4 92 15.4 26 15.9 44 17.2 162 -2.264** High prestige 23.9 317 22.5 111 28.8 267 25.7 692 Percentage internet-related Low prestige 8.7 92 3.8 26 2.2 44 6.1 162 -1.7403* High prestige 9.1 317 7.2 111 13.8 267 10.7 692 Total 12.0% 409 8.0% 137 18.7% 311 13.7% 854
VI. RESULTS
Table VIII presents five different OLS regressions. For regressions 1 and 2, the Carter-Manaster proxy was used as a measure for underwriter prestige. Prestigious underwriters were assigned to those with a Carter-Manaster proxy of 8 or higher. In regression 2, time fixed effects and industry fixed effects were added to the regression, which resulted in an increase in the adjusted R-squared from 0.0699 to 0.0766. However, the adjusted R-squared remained low. Regression 1 actually disclosed that the first-day returns of IPOs were 3.567% higher for prestigious underwriters than for non-prestigious underwriters, significant at 1%. Adding the time fixed effect and industry fixed effects increased this outcome to a higher return of 4.641%, which is also significant at 1%. The betas of the vc-backed dummy and the tech/internet dummy were significant, but ln(assets) and ln(age) were far from significant. These independent variables may or may not be relevant, which is consistent with the suspicions derived from the correlation matrix.
Regression 3 and 4 were based on the Megginson-Weiss measure for underwriter prestige. Regression 3 disclosed that the first-day return of IPOs was 0.383% higher for every percentage point increase in relative market share of the underwriter, significant at 1%. In regression 4 the time fixed effects and industry fixed effects were added to the regression. First-day return increased by 0.434% for every percentage point increase in relative market share of the underwriter.
Regressions 5 and 6 did not differ much from the other regressions. Regression 5 showed that one increase in tier, for example when underwriters move from the major bracket to the bulge bracket, resulted in an increase of 2.633% in first-day returns, significant at 1%. Regression 6 shows an increase of 3.746% in first day returns, when the fixed effects were added to the regression.
Overall there are only small differences notable between the various proxies. This is not a surprising effect since all proxies give an indication of the level of prestige of an underwriter and are correlated. The MW proxy had slightly higher explanatory power and stays statistically significant at the 1 percentage level. Based on these findings, the MW was used as proxy for underwriter prestige in the remaining regressions. However all proxies do show a significant positive relationship between IPO underpricing and underwriter prestige over the period 2005 – 2013, which is in line with the hypotheses set in a previous section. All regressions are also significant different from zero, tested with an F-test shown in table VIII. Adding fixed effects give a slight increase in adjusted R-squared. Loughran and Ritter (2004)
argued that IPOs underwritten by top-tier underwriters were underpriced more in the 1990s and, especially, in the bubble period because of spinning and more highly ranked analysts. They notified that highly ranked analysts are most of the times employed by top-tier underwriters. The findings from Table VIII are in line with the hypothesis of Loughran and Ritter (2004). A positive significant relationship was found from 2005 to 2013.
Table VIII: IPO Underpricing regressions on prestige measures
The sample includes 854 IPO observations from 2005 to 2013 with an offer price of at least 5$. The dependent variables in the regressions are three different proxies to measure prestige. Colomn 1 and 2 regress the Carter-Manaster proxy on first-day return. Column 3 and 4 regress the Megginson-Weiss proxy on the first-day return. Column 5 and 6 regress the Johnson-Miller proxy on the first-day return. The Carter-Manaster proxy equals one (zero otherwise) if the lead underwriter has a rank of 8 or more. The Megginson-Weiss proxy is the relative market share of the lead underwriter. The Johnson-Miller proxy is ranked 3 if the lead underwriter is in the bulge bracket, ranked 2 if in the major bracket, ranked 1 if in the sub-major bracket, and all other lead underwriters are ranked 0. Ln(Assets) is the natural logarithm of the firm’s pre-issue book value of assets in millions of dollars. No. Lead Underwriters is the number of lead underwriters that were assigned to the IPO. Ln(Age) is the natural logarithm of the firms age plus 1. VC-back is a dummy that equals one (zero otherwise) if the IPO was backed by venture capital. Tech/internet dummy is one (zero otherwise) if the firm is a technological or internet related firm. The time fixed effects are based on the IPO year and industry fixed effects based on the 12 included industries. T-statistics are computed using heteroscedasticity-consistent standard errors. *, **, and *** indicate significance at 10%, 5%, and 1%, respectively
1 2 3 4 5 6 Intercept 0.186 3.731 0.784 4.386 -1.565 1.725 (0.07) (0.97) (0.31) (1.17) (-0.55) (0.44) Carter-Manaster proxy 3.567** 4.641** (1.87) (2.16) Megginson-Weiss proxy 0.383*** 0.434*** (2.72) (2.88) Johnson-Miller proxy 2.633*** 2.746*** (3.19) (2.99) Ln(Assets) -0.911 -0.535 -0,901* -0.526 -1.514** -1.003 (-1.47) (-0.79) (-1.76) (-0.80) (-2.23) (-1.41)
No. Lead Underwriter 1.382** 0.207 1.278** 0.143 1.100* 0.018
(2.20) (0.33) (2.09) (0.23) (1.82) (0.03) Ln(Age) 0.717 -0.979 0.637 -1.001 0.701 -0.88 (1.28) (-1.19) (1.14) (-1.22) (1.21) (-1.06) VC-backed 8.879*** 11.617*** 8.680*** 11.538*** 8.405*** 11.362*** (4.88) (5.65) (4.81) (5.63) (4.61) (5.4) Tech/internet dummy 7.586*** 0.825 7.377*** 1.160 7.142*** 0.45
