Underpricing of IPOs : do top-banks leave more money on the table?

Amsterdam Business School Faculty of Economics and Business

Master of Science in

Business Economics: Finance

Master Thesis

Underpricing of IPOs: Do Top-Banks leave more

money on the table?

Name: Shahin Dashti Student ID: 10840893 Professor: Dr. Liang Zou

II

Statement of Originality

This document is written by Student Shahin Dashti 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 creat-ing it.

The Faculty of Economics and Business is responsible solely for the supervision of comple-tion of the work, not for the contents.

III

Table of Contents

2. Literature Review and Underpricing Theories ... 3

2.1 Characteristics of IPOs ... 4

2.2 IPO Underpricing ... 8

2.3 Asymmetric Information Theory... 10

2.4 Institutional Theory ... 14

2.5 Control Theory ... 16

2.6 Behavioural Theory ... 17

2.7 Reputation of Underwriter ... 19

3. Methodology ... 21

3.1 Dataset and Sample ... 21

3.2 Measurement of Underpricing ... 22

3.3 Measurement of Underwriter Reputation ... 23

4. Empirical Results and Analysis ... 25

4.1 Descriptive Statistics ... 25

4.2Regression Results ... 31

4.3Empirical Validation from Academic Literature ... 38

5. Conclusion ... 40

References ... 42

IV

List of Tables

Table 1: Overview of the IPO activity in the U.S. from 2004 – 2011 ……….. 28 Table 2: Sample characteristics sorted by Top-Tier and Non-Top-Tier IPOs ………. 30 Table 3: Multicollinearity test ……….……… 32 Table 4: OLS Regressions of the percentage first-day return on the Top-Tier variable and oth-er control toth-erms ………... 35 Table 5: Carter-Manaster ranks for all underwriting banks in the sample ……… ………… 51 Table 6: OLS Regressions of the market adjusted percentage first-day return on the Top-Tier variable and other control terms ……….…… 54

V

List of Figures

Figure 1: Involved parties during the IPO process ………..…..………. 7 Figure 2: Number of IPOs per sector in the U.S. from 2004 – 2011 ……….…….. 26 Figure 3: Number of IPOs in the U.S. from 2004 – 2011 and average first-day return ……. 27

VI

List of Abbreviations

CAPM Capital Asset Pricing Model DJIA Dow Jones Industrial Average IPO Initial Public Offering

MLP Master Limited Partnership

NTT Non-Top-Tier

OLS Ordinary Least Squares

REIT Real Estate Investment Trust

SEO Seasoned Equity Offering

TT Top-Tier

1.

Introduction

The decision of private firms to go public is motivated by a variety of reasons, of which the access to new sources of finance is often claimed to be one of the most important (Röell, 1996). Whereas the spectrum of usage for fresh funding is related to company specific char-acteristics, one could still argue, that it is to the firm’s interest to maximize the proceeds from the Initial Public Offering (IPO). One measure for this could be to issue stocks at an of-fer price that reflects the intrinsic value of the firm in the aggregate. Investors would then earn the required rate of return considering risk exposures of the firm, while managers would raise a reasonable amount of equity with which they pursue the desired goals for the company.

In fact, pricing mechanisms and their outcomes in Initial Public Offerings (IPOs) have been examined by academics in detail for a long time, proposing that offerings are on average underpriced. A variety of literature from the past decades reveals that IPO stocks tend to experience very high returns during their first day of trading; this suggests that they are of-fered at a larger discount than necessary in regard to risk factor exposures of the firms.1 Or put simply: Stocks of IPO firms are probably sold too cheap. This phenomenon, commonly known as “underpricing”, was first explicitly documented by Ibbotson (1975), and later con-firmed for different periods and regions by numerous authors.2 As it seems that first-day returns on average exceed the theoretically required risk adjusted rate of return, scientists are confronted with a very puzzling question: Why do firms and associated lead underwrit-ers “leave money on the table” by issuing shares at a large discount, making an IPO very costly for the firm? At least one could ask whether this is intended by the firms as a type of premium to investors and investment banks or if it is just due to the inability of the under-writer to price the shares adequately. Either way, this raises the question if there are pre-dictable differences in the outcome of an IPO simply by choosing the “right” lead

1

See IBBOTSON (1975). The author finds little evidence that initial returns in IPOs can be explained using a one-factor model as the CAPM. Ritter (1984) tests whether initial returns are explainable through the risk composi-tion of the firm, which he has to reject.

2

See also among others IBBOTSON, JAFFE (1975); RITTER (1984); CARTER, MANASTER (1990); HANLEY (1993);

CARTER, DARK, SINGH (1998); LOUGHRAN, RITTER (2002); LOUGHRAN, RITTER (2004); LJUNGQVIST, WILHELM (2003); CHAMBER, DIMSON (2009); LIU, RITTER (2011). See ENGELEN, VAN ESSEN (2010) for a detailed

2 er, since he prices the shares. In other words: Do firms experience different levels of under-pricing once they choose a Top-Tier investment bank as lead underwriter? This constitutes the underlying question that I want to investigate in this thesis. The results of this thesis can help companies which are about to launch an IPO in their decision making process if a Top-Tier investment bank will really act in their best interest.

I argue that Top-Tier investment banks on average underprice more than less prestigious investment banks, leading to higher first day returns of IPOs with an prestigious (Top-Tier) investment bank as the underwriter. For a sample of 905 IPOs in the U.S. from 2004 until 2011, I document average first-day returns of 10.8%. Furthermore, stocks of IPO firms with prestigious (Top-Tier) lead underwriters on average yield first-day returns of 11.7% com-pared to average returns of 7.3% if the offering was arranged by non-prestigious (Non-Top-Tier) lead underwriters. Most important: OLS regressions with the first-day return as de-pendent variable and multiple different indede-pendent controls, confirm this difference. Coef-ficients for the Top-Tier underwriter dummy variable obtain values between 4.03% and 4.36% paired with p-values from 0.0028 to 0.0107. Thus, besides a relatively stable magni-tude, the Top-Tier variable turns out to be significant at the 99% confidence level in each regression model. This underlines the robustness of the proposed effect from the choice of a Top-Tier investment banks as lead underwriter. Even though firms differ significantly in size for Top-Tier and Non-Top-Tier underwritten IPOs, these differences seem to not sufficiently explain increased underpricing by Top-Tier lead underwriters. Furthermore, the same out-come can be shown once I adjust my return data for market movements on the respective IPO date. All in all my results support the hypothesis, that Top-Tier investment banks under-price more than Non-Top-Tier investment banks, inducing significantly higher returns on the first trading day for the latter ones. My work is in line with a broad base of prior findings on the role of the lead underwriter as a determinant for initial returns in IPOs, but should not be regarded as a dismissal of opposing studies that show up different results. In the end, it might still be the case that other variables that are not included in my analysis, explain the seemingly strong effect of Top-Tier lead underwriters.

Finally, the question for possible reasons of the outcomes, mentioned above, remains. As my work is largely based on previous secondary research conducted on the subject, I do not aim to add new hypotheses that could exemplify this phenomenon. Instead I will pick up on a

3 few explanatory points, in order to further align my results with previous findings and inter-pret it for existing frameworks of scientific literature.3 Furthermore, I will replace some ex-planatory variables of Loughran and Ritter’s model with new independent variables that I found to be more relevant in order to explain the phenomenon of underpricing. Neverthe-less, this is not the core aspect of my work, as the full coverage of hypotheses and models around the role of the underwriter in an IPO would clearly go beyond the scope of my thesis. My thesis proceeds as follows. Chapter 2 reviews the milestone literature that emerged over the last few decades and briefly discusses several theoretical approaches towards IPO un-derpricing. This chapter also presents a theoretical approach developed by Loughran and Ritter (2004). Chapter 3 presents the methodology behind my empirical analyses. The results in chapter 4 are split into three parts. The first part covers descriptive statistics on underpric-ing and IPO activity in the U.S. from 2004-2011, includunderpric-ing splits of IPOs into those conducted by Top-Tier Investment Banks and Non-Top-Tier Investment Banks. The second part illus-trates the results from different regression models, showing a significant and robust rela-tionship between underwriter reputation and underpricing. The last part aligns my empirical findings with those of recent scientific literature on underpricing and the role of the lead underwriter. This has two main reasons. Not only do they provide support for my own re-sults, but also do they underpin critical aspects for theoretical explanation approaches. Chapter 5 concludes my thesis.

2.

Literature Review and Underpricing Theories

In the introductory chapter, the underlying motivation for this thesis has been outlined and the main findings have been stated. The following chapter will outline the rationale behind the process of going-public for firms. Furthermore, the IPO underpricing phenomenon will be introduced, and different theories, as well as empirical evidence, will be stated to explain the aforementioned underpricing occurrence. In addition to that, the rationale of TT-underwriters for IPO underpricing will be elaborated. Due to brevity reasons, it is almost

3

My analyses are largely oriented on the work of Loughran and Ritter (2004). As the two authors provide em-pirical evidence for the period of 1980 until 2003, I now caught up large parts of their methodology and ex-panded it for the period from 2004 until 2011. The mentioned paper will also be the core source for potential explanations why Top-Tier Investment Banks seem to underprice more.

4 impossible to review all theoretical approaches towards IPO underpricing. However, I tried to include the seminal papers that have shaped the ongoing research.

2.1 Characteristics of IPOs

Initial Public Offerings

Accessing new sources of capital is essential for firms: the issuance of stocks to investors is one way to access these new sources. Therefore, when a private company sells stocks for the first time to the public on the primary market, it is referred to as an initial public offer-ing. The process of going-public involves different parties with different objective functions and incentives. During the last few decades, four stylized facts gained a wide acceptance amongst scholars and practitioners, such as the short-run underpricing, the long-run under-performance, the cyclicality of the number of issues, and the substantial costs associated with the IPO (Berk and DeMarzo, 2014). Regarding the issuing method, multiple methods exist in which IPO stocks can be sold to the public. The most prominent issuing methods are fixed price offerings, auctions, hybrid mechanisms, and bookbuilding; whereas the latter is the predominant method internationally and especially in the U.S. (Sherman, 2001).

The Rationale behind Going-Public

The IPO literature has become fairly mature as well as sophisticated; there are now several theories as to why firms go public. The main advantages are better access to capital and greater liquidity. The disadvantages, however, are the substantial costs associated with the IPO. The costs can be roughly grouped into six categories: gross spread, other direct expens-es, indirect expensexpens-es, abnormal returns, underpricing, and green-shoe provisions (Ross et al., 2013). However, it is a common conception that underpricing costs exceed all other costs during the issuance process (Berk and DeMarzo, 2014).

The Market Timing Theory states that entities issue stocks on the primary market when its equity is overvalued. For instance, Bayless and Chaplinsky (1996) state that firms have a window of opportunity. These are certain circumstances that support the going-public-process such as: good general stock market conditions, a good industry condition, and the frequency, size and success of previous IPOs in the financial cycle. Furthermore, Choe et al.

5 (1993) find that firms are likely to issue during hot market phases when the number of IPOs is on a high absolute level.

The Valuation Theory claims that the monitoring of outside shareholders after the IPO is value enhancing for the company and supports the going-public desire of the owners (Holmstrom and Tirole, 1993). Additionally, Amihud and Mendelson (1988) find that the in-creased liquidity associated with the listing also contributes to an increase in firm value. The Life Cycle Theory (Zingales, 1995) states that the IPO is linked to the life cycle of a com-pany; earlier in its life cycle, entities tend to be private, then when they become sufficiently large it is optimal for them to go public. According to the Life Cycle Theory, an IPO might be a means of the entrepreneur to regain control from angle investors or venture capitalists (Black and Gilson, 1998). Furthermore, Zingales (1995) observes that public targets are easi-er to spot for potential acquireasi-ers and obtain higheasi-er valuations; hence entrepreneurs facili-tate the acquisition through the IPO.

Additionally, further reasons behind going-public include the desire to raise capital for the firm and to create a public market where the equity holders can convert their shares into cash at a future date. Finally, there are also non-financial motives, such as increased publici-ty and a better brand perception by the public (Ritter and Welch, 2002).

The Key Parties and the IPO Process

The key parties involved in the IPO process are the issuer, the underwriting bank, and the outside investors, which are all characterized by differing objective functions.

The issuer is the underlying firm that decides to go public and therefore sell its shares to outside investors. The main tasks for issuers are to make commitments to the initial public offering, to provide other involved parties with the necessary data, to hand in legal docu-ments to the specific institutional bodies, and to choose the underwriting investment bank. Two important criteria for the issuer in the underwriter selection are the reputation and the performance of the investment banks during past IPOs (Berk and DeMarzo, 2014). Loosely speaking, the main goal of the issuer is to maximize its proceeds from the IPO in addition to a successful and smooth launch.

6 The underwriter is the financial intermediary, typically an investment bank, which performs the IPO on behalf of the issuing company. An investment bank provides its clients primarily with merger and acquisition advisory services, capital raising services, trading of securities, brokerage and research. The main tasks of the underwriter during an IPO are the due dili-gence, the determination of the offer size (therefore indirect the capital structure), the preparation of the marketing material, and assistance in the preparation of regulatory filings (Berk and DeMarzo, 2014).

Furthermore, the underwriting bank issues the “red herring” (a preliminary prospectus with key metrics to potential outside investors), organises road shows, and is responsible for the bookbuilding. The latter is the collection of information from outside investors about the demand for the IPO stock to set the price range and the offering size.

During the offering period, one common pricing mechanism is the firm commitment, where the underwriter buys all stocks from the issuing firm at a prespecified price and sells them to outside investors for the offer price (Dunbar, 1998). If the outside demand for the issued shares is exceptionally high, the investment bank is allowed to issue additional shares, usual-ly up to 15% (green-shoe option).

Moreover, the underwriting bank is also involved in aftermarket activities. If the first trading price falls below the offer price, the bank places stabilisation bids to avoid a declining stock price (Aggarwal, 2000). A further well observed activity in the after-market is short-covering. The investment bank short-sells the IPO stock prior to the first trading, benefiting from a declining stock price.

It is commonly observed, especially for large IPOs, that several investment banks built an underwriter syndicate with one or more lead underwriters to perform the issuance.

7 Figure 1: Involved parties during the IPO process

Source: Bodie, Kane, Markus, 2011

The investors involved in the IPO process can be grouped into institutional and retail inves-tors. Usually, retail investors are private investors that place smaller bids, while institutional investors are for instance mutual funds, banks, pension funds, hedge funds, and insurance companies with higher financial power (Berk and DeMarzo, 2014). Institutional investors usually place higher bids due to the pooling of investments and oftentimes possess informa-tional advantages. Nevertheless, both investors’ goal is to get a large allocation, to receive underpriced IPO stocks and to maximize their return (Eckbo, 2008).

Long-Run Performance

The IPO literature suggests that IPOs are a relatively poor long-term investment. Ljungqvist (1997) for instance, finds that German IPOs underperform their index benchmark by 12.1% after three years, whereas Stehle et al. (2002) observe a 6% underperformance three years after the offering. The findings are in accordance with various other academic papers. Possi-ble explanations for long-run underperformance are given by Miller (1977) who attributes a ‘clientele effect’ to IPOs. He states that only optimistic investors are involved in IPOs, but their overly optimistic view is corrected when more information about the true value of a firm is revealed. Heaton (2002) claims that managers tend to overinvest the proceeds ob-tained from the IPO and hence decrease the firm value. Teoh et al. (1998) find that poor

8 long-run performance is linked to early optimistic accounting to look financially healthy when the IPO is conducted.

2.2 IPO Underpricing

IPO underpricing is a phenomenon that has been intensively discussed over the last few decades and appears to be independent of time period and geographical area as well as the issuing method. The first empirical and theoretical models were developed in the early 1970s and extended over the following years.

In general, several means exist to define underpricing. The most prominent measures are on the one hand, the dollar amount left on the table and on the other hand the percentage dif-ference between the offer price and the first trading day closing price. As a result of under-pricing, significant abnormal returns are observable on the first day of trading. This indicates a wealth loss for the issuer of the IPO shares, since the stocks are sold at an offer price which could have been higher. Thus, money is left on the table by the issuer. A second side-effect is that retained shares are diluted after the IPO; the dilution can increase even further if green-shoe options are exercised by the underwriter.

The degree of underpricing does not only depend on geographical differences but also on the development of the capital market. Well-developed capital markets are expected to in-corporate the initial returns fairly quickly, whereas less-developed markets are expected to need a longer time period. A common observation is that economies with well-developed capital markets seem to have more moderate underpricing (Michelacci and Suarez, 2004). Additionally, cross-country differences may, in part, be caused by different institutional frameworks. Still, they are unlikely to fully explain the degree of underpricing that was doc-umented over the last few decades (Eckbo, 2008).

While IPO underpricing is a widely accepted phenomenon among academics, the reason behind the underpricing is an ambiguous topic. The first seminal paper for IPO underpricing was given by Reilly and Hatfield in 1969, who found a 20.2% underpricing for DJIA stocks over a sample period from 1963-1966. Further empirical evidence for the U.S. stock market was given by Logue in 1973 and Ibbotson in 1975. In 2002, Ritter and Welch documented a

9 significant relationship between different years and the degree of underpricing. Their results show an average underpricing of 7.4% in the 1980s, 11.2% in the 1990s, and 18.1% in the mid-90s for the U.S. market.

Additional evidence for underpricing was especially found during hot markets. From 1999 to 2000 (dot-com crisis) more than USD 62 billion were left on the table by issuing firms; the average IPO underpricing laid between 71% and 57% respectively (Eckbo, 2008). The huge proceeds that are foregone by issuers are on the first sight counterintuitive. But underpric-ing is not only geographically concentrated on the U.S. stock market. Ritter (2003) summa-rizes underpricing findings from various countries. For instance, Datar and Mao as well as Isa and Young (cited in Ritter, 2003) found an average underpricing of 256.9% and 104.1% be-tween 1990 and 2000 in China and Malaysia respectively. Consequently, underpricing seems to be a global phenomenon; albeit Ljungqvist et al. (2003) note that IPO practices seem to converge and become more homogenous.

In a different study, Ljungqvist (1997) found an average underpricing of 9.2% for 185 com-panies over a sample period from 1973-1990. Furthermore, he claims that the underpricing of German stocks is generally lower than that of U.S. stocks, which may be attributed to size and age differences.

However, fundamental misevaluation or asset-pricing risk premia are unlikely to explain the high first day return of new issued stocks (Ritter and Welch, 2002) and thus several theories emerged over the last few decades to explain underpricing.

In an earlier study, Loughran and Ritter (1995) conclude that underpricing is a result of the market that systematically misestimates the autocorrelation of earnings growth. Further-more they point out that the market overweight recent improvements in the operating per-formance.

The theories can be roughly grouped into the following four distinct categories: Asymmetric Information Theory, Institutional Theory, Ownership and Control Theory, and Behavioural Theory. Among the four different explanation approaches, the Asymmetric Information The-ory seems to be the best established so far, but there is no general agreement among schol-ars as to which explanation is correct (Eckbo, 2008). The following chapters will outline the

10 main theories of IPO underpricing that underpin the subsequent analysis and discuss some empirical evidence.

2.3 Asymmetric Information Theory

The IPO process involves a variety of different parties, such as the issuing firm, the under-writing investment bank and the investors. Central to the Asymmetric Information Theory is the idea that one of the abovementioned parties is better informed than the others, and that their objective functions are not aligned.

Rock’s Winner’s Curse

One prominent asymmetric information model was first discussed by Rock in 1986. His so called winner’s curse is built upon the well-established asymmetrical lemons problem by Akerlof (1970). Rock (1986) claims that some investors are better informed about the under-lying value of the IPO firm than other investors. Thus, informed investors bid only for IPOs that are perceived to be underpriced and avoid overpriced IPOs. On the other hand, unin-formed investors bid for both types of IPOs. This consequently leads to the following effects: if the demand for a specific IPO share is high (underpriced stock), every investor will obtain his demanded shares on a pro rata basis to his bids. If the demand for a stock is low (over-priced stock) the uninformed investor will obtain all the stocks he has bid for. The winner’s curse therefore claims that the uninformed investors always lose, since they receive shares with negative expected average returns. Thus, uninformed investors need to be attracted by stocks that are on average underpriced, otherwise they will not participate in the IPO mar-ket. Moreover, uninformed investors are needed because the demand of informed investors is not high enough to cover the supply.

Rock’s winner’s curse was tested and supported by several empirical studies. Some re-searchers tried to separate the investors into informed and uninformed investors to obtain useful proxies. Koh and Walter (1989) found that after making adjustments for allocations, returns fall materially in the Singaporean stock market. Amihud et al. (2003), state that unin-formed investors earn a negative allocation-weighted return on the Israeli stock market. Levis (1990) found similar results for the U.K studying 123 IPOs.

11 Another approach to test the winner’s curse theory is modelled by distributing the infor-mation among investors equally. Hence, if all investors obtain the same inforinfor-mation, under-pricing should not occur. Michaely and Shaw (1994) focus on master limited partnerships (MLP) which excludes institutional investors due to tax reasons, leading to a rather homoge-nous investor group. Their results indicate an initial return of -0.04%, compared to 8.5% of non MLP underpricing over the same sample period, supporting Rock’s winner’s curse. However, Rock’s theory might be a good approach if one is seeking to explain the presence of underpricing between 5-10%; albeit the winner’s curse might be not useful to explain why underpricing of 15% or more occurs, observable during the dot-com bubble (Eckbo, 2008). Also, Wasserfallen and Wittleder (1994) question the usefulness of Rock’s model for the German stock market, since overpricing is rarely found in their study and, if it does occur, the amount is quite small. Therefore, they conclude that protection of uninformed investors from overpriced IPO stocks through average underpricing does not seem to be required. In a different approach, Beatty and Ritter (1986) tested the winner’s curse by underpricing against ex-ante uncertainty. They modelled the information production during the IPO pro-cess with option pricing. Valuing a company can be seen as buying a call option on the IPO stock. If the result of the valuation exceeds the offer price (strike) the option is exercised and the IPO stock is bought at the offer price. Considering the fact that the value of a stock op-tion increases if the volatility increases (ceteris paribus) the “IPO call opop-tion” increases in value the greater the valuation uncertainty. Hence, if valuing the company will be worth more, because the uncertainty of the valuation is higher, more people will become well in-formed. This consequently leads to a higher underpricing, as the winner’s curse problem is worsened due to higher information asymmetry. Thus, the winner’s curse theory is support-ed if underpricing increases in valuation uncertainty.

Information Revelation Theory

The centre of the revelation theory is based upon the way in which informed investors re-veal their information about the assumed value of an IPO stock. Regarding the assumption that some investors are better informed than other investors and the issuer, the key task for the underwriting bank is to obtain the information from the informed investors before the

12 price is set. However, informed investors have no incentive to reveal their information be-cause this would lead to a loss in underpricing for them (Benveniste and Spindt, 1989). The challenge for the underwriter is to design a mechanism, inducing investors to expose their information truthfully and trustworthily. One solution to this problem can be the book-building process, as stated by Benveniste and Spindt (1989) and Spatt and Srivastava (1991). During the bookbuilding period, the underwriter can collect enough information from inves-tors to allocate the shares. Invesinves-tors with low bids will obtain few shares, whilst invesinves-tors with high bids will obtain many shares. Consequently, investors that gave misleading infor-mation are left with few shares. Moreover, if the IPO stock is underpriced, the investors have the incentive to reveal their true information, since they want to participate in the IPO (Eckbo, 2008).

To test the Information Revelation Theory, Cornelli and Goldreich (2001, 2003) studied the IPO books of leading underwriting banks in 37 cross-border IPOs outside the United States. They examine two different types of bids: market orders and price-limited bids. Price-limited bids receive a 19% higher allocation than investors that place market orders. These findings support the Information Revelation Theory, because the revelation of useful information is rewarded. Nevertheless, Jenkinson and Jones (2004) use a different database consisting of 27 IPOs managed by European banks, and find less supportive evidence for the Information Revelation Theory.

In a study conducted by Hanley and Wilhelm (1995), the researchers show that institutional investors are being rewarded to a greater extent than retail investors. Their results are based on the observation of 38 U.S. IPOs from 1983-1988, and indicate that institutions are allocated 66.8% of the average IPO. According to Hanley and Wilhelm (1995), better in-formed investors are rewarded with higher allocations for superior information. However, their results show that underpricing is necessary for the investor to gain from the large allo-cation. Similarly, Jenkins and Ljungqvist (2001) found that institutional investors receive greater allocations by about 50%-66.6% internationally when banks have discretion in how the stocks are allocated.

The Information Revelation Theory claims that if the offer price of an IPO is revised upwards, positive information has been revealed. But according to Hanley’s (1993) Partial Adjustment

13 Theory, positive information will only partially adjust the offer price. Therefore, if the offer price is revised upwards, the IPO underpricing should be larger than if the offer price has not been revised. The investor is compensated for revealing his information, enabling the issuer to revise the price partly upwards.

Principal-Agent Theory

IPO underpricing represents a wealth transfer from the issuer to the investor; hence inves-tors compete for allocations of underpriced stocks. This competition is observed to be influ-enced by payments from the investor to the underwriting bank. Examples of side-payments are excessive trading commissions, paid for future, unrelated transactions (Loughran and Ritter, 2002). Additionally, underpriced stocks were given to executives of other companies in the hope to win future projects (spinning). Nevertheless, all the afore-mentioned practices are in conflict with the goal of the issuer, who wants to maximize his proceed from the IPO.

Baron (1982) associates the underpricing of IPOs with a second-best solution in a principal-agent model. He argues that the issuer delegates the pricing decision of the IPO to an in-vestment bank that is better informed about the investors demand. This superior infor-mation is then used by the underwriting bank to select a contract of IPO prices. The IPO price then depends upon the demand for the new stocks. If the demand is low, the offer price will also be low. This optimises the (unobservable) selling effort by making it depend-ent on market demand. In comparison to a world with symmetric information and a first-best solution, the second-first-best contract that is incentive-compatible involves underpricing and allows the underwriter to capture positive rents.

Boehmer and Fishe (2001) find that underwriters have an incentive to sell stocks too cheap, since the underpricing leads to a higher trading volume in the aftermarket and more liquidi-ty. Nevertheless, contradictory studies by Muscarella and Vetsuypens (1989) find that the Principal-Agent Theory cannot explain underpricing exclusively. They look at 38 IPOs where the investment bank is the underwriting bank for their own IPO and hence no principal-agent problem should be present. However, the IPOs show that underpricing is still present as in any other IPO, even without the occurrence of principal-agent problems.

14

Signalling Theory

Core to the Signalling Theory is the idea that underpricing is a signal of the firm quality. The theory is based on the idea that the issuer possesses superior information about future cash flows and prospects compared to the underwriting bank. Therefore, underwriting is a means used to signal the company’s high value with high first day return, even if underpricing is costly on a first sight (Eckbo, 2008). For instance, the (high-quality) firm uses underpricing as a means to signal quality, since low-quality firms cannot underprice because their firm value would not appreciate in the secondary market, and it would be too costly for them. The high-quality firms will recoup their upfront underpricing in the future by, for example, fa-vourable dividend announcements (Allen and Faulhaber, 1989) or induced analyst coverage due to high initial returns (Chemmanur, 1993). It remains questionable why a firm should use underpricing as a means to signal quality to the market. On the one hand, underpricing is costly for the firm. On the other hand, cheaper means (e.g. a positive dividend announce-ment) exist to signal a high firm quality.

Moreover, Michaely and Shaw (1994) and Jegadeesh et al. (1993) show that underpricing is unrelated to future equity offerings, hence contradicting the signalling explanation for IPO underpricing. Furthermore, evidence against the signalling theory was also given by Es-penlaub and Tonks (1998) on the UK IPO market between 1986 and 1991.

2.4 Institutional Theory

In general, the institutional theory concentrates on three distinct institutional features that are commonly observable in the IPO market, such as: bank’s price stabilising activities, taxes, and issues related to litigation.

Bank’s Price Stabilisation

The underwriter can take several actions when trading of the IPO share begins in the after-market. In a controversial research paper by Ruud (1993), the author claims that underpric-ing is an unintentional side-effect of the biddunderpric-ing effort by the underwritunderpric-ing bank. Studyunderpric-ing 463 IPOs from 1982 to 1983, Ruud argues that underwriters bid on the stocks to stabilize the price if the IPO price is likely to fall below the offer price. Regarding the return distribution,

15 the left tail would hence be eliminated, causing the mean return to increase. If that would not be the case, the mean return would be close to being zero, which would result in no un-derpricing. Ruud argues that underpricing peaks sharply at zero but seldom falls below zero. Hence, empirical results may not be the unconditional expectation of the true initial returns, but the mean conditional initial return upon underwriter intention in the secondary market. Ruud’s view was heavily criticized (DeGeorge, 1995) and shows little support from other ac-ademics. Additionally, the lack of available data limits the scope of empirical evidence of the impact on price stabilizing bids by the lead underwriter. Oftentimes, only market regulators have exclusive access to this kind of data; other players who are involved in the IPO process are excluded.

Tax

Supporters of the tax theory claim that underpricing is advantageous from a tax point of view. The rationale behind that theory is as follows: if the salary is more heavily taxed than capital gains, an incentive is created by paying employees with appreciating assets, such as an underpriced stock.

In a study focussing on the Swedish stock market, Rydqvist (1997) shows a relationship be-tween the tax rate and IPO underpricing. He finds that an increase in the tax rate for appre-ciating assets leads to a decrease in underpricing. Regarding the U.S. market, Taranto (2003) found similar results.

All in all, it can be concluded that tax benefits from underpricing may help to examine cross-sectional underpricing returns. Nevertheless, it is unlikely that tax alone can explain the phenomenon of IPO underpricing (Eckbo, 2008).

Legal Insurance

The underlying idea of litigation as a reason for underpricing dates back to Logue (1973) and Ibbotson (1975), and was further outlined by Tinic (1988), as well as Hughes and Thakor (1992). These findings claim that underpricing is a means to avoid the likelihood of future lawsuits from disappointed shareholders against the managers.

16 In 2002, Lowry and Shu found a simultaneous correlation between litigation risk and pricing. According to their lines of argumentation, issuers choose a certain level of under-pricing to mitigate lawsuits and litigation, and the level of underunder-pricing is related to the probability of litigation. Using a two-stage least square method, Lowry and Shu (2002) show that the level of underpricing is positively related to the probability of being sued.

However, the risk for managers being sued by disappointed shareholders is often empirically proven to be statistically insignificant and more U.S. centric, whereas IPO underpricing is a global phenomenon and therefore excluded further on (Ljungqvist, 1997; Drake and Vet-suypens, 1993).

2.5 Control Theory

The Control Theory focuses on the managers’ behaviour in using underpricing as a tool to strategically select appropriate investors and to allocate control over the company (Jensen and Meckling, 1979). Supporters of the Control Theory state that underpricing is a means to influence the shareholder base to limit intervention by outside parties.

Underpricing as a Means to Retain Control

The core of this theory is that underpricing is a means to retain control after the company went public by creating excess demand with a low initial stock price. Brennan and Franks (1997) claim that underpricing enables the managers to obtain a widely dispersed share-holder group and hence maintain their private benefits. The underpricing results in an in-creased demand by retail investors, which consequently leads to smaller stakes for every investor and a lower level of monitoring (Shleifer and Vishny, 1986). Furthermore, Booth and Chua (1986) also support the underpricing theory of Brennan and Franks, but find different motives for underpricing. They claim that managers desire a more dispersed ownership structure because this causes more liquidity in the secondary market.

Underpricing as a Means to Reduce Agency Costs

Given the case that managers have substantial stakes in their company, they might want to reduce the agency costs, provided that the agency costs are higher than their control bene-fits.

17 Soughton and Zechner (1998) claim that due to the monitoring incentive of a large investor, the manager might prefer to allocate the shares to a single large outside investor instead of various small investors. Subsequently, the large investor would reduce the agency costs through its monitoring effort, at the same time reducing the private benefits of control. As long as the manager’s agency costs are reduced by more than the private benefits, the man-ager has the incentive to allocate his shares to a single outside investor. The incentive for the investor is in turn created towards underpricing. Consequently, IPO underpricing might be the result of a manager who tries to reduce his agency costs.

2.6 Behavioural Theory

First day returns of newly issued shares increased materially in the late 1990s (dot-com bub-ble). As a result many researchers became doubtful about whether IPO underpricing can be fully attributed to informational frictions. For instance, between 1999 and 2000, USD 62 bil-lion were left on the table by issuing firms (Eckbo, 2008).

Behavioural theories build the next group of explanations for underpricing in this thesis. In comparison to the aforementioned theories, they are still an immature area of research. At the centre of the behavioural theories are the assumptions that either irrational investors bid up the price of IPO shares beyond their actual value, or that the issuers behave irration-ally, which consequently implies a failure in putting pressure on the investment banks to reduce underpricing (Eckbo, 2008).

Cascade Models

Welch (1992) shows the presence of cascade effects if investors make their investment deci-sions sequentially. Welch (1992) states that, if an investor enters the IPO at a later point in time, he bases his valuation on the bids of earlier investors and disregards his own infor-mation. In other words: the investor optimally ignores his own information and imitates ear-lier investors. For instance, a positive cascade is formed if an investor makes a bid and com-mits to his bid. Since the later investors follow the earlier investors, the earlier investor can demand a benefit in terms of underpricing. Hence, cascades might be an explanatory ap-proach for IPO underpricing.

18 Welch (1992) argues that in general, issuers are better off if cascades are formed, because free information would maximize the issuer’s disadvantage in comparison to investors which would recognize the true value of the entity. However, this theory is in general limited if the underwriter keeps the bidding information secret, as it is the case if the bookbuilding meth-od is used. The bookbuilding methmeth-od emphasises the importance of underwriters during the IPO process but also increases the agency problem.

Welch’s (1992) approach has gained little empirical attention. This might be due to limita-tions arising from the discrete bookbuilding method or a lack of profound data.

Investors Sentiment

Investor Sentiment Theories focus on the influence of investors’ sentiment and other irra-tionalities on the IPO market. IPO firms are prone to investors’ sentiment because their stocks are fundamentally more difficult to value, since they are relatively young, immature, and lack some crucial information (Berk and DeMarzo, 2014). Ljungqvist et al. (2006) state that sentiment investors hold too optimistic beliefs about the future development of IPO firms. The objective of the IPO firm would be to issue as many shares as possible to exploit the overly optimistic sentiment, but this would go on the expense of the stock price. Ljungqvist et al. (2006) state that the IPO firm sells the stock to institutional investors who then gradually sell the stock to sentiment investors to keep the price up. The risk of selling the stock gradually is compensated by underpriced IPO shares, since the stock price might decrease during the holding period.

Ofek and Richardson (2003) support the findings of Ljungqvist et al. (2006), and point out that abnormal initial returns occur when IPO shares are sold from institutional investors to retail investors on the first day of trading.

Dorn (2002) uses German data on IPO trading by 5000 retail customers and finds that IPO stocks are passed from institutional investors to retail investors, which could support the view that institutional investors restrict supply to keep the stock price high. Furthermore, he documents that retail investors overpay for IPOs following high periods of underpricing and for IPOs that are remarkably advertised. Cook et al. (2006) also document a significant posi-tive relationship between promotional activities and IPO stock valuations.

19 Da et al. (2011) investigate another explanation approach for sentiment-driven first day re-turns. They point out that IPO stocks gain high attention from retail investors, which leads to a retail investor buying pressure on the first day of trading, therefore driving the price up.

Prospect Theory and Mental Accounting

In 2002, Loughran and Ritter propose an alternate behavioural explanation approach for IPO underpricing. They state that the underpricing is a result of a behavioural bias among the managers of the firm instead of a bias among the investors. Loughran and Ritter (2002) ar-gue that managers accept underpricing because of the price and therefore wealth increase of the retained shares after the first day of trading. Hence, if the issuer is biased by mental accounting (Thaler, 1980), he offsets the money left on the table with the appreciation of the retained shares. If the gain exceeds the underpricing loss the manager is satisfied with the investment bank’s performance.

In 2005, Ljungqvist and Wilhelm test whether the managers make subsequent decisions based on their perceived performance of the IPO. As a proxy, they analyse the hiring rate of the IPO underwriter to lead later seasoned equity offerings. Their results show that if man-agers are satisfied with their underwriter’s performance, they are less likely to hire a differ-ent underwriter in a SEO. Hence, their results lead to the conclusion that there exists ex-planatory power in the behavioural model.

2.7 Reputation of Underwriter

Loughran and Ritter (2004) test three different models that should explain increased pricing over the past decades. The first two hypotheses are not directly related to under-writer reputation, since they focus on risk characteristics of IPO firms in a way that the pro-portion of riskier firms increases and therefore also average underpricing (Ritter, 1984), and on reduced ownership of executives of IPO firms leading to higher ownership fragmentation and finally increased acceptance of underpricing (Ljungqvist and Wilhelm, 2003). Both mod-els are especially interesting for developments in the 90’s and the dot-com-bubble in the early 2000’s, where underpricing dramatically increased.For both models, the authors are not able to record empirical evidence, which is why they conclude that these are not able to explain underpricing, especially the sharp increase in the early 2000’s.

20 Instead they develop a third hypothesis, which also includes the underwriter side as one driver for underpricing, and is called the “changing issuer objective function”. The basic thought behind their model is that issuing firms are more willing to accept underpricing. The proposed reason for this phenomenon can be grouped into the following two hypotheses: The analyst lust hypothesis and the spinning hypothesis.

Analyst Lust Hypothesis

IPO firms increasingly put efforts on analyst coverage. In doing so they aim to get coverage from bullish analysts with a high reputation, meanwhile they disregard to avoid underwriters that aggressively underprice IPOs. The above mentioned is called the “analyst lust hypothe-sis”. Analyst coverage is seen as important for these firms for different reasons. On the one hand, the incremental growth rate assumption for the valuation of firms becomes more rel-evant, since IPO valuations increase. The relative value of these growth opportunities in-creases against the value of existing assets, leading to a higher relevance of analyst cover-age. On the other hand, the development in telecommunications and media has improved the availability of analysts’ recommendations. Thus, it becomes more interesting to have coverage form highly reputable analysts. Various authors deliver supporting empirical evi-dence for the link between high reputation analysts and a better standing of investment banks in the market for equity underwriting services. Among others one should mention Dunbar (2000) and Cliff and Denis (2004).

Spinning Hypothesis

The “spinning hypothesis” constitutes the second part of the approach of Loughran and Rit-ter (2004). This hypothesis states that IPO firms executives and venture capitalists partici-pated in IPOs with high first-day returns via personal brokerage accounts at the underwriting bank. Since the lead underwriter also allocates large parts of the offered shares, it occurred that executives and venture capitalists of the firms received a part of these shares. Once there is severe underpricing in place, it will lead to capital gains on the personal accounts of these actors and therefore to substantially higher personal wealth. Ergo it becomes obvious that executives and venture capitalists, which are capable to choose the lead underwriter, have an incentive to employ investment banks with a reputation for high underpricing. Since

21 these decision makers are more interested in their own monetary wealth, than maximizing proceeds to the firm, this principal-agent problem leads to higher average underpricing. But why do investment banks underprice at all? Loughran and Ritter (2004) assume that the lead underwriter benefits from high first-day returns. Firstly, once an underwriter receives fees not only from the issuer side, but also from the demand side of the market, i.e. the in-vestors, he would have an incentive to choose low offer prices in order to market the shares easier. Secondly, the underwriter might receive commissions out of trades by the investors in the secondary market after the IPO. Hence, the underwriter would indirectly profit from high initial returns if this commission consists of a percentage value of the investor’s trading profits. Last but not least, banks are interested in future revenues from their customers, and thus allocate other underwritten IPOs with high underpricing mostly to clients with high ex-pected revenue streams.

All in all these hypotheses build the basic framework, which could explain higher under-pricing by Top-Tier investment banks. Banks with a reputation for high quality analysts and severe underpricing should increase their market shares and finally obtain higher reputation values, since issuers have an incentive to choose this kind of underwriters. Empirical evi-dence supports these models in a number of ways. Ultimately, Top-Tier investment banks would leave more money on the table than their Non-Top-Tier counterparts.

3.

Methodology

In the previous chapters, the rationale for firms to issue shares on the primary market as well as several theoretical frameworks and previous empirical findings for IPO underpricing have been discussed to provide a possible explanation for the underpricing phenomenon. The following chapter states the different research hypotheses, data and methodology of this thesis.

3.1 Dataset and Sample

The following shows up the methodological framework I used for my analyses, which is in large parts based on prior work of Loughran and Ritter (2004). This is for the reason that the

22 authors provide evidence of the relationship between underpricing and the reputation of an underwriter for the period from 1980 until 2003. In order to be in line with recent literature and scientific standards, I extend the relevant parts of their analyses for the period from 2004 until 20114, while primarily following their methods.

A sample of 905 IPOs from the U.S. for the years 2004 until 2011 builds the basis for the analyses that follow in the consecutive chapters. The data source is a deal report from Thomson One, which amounts to a total of 2958 IPOs in the U.S. for the same period. Based on this dataset the above mentioned sample of 905 IPOs remains after eliminating IPOs for which no return data is available and applying very similar filters as Loughran and Ritter (2004) did. Therefore, I exclude IPOs with an offer price below USD 5.00 per share. Further data such as share prices (offer and close), number of underwriters, number of offered shares, underwriter information, company data, industry classifications and return data for market indices are also provided by Thomson One and are directly allocated to respective deals via unique deal numbers, ensuring the consistency of my dataset. Unlike Loughran and Ritter (2004), I will not exclude REITs in my dataset. Industry groups for internet, technology firms and REITs are built using mid description and macro description classifications, respec-tively. The definitions are based on Loughran and Ritter (2004) but deviate, since both au-thors use SIC codes for technology firms, which do not fully match the same wording and allocation compared to classifications of Thomson One. Nonetheless they should be close enough. The detailed listing of all classifications for the internet- and technology-industry groups can be found in the appendix.

In the next passages I will first define how underpricing is measured and consecutively pre-sent the reputation measure that clusters banks into Top-Tier (TT) and Non-Top-Tier groups (NTT).

3.2 Measurement of Underpricing

Underpricing is defined and measured by the percentage price change from the offer price to close price of the very first trading day (I often refer to this as the “initial return” or “first-day-return” in the following) which is given by the following formula:

4 _{My analyses are due to 2011 since the Carter-Manaster ranks of underwriter reputation are available until}

23 𝐼𝑅𝑖,𝑡,𝑑 = [

𝑃_𝑖,𝑡

𝑃𝑖,0− 1] × 100

Where

𝐼𝑅𝑖,𝑡,𝑑: The initial return of stock i at IPO date t for the first day d. 𝑃_𝑖,0: The offer price of stock i

𝑃_𝑖,𝑡: The closing price of stock i at the first trading day after the IPO

With regard to prior literature I also use daily raw returns with no adjustment for market movements, for the analyses presented in the ensuing chapter. This has two main reasons. First, I want to stick to the methodology of the previously mentioned anchor paper by Loughran and Ritter (2004). The second reason is the average magnitude of returns I record for the first trading day of IPO stocks, which is much higher than average one day market movements. Already the very first specific paper on underpricing by Ibbotson (1975) shows that market returns are not able to explain much of the observed initial returns, using a two parameter model. Moreover, the main regression results I obtain are robust against adjust-ments for market moveadjust-ments and are shown in the appendix of my thesis.

3.3 Measurement of Underwriter Reputation

Once initial returns are measured, the next step would be to allocate all IPOs to either the Top-Tier or the Non-Top-Tier underwriter segment. To do so, I adopt underwriter reputation ranks initially established by Carter and Manaster (1990) and later updated including few adjustments in their methodology by Jay R. Ritter, which are publicly available on his web-site.5 The idea behind this ranking system is to assign a rank between 0 and 9 to each in-vestment bank based on the reputation it enjoys as an underwriter, where 9 is the best and 0 the worst rank a bank can obtain. These ranks are generated on the basis of tombstone announcements for IPOs. Carter and Manaster (1990) assume a strict hierarchical structure for the investment banking industry as a whole that is reflected in the order, associated banks of the IPO are listed within these announcement. Therefore Top-Tier investment

5 _{Data can be downloaded from the following link: https://site.warrington.ufl.edu/ritter/ipo-data/. The name of}

24 banks are likely to be listed in the upper brackets of the announcement, whereas Non-Top-Tier banks are placed in a lower section of the list of syndicated underwriters. A list of all lead underwriters in my sample and their Carter-Manaster ranks can be found in the Appen-dix.

So far the reader should understand the basic thought: the higher the rank, the better the reputation of the underwriter in this model. Following the logic of Loughran and Ritter (2004), I define banks with ranks of 8 and more as Top-Tier underwriters (TT) and banks with ranks lower than 8 as Non-Top-Tier underwriters (NTT). In case multiple investment banks are listed as lead managers, I assume the first mentioned bank to be the relevant lead un-derwriter for the reputation based allocation of IPOs to the respective groups. The distribu-tion of the number of deals that are allocated to these two groups is asymmetric. My total sample of 905 IPOs, counts 706 deals for the Top-Tier group, while the remaining 199 deals form the Non-Top-Tier group. This feature already indicates a balanced concentration in the market for equity underwriting. Market share based measures for underwriter reputation, as the Megginson-Weiss system, are not used here. Cooney et al. (2004) identify two potential drawbacks for measures built on market shares of underwriters. On the one hand market shares can vary rapidly from one year to another, even though it is not clear why the pres-tige of an underwriter should change at the same pace. On the other hand above mentioned measures implicitly assume that market shares are equivalent to underwriter prestige. This assumption is critical as, for instance, niche players might enjoy a good reputation in the market while they execute deals much more selective than global fully fledged investment banks. Nonetheless, the same authors also expose some handicaps for the Carter-Manaster ranking system based on which Jay R. Ritter provides reputation ranks that I adopt for my analyses. Most notably the strict hierarchical structure that is assumed has an important implication that might uncover some inconsistencies in this framework. If a strict hierarchy in the investment banking industry holds and the individual players defend their positions, no bank in the same brackets of a tombstone announcement with different Carter-Manaster ranks should exist. This in fact is not the case, as shown by Cooney et al. (2004). Further-more, empirical results from the initial Carter-Manaster methodology might be biased to-wards penny-stock underwriter. A bank that underwrites offerings only syndicated by other low reputation banks, can easily obtain a high reputation rank, even if it is commonly known

25 to be non-prestigious. This might be the case for some penny-stock offerings. Nevertheless, Loughran and Ritter (2004) already controlled for this side effect via adjustments in their methodology once they updated the original Carter-Manaster rankings from Carter and Manaster (1990) and Carter et al. (1998) for more recent periods. Regardless of few poten-tial inconsistencies in the ranking system I deploy, there is no implication that these ranks are incorrect or inadequate. They only question is if the initial assumption of a strictly hier-archical structure in the investment banking industry holds, since also upgrades from lower to higher Carter-Manaster ranks appear.

4.

Empirical Results and Analysis

The previous abstracts should have shown the reader how I measure underpricing in general on the one side and how I cluster all IPOs of my sample into Top-Tier and Non-Top-Tier un-derwriter segments on the other side. What follows, presents the results I obtain in applying this methodology to my dataset. The next chapter first provides some information on the IPO activity in the U.S. in general for the years 2004 until 2011 and afterwards provides fur-ther evidence for a positive relationship between underwriter reputation and underpricing.

4.1 Descriptive Statistics

The number of IPOs varies across the sectors. Figure 1 gives the number of IPOs per sector from 2004 to 2011. Most of the IPOs occurred in the high technology sector with 169 IPOs, followed by 158 IPOs in the healthcare sector and 154 IPOs in the financial sector. The num-bers of IPOs in the other sectors are continuously falling, whereas the material industry showed the lowest number of IPOs, with only 40 over the sample period.

26 Figure 2: Number of IPOs per sector in the U.S. from 2004 – 2011

IPO activity in the U.S. varies substantially between 2004 and 2011, showing up a dynamic period in the run up to the financial crisis prior to the year 2008, followed by modest years. Although the recovery from the crisis after the Lehman bankruptcy seems to be underway, the number of IPOs remains far lower than pre-crisis levels. To be more precisely: I record on average 163 IPOs per year for the period 2004-2007, followed by 26 and 39 IPOs for the years 2008 and 2009 respectively, rebounding to a mean of 95 IPOs for 2010 and 2011. Fig-ure 2 shows the number of IPOs for each year from 2004 to 2011, as well as the average first-day return I record for the corresponding year.

High Technology; 169

Healthcare; 158

Financials; 154 Energy and Power;

98 Consumer Products

and Services; 76 Industrials; 61

Real Estate; 59 Telecom, Media and

Entertainment; 48

Retail; 42

27 Figure 3: Number of IPOs in the U.S. from 2004 – 2011 and average first-day return

Average first-day returns are 10.8% for the whole sample presented above. This finding is roughly in line with data from Jay R. Ritter’s analyses, which state an average first-day return of 10.9% for the period under observation.6 Once we take a closer look at individual years, it seems to be puzzling that, even though the number of IPOs clearly diminishes from 2007 to 2008, there is also sharp decline in underpricing from 13.9% in 2007 to 4.9% in 2008, while the number of IPOs for 2009 goes up slightly but the underpricing rises sharply back to the level of 2007. On a market-return adjusted basis, average first-day returns are even higher at 5.1% and 14.1% respectively, thus indicating that average market performance on the IPO dates was negative. Furthermore, over the whole period market-returns seem to have little impact on the underpricing of IPO stocks, since average as well as median first-day returns remain nearly at the same levels after a market return adjustments. This again could ques-tion, whether the observed returns can be explained by exposures to market risk factors. Nevertheless, this should not be the point here. Rather should the reader take into account that IPO stocks seem to realize relatively high returns on their first trading day over longer periods of time, while market returns are far less. The next step would be to investigate pos-sible determining factors for the magnitude of underpricing, which in turn might point to

6

Jay R. Ritter provides data on IPOs in the U.S. on his website with the following link: http://site.warrington.ufl.edu/ritter/ipo-data/. Data is located in the section “IPO data” and there in the sub-section “IPOs 2013 Underpricing”.

0% 2% 4% 6% 8% 10% 12% 14% 16% 0 50 100 150 200 250 2004 2005 2006 2007 2008 2009 2010 2011 Number IPOs Average first day return

28 possible explanations for this phenomenon. Previously presented results are shown below in table 1 on a year-by-year basis.

Table 1: Overview of the IPO activity in the U.S. from 2004 – 2011

The sample contains all IPOs after subtracting offers as described in chapter 3.1. First-day returns are defined as percentage change from the offer price to the first trading day closing price. Market adjusted returns are calculated, subtracting the S&P500 return from the first-day return at each respective IPO date. Money on the table is defined as the absolute USD price change at the first trading day multiplied by the offered shares.

Year No. of IPOs First-Day Return First-Day Return (market

adjusted)

Money on the table

Mean Median Mean Median Mean

2004 2005 2006 2007 2008 2009 2010 2011 193 147 145 165 26 39 101 89 10.9% 8.8% 12.2% 13.9% 4.9% 13.7% 7.1% 11.1% 5.8% 4.3% 5.6% 4.7% -0.6% 7.9% 3.8% 4.6% 10.9% 8.8% 12.1% 14.0% 5.1% 14.1% 7.0% 11.1% 5.7% 4.6% 5.4% 4.9% -1.7% 8.1% 3.9% 4.6% USD 17.1m USD 13.2m USD 22.6m USD 25.8m USD 16.1m USD 69.6m USD 6.1m USD 25.0m All 905 10.8% 5.0% 10.8% 4.7% USD 20.3m

The observed firms leave on average USD 20.3 million on the table, symbolizing the mone-tary impact of underpricing in dollar values, as firms may miss to collect potential proceeds from their IPO. If the case were true that IPO stocks are issued at too low offer prices to re-flect the firm value adequately, this would simply mean that firms do not raise the amount of capital they could have. The difference in post-issue valuations using offer and first-day close prices of stocks again underlines the suspicion that offer prices on average do not re-flect the intrinsic firm value.

However, it is essential to not to regard these figures as ex-ante evidence for mispricing of IPO firms by banks, since these returns could still be some kind of risk premium. Also would it be a hasty conclusion to state underpricing as abnormal return and even more to question market efficiency as defined by Fama (1970) in some way.7 Hence, the question would rather

7 _{See FAMA, EUGENE F. (1970) for definitions of market efficiency. Ibbotson (1975) and Schwert (2002) test}

29 be how different factors might influence underpricing. Once, these factors are explored in detail, possible approaches that explain underpricing could be formed.

The choice of the lead underwriter might be one important starting point to investigate a firm’s IPO in more detail. Especially in regard to underpricing, one could ask whether Top-Tier banks for underwriting services, price firms differently than those considered to be Non-Top-Tier. Do Top-Tier underwriters price shares of firms more accurately? Or do they under-price more to benefit from high returns on the first trading day as some kind of indirect compensation? Finally, to put it in a nutshell: Does the choice of the lead underwriter have a significant impact on first-day returns of IPO stocks? This question will be the core question of the analyses that follow in the consecutive passages. In chapter 2.7, I have already pre-sented some explanation approaches that provide a rationale to the results I obtain. Yet these should not be the main focus of my thesis here, as this would go beyond my scope. In order to show up some initial differences between IPOs conducted by the TT group and the NTT group of underwriters, I sort my sample using the reputation ranks for the lead un-derwriters of each IPO. Thereof, I obtain two sets of IPOs that are considered to be a) Top-Tier IPOs and b) Non-Top-Top-Tier IPOs. Table 2 describes both groups using different parame-ters that are later used in my regressions. Already at this point some interesting differences show up for both groups of IPOs. First of all, the average first-day return for Top-Tier offer-ings is significantly higher than the one of their counterparts. My sample records an average first-day return of 11.7% for Top-Tier IPOs versus 7.3% for the other group, thus indicating that the choice of lead underwriters matter for underpricing. This result is further confirmed, using median first-day returns as measure. First-day returns are on average 4.4 percentage points higher for IPOs with high reputation underwriters, which is equivalent to 60% higher underpricing versus Non-Top-Tier underwritten offerings. At this point one could claim that the choice of the lead underwriter seems to have a direct influence on the magnitude of first-day returns, thus suggesting that Top-Tier investment banks underprice more.

However, one has to consider other firm specific features that might drive initial returns as some kind of risk premium. Therefore, other controls are included for both sorts of IPOs. In line with Loughran and Ritter (2004) I adopt variables which describe the sample also in a

still hold in regard to underpricing. Even though high first-day returns seem to be in place, it is difficult for an individual investor to participate in IPOs.