The underpricing and aftermarket performance of IPOs during the period 2001 – 2012

(1)

The underpricing and aftermarket performance of IPOs

during the period 2001 – 2012

Amsterdam Business School

Name

Marcha van der Boon

Student number

10218114

MSc in

Business Economics

Track

Finance

Number of ECTS

15 Supervisor

Ilko Naaborg

Completion

7 July 2016

Abstract

This thesis analyses the impact of the level of underpricing on the aftermarket performance of IPOs in the United States during the period 2001 to 2012. It makes a distinction between VC-backed, PE-backed and non-sponsored IPOs. The sample consists of 1368 IPOs in 48 different industries that are listed on NYSE, NASDAQ or AMEX. It consists of 510 VC-backed and 374 PE-backed IPOs. A multiple regression model and an ordinary least squares method are used to test the hypothesis. The aftermarket performance is the dependent variable and is measured by the 3-year buy-and-hold return. The underpricing is the main independent variable and is measured by the percentage difference between the offer price and the first day close price. The results show a positive, significant relation between underpricing and aftermarket performance. They also show that the underpricing of non-sponsored IPOs has the highest impact, followed by the underpricing of PE-backed and then VC-backed IPOs. However, these results are not significant. The results are more or less robust. However, in some years of the research period there are few companies that went public, resulting in a small sample size. In order to avoid the small sample size problem, future research can take a longer time period or could place the time period forwards. They can also include exchanges from other countries to generalize the results or to investigate possible differences.

(2)

Statement of Originality

This document is written by Marcha van der Boon who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

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

(3)

Table of Contents

1. Introduction ... 4

2. Literature review ... 6

2.1. IPOs ... 6

2.1.1. IPO firm characteristics ... 7

2.2. Underpricing ... 7

2.2.1. Asymmetric information ... 8

2.2.2. Symmetric information ... 10

2.2.3. Underwriter theory ... 10

2.3. Aftermarket performance ... 11

2.4. Venture capital and private equity ... 12

2.4.1. Venture capital ... 12

2.4.2. Private equity ... 13

2.5. Conclusions of the literature ... 13

3. Methodology, Hypothesis and Data ... 15

3.1. Methodology ... 15

3.2. Hypothesis ... 18

3.3. Data and descriptive statistics ... 19

4. Empirical results ... 25

4.1. Empirical results ... 25

4.2. Robustness check ... 30

5. Conclusion and discussion ... 35

References ... 37

Appendix A ... 40

Appendix B ... 41

Appendix C ... 42

(4)

1. Introduction

A private company that is in need of extra capital has several options to attract funds to develop their business activities. One way is to borrow money from banks and add debt to its balance sheet. Another way is to attract venture capital from an investment bank or venture capitalist or to raise capital by private equity firms (Berk and DeMarzo, 2011, pp. 771 – 773). Nevertheless, the private company has also the possibility to go public by using an initial public offering (IPO). An IPO is the first sale of common stock or shares by a private company to the public and is one of the most important decisions during the life of the company (Lewellen, 2006). It generates funds to invest in new projects and growth opportunities (Ritter, 1991).

If the offer price of the IPO is lower than the value of the new issues, there is some “money left on the table” by the issuers. This means that the IPO is underpriced (Ritter, 1991). Lewellen (2006) states that the IPOs of operating companies are on average underpriced in all countries. However, they say that the IPOs of non-operating companies, such as closed-end funds, are generally not underpriced. Ritter and Welch (2002) and Beatty and Ritter (1986) find an average first-day return of 18.8% and 19.8% respectively. In addition to this evidence, Loughran and Ritter (2004) find that the level of underpricing changes over time.

The aftermarket performance is defined by the price level performance of a newly issued stock after its IPO and begins on the first day of trading on the exchange. In recent years, academics pay special attention to the stock price performance of IPOs in the years after the offering, because there is considerable variation in the aftermarket performances of the IPOs year-to-year and across industries (Ritter and Welch, 2002). Several researchers find that IPO firms significantly underperform in the three to five years after going public. Ritter (1991) finds that the average holding period return is 34.47% in the years after going public. However, he finds that a control sample of some listed stocks produces an average holding period return of 61.86% over this same 3-year holding period.

Therefore, according to the literature, there are two puzzles regarding the pricing of IPOs. The first puzzle shows the short-run underpricing and the second puzzle shows the long-run underperformance. The short-long-run underpricing implies that IPOs provide significant abnormal returns on the first day of trading (Ritter and Welch, 2002). The long-run underperformance implies that the abnormal returns of these IPOs may be negative in the aftermarket (Ritter, 1991).

Although various researchers try to explain either short-run underpricing or long-run underperformance, they do not focus on the impact of the underpricing on the aftermarket performance. Therefore, this research investigates the following research question: “Does the

level of IPO underpricing have an impact on the aftermarket performance during the period 2001 – 2012?” This is interesting, because if the IPO market is not efficient in the short-run, there is

(5)

no reason to assume that it is efficient in the long-run. This means that the aftermarket stock price will not necessarily revert to the fundamental price. In addition, it is interesting to see what the effect of the underpricing at the beginning of the IPO is on the aftermarket

performance of the IPO at a later point in the future. This thesis studies the implications of the underpricing on the subsequent developments of the IPO firm. Although this is an interesting question, researchers have not investigated this yet.

An addition to this uninvestigated issue is the distinction between venture-capital-backed (VC-venture-capital-backed), private-equity-venture-capital-backed (PE-venture-capital-backed) and non-sponsored IPOs. This is interesting, because Levis (2011) finds obvious differences across these three groups. They are different in terms of first-day returns, industry classification, operating characteristics and market size. The findings in this thesis will be relevant for investors, issuers and underwriters, because they are the main stakeholders who are involved in the IPO. The results will show if the

uncertainties about a low offer price will be compensated.

This thesis uses a multiple regression model and an ordinary least squares method to test whether the level of IPO underpricing has a significant impact on the aftermarket

performance. The model includes two dummy variables to control for industry and year fixed effects. The data consist of 1368 IPOs that went public during the period 2001 to 2012 in the United States in 48 different industries. The sample consists of 510 VC-backed IPOs and 374 PE-backed IPOs. All IPOs are listed on NYSE, NASDAQ or AMEX. The IPO data are collected from different databases. The aftermarket performance is measured by the 3-year buy-and-hold return (Ritter, 1991). In order to calculate this, the monthly stock prices are collected from the database Datastream (Datastream, 2016). The underpricing is measured by the percentage difference between the offer price and the first day close price (Ritter and Welch, 2002), these data are collected from the database Thomson One (Thomson One, 2016).

This thesis is structured as follows. After the initial introduction section, the second section discusses the literature review, it includes the main theories in the existing literature and the empirical evidence found in line with these theories. The third section explains the methodology and shows the data sources, these are necessary for testing the hypothesis. Then, the fourth section presents the empirical results and some robustness checks. Finally, the fifth section presents the conclusion and discussion.

(6)

2. Literature review

This section includes the main theories in the existing literature and the empirical evidence found in line with these theories. The first section discusses IPOs in general, the second section discusses underpricing, the third section discusses the aftermarket performance and the fourth section discusses VC-backed and PE-backed IPOs.

2.1. IPOs

An IPO is the first sale of common stock or shares by a private company to the public. If a public company sells additional equity capital on the stock market, it is called a seasoned equity offering (SEO) (Lewellen, 2006). Researchers discuss both advantages and disadvantages of IPOs. Some advantages of going public are greater liquidity and better access to capital. Berk and DeMarzo (2011, pp. 776 – 777) state that companies that go public give their private investors the ability to diversify. In addition, public companies typically have access to much larger amount of capital through the public markets, both in the initial public offering and in subsequent offerings. However, a major disadvantage of an IPO is that when investors diversify their holdings, the equity holders of the corporation become more widely dispersed (Berk and DeMarzo, 2011, pp.776 – 777).

The main question for a company is the decision to stay private or go public. The decision to go public arise the following question “why do firms go public?” (Ritter and Welch, 2002). Berk and DeMarzo (2011, p. 771) say that the initial capital that is required to start a business is usually provided by the entrepreneur herself and her immediate family. However, few families have the resources to finance a growing business, so growth almost always

requires outside capital. Therefore, the main reason to go public is to attract outside capital and the desire to raise equity capital for the firm. In addition, they create a public market in which shareholders can convert some of their wealth into cash at a future date (Ritter and Welch, 2002). However, they also argue that increased publicity and other non-financial reasons are less relevant for most firms.

Some theories that empirical researchers mention about the decision to go public are the life cycle theories. First of all, Zingales (1995) observes that a potential takeover target has an advantage when it is public, because it is much easier for a potential acquirer to spot. Entrepreneurs of a public company can get a higher value in an acquisition than what

entrepreneurs would get from an outright sale. According to CFOs, the creation of public shares for acquisitions is the main motivation for going public (Ritter and Welch, 2002). Second, Chemmanur and Fulghieri (1991) find that IPOs can have more dispersion of ownership. Pre-IPO investors hold undiversified portfolios, allowing them not to pay as high a price as diversified public-market investors. Overall, a firm will be private early in its life cycle, but it becomes optimal to go public if the firm grows sufficiently large.

(7)

Other theories that empirical researchers mention about the decision to go public are the market-timing theories. Lucas and McDonald (1990) develop an asymmetric information model. In this model firms postpone their equity issue if they know they are currently

undervalued. If a firm’s value is too low according to a bear market, they will delay their IPOs until a bull market provides more attractive pricing. They also argue that firms avoid their IPOs in periods where few other good-quality firms go public.

2.1.1. IPO firm characteristics

Ritter (1991) demonstrates that IPO firm characteristics are important, because they are of interest for the long-run performance of the firm. He uses a sample of 1526 IPOs of common stock in the period 1975 to 1984. The IPOs in this sample are listed on NYSE, NASDAQ and AMEX. The firms on these exchanges have different characteristics and hence different aftermarket performances. He uses 3-year buy-and-hold returns in order to calculate and compare the long-run performances of these IPOs. One important firm characteristic that Ritter (1991) mentions, is the age of the company. The age and aftermarket performance have a strong monotone relation. He finds that young growth companies have a lower aftermarket

performance. Other influential characteristics are the total assets before the IPO and the proceeds of the IPO. Ritter (1991) finds that more total assets and higher offers have the worst aftermarket performance. Third, the offer price of the IPO is also relevant for the long-run performance, because smaller offerings are more speculative (Ritter, 1991). Finally, he argues that the long-run performance of IPOs varies widely across years and industries. He finds that financial institutions have the best long-run performance and oil and gas firms show the worst long-run performance.

2.2. Underpricing

The underpricing is defined by the percentage difference between the offer price and the first day close price (Loughran and Ritter, 2004; Ritter and Welch, 2002; Beatty and Ritter, 1986). The terms underpricing and first-day returns are used interchangeably. If the offer price of the IPO is lower than the value of the new issues, there is some “money left on the table” by the issuers (Ritter, 1991). Lewellen (2006) states that the IPOs of operating companies are on average underpriced in all countries. However, they say that the IPOs of non-operating companies, such as closed-end funds, are generally not underpriced. Ritter and Welch (2002) argue that the underpricing of IPOs appears to be a short-run phenomenon. They find an average first-day return of 18.8% in their sample of 6249 IPOs from 1980 to 2001.

Approximately 70% of the IPOs end the first day of trading at a closing price greater than the offer price and about 16% have a first day return of exactly 0. Beatty and Ritter (1986) also provide evidence that IPOs are underpriced. They argue that this underpricing is implemented

(8)

by the investment banking industry and report that in a sample of 5000 firms that went public during 1960 to 1982 in the United States, the average IPO was trading 19.8% higher than its offering price shortly after public trading started. However, Loughran and Ritter (2004) emphasize another phenomenon, namely that the level of IPO underpricing changes over time. They indicate that the average first-day return on IPOs of 7% in the 1980s doubled to almost 15% during 1990 to 1998 and jumped to 65% during the internet bubble years of 1999 to 2000. During 2001 to 2003 it reverts to 12%.

The average IPO underpricing is not explained by simple fundamental market

misvaluation or asset-pricing risk premia, because if diversified IPO first-day investors require compensation for bearing systematic or liquidity risk, second day investors should also require this premium. Nevertheless, they do not seem to require this premium (Ritter and Welch, 2002). Therefore, in order to understand the underpricing puzzle, they focus on the setting of the offer price. At this point the underwriter suppresses the interplay of supply and demand.

There are three general IPO underpricing theories in the literature. Ibbotson (1975) categorize theories of underpricing on the basis of whether asymmetric information or symmetric information is assumed. The next two subsections discuss the theories based on asymmetric information and symmetric information. The third subsection discusses the underwriter theory.

2.2.1. Asymmetric information

The asymmetric information theory can be divided into theories in which IPO issuers are more informed than investors and into theories in which investors are more informed than the issuer (Ritter and Welch, 2002).

The first category, where the issuer is more informed than the investor, shows a lemons problem for rational investors. A theory based on this assumption is the signaling model

(Akerlof, 1970). The problem is that the investor will not buy any shares, because he is not able to make a distinction between low- and high-quality issuers. However, the high-quality issuers have the possibility to signal their quality in order to distinguish themselves from the low-quality issuers (Ritter and Welch, 2002). If high-low-quality issuers sell their shares at a lower price than the market expects, they have a credible signal. Low-quality issuers will deter from

imitating this price, because they cannot afford such a low price (Ritter and Welch, 2002). High-quality issuers can afford this underpricing, because they expect to capture larger revenues through SEOs. SEOs have more favorable prices (Welch, 1989). In contrast, low-quality issuers do not signal, because they do not expect to recover their initial investment through

aftermarket SEOs (Ritter and Welch, 2002). In signaling models it is common that high-quality issuers will attempt to signal their quality by throwing money away, for example by leaving

(9)

money on the table in the IPO. In the end, only issuers with worse-than-average quality are willing to sell their shares at the average price (Ritter and Welch, 2002).

Several researchers find evidence in favor of the signaling theory, because the issuing firm will recover the lower initial income by conducting future equity issues. Welch (1989) presents the signaling model, which suggests that IPO firms use a multiple issue strategy. His sample consists of 1028 IPOs during the period 1977 to 1982. He finds that a third of all these IPOs had issued a SEO by 1987. Therefore, he shows evidence that many firms raise substantial amounts of additional equity capital in the years after their IPO. Jegadeesh, Weinstein and Welch (1993) also find consistent results for the signaling model. Their sample consists of 1985 IPOs during the period 1980 to 1986. They exclude IPOs issued after 1986, because they require data on SEOs for a three-year period after the IPO. They use a multiple regression model and find that firms with higher first-day returns are more likely to issue SEOs and to return with larger offerings. However, Michaely and Shaw (1994) reject the signaling theory. Their sample consist of 947 IPOs during the period 1984 to 1988. They also use a multiple regression model, but they find no evidence of either more SEOs or a higher dividend level for more underpriced IPOs.

The second category, where investors are more informed than the issuer, the issuer faces a placement problem. The issuer does not know the price the market is willing to bear. A theory based on this assumption is the winner’s curse model (Rock, 1986). Another assumption of Rock’s model is that some investors can become informed for a cost c, but investors who do not incur this cost are uninformed (Beatty and Ritter, 1986). Rock (1986) argues that the issuing firm has to set an offer price and then solicits purchase orders from the public at this price. If there is an excess demand for the issue, the shares are rationed. Rock (1986) says that informed investors will only submit purchase orders if the offering is underpriced. However, uninformed investors will submit purchase orders for both underpriced and overpriced issues. Therefore, both informed and uninformed investors will submit purchase orders for

underpriced issues. In this case, uninformed investors will be assigned only some of the shares that trade at a premium in the aftermarket (Rock, 1986). However, for overpriced issues uninformed investors are assigned all issues that trade at a discount in the aftermarket. This activity exacerbates an adverse selection problem for uninformed investors (Beatty and Ritter, 1986). Rock (1986) calls it a winner’s curse, investors are concerned that they will only acquire all the shares if they belong to the most optimistic investors.

Several researchers find evidence in favor of the winner’s curse model, because the uninformed investors are required in order to sell all offered shares. Since uninformed

investors require positive expected returns, issuers have to underprice their shares (Ritter and Welch, 2002). Koh and Walter (1989) test Rock’s model and use a sample of 70 companies listed

(10)

on the Singapore Stock Exchange (SSE) during the period 1980 to 1987. They have rationing information and by simulating the returns, they find that shares are on average underpriced.

2.2.2. Symmetric information

The symmetric information theory assumes that issuers and investors have the same information.

Tinic (1988) uses 204 IPOs issued before and after the Securities Act 1933 and argues that issuers underprice their shares to reduce their legal liability. Hughes and Thakor (1992) also find evidence for this and state that an offering that is underpriced is less likely to be sued. For instance, an IPO that starts trading at $35 with an offer price of $25 is less likely to be sued than if it had an offer price of $35. This is because the aftermarket price is more likely to drop below $35 than below $25. Nevertheless, Drake and Vetsuypens (1993) examine 93 IPOs and reject the legal liability theory. They assert that the underpricing of IPOs does not protect them from being sued. They claim that sued IPOs have higher instead of lower underpricing.

However, Lowry and Shu (2002) investigate 1841 IPOs during 1988 to 1995 and comment that issuers will underprice their shares more, if it is more likely that they will be sued in the future. Ritter and Welch (2002) think that underpricing is not a cost efficient manner to avoid

subsequent lawsuits. Keloharju (1993) studies 80 IPOs between 1984 and 1989. He finds convincing evidence to reject the legal liability theory, because there are similar levels of underpricing in countries in which U.S. litigation tendencies are or are not present.

2.2.3. Underwriter theory

The underwriter theory is based on the underwriter’s overall reputation and status. In the IPO market the underwriter is the intermediary between the issuer and the investors. An

investment-banking firm underwrites many offerings over time, allowing them to develop a reputation and earn a return on this reputation (Liu and Ritter, 2011). The reputation of the underwriter could have an impact on the level of IPO underpricing. Liu and Ritter (2011) argue that this reputation is based on underwriter quality, industry expertise and analyst coverage from influential analysts. They use a multiple regression model and a sample composed of 7319 IPOs in the United States during the period 1980 to 2008. They find that prestigious

underwriters are associated with more underpricing, because they are selling differentiated products. Beatty and Ritter (1989) implement a weighted least squares method and use a sample of 1028 firms in the period 1977 to 1982. They mention three necessary conditions to enforce whether it is interesting for an underwriter to underprice an IPO. The first condition states that the underwriter is uncertain about the aftermarket price of the stock, the second condition states that the underwriter has to maintain his reputation and the third condition states that the reputation will be destroyed if the underwriter underprice too little or too much

(11)

(Beatty and Ritter, 1989). They also find that underwriters will lose market share if they overprice or underprice their shares excessively.

2.3. Aftermarket performance

The aftermarket performance is defined by the price level performance of a newly issued stock after its IPO and begins on the first day of trading on the exchange. In recent years, academics pay special attention to the stock price performance of IPOs in the years after the offering, because there is considerable variation in the aftermarket performances of the IPOs year-to-year and across industries (Ritter and Welch, 2002). Ritter and Welch (2002) argue that, according to the efficient market, the aftermarket stock price should just reflect the intrinsic value of the shares. However, several researchers find that IPO firms significantly underperform in the three to five years after going public. This means that the abnormal returns of these IPOs may be negative in the long-run, they appear to be overpriced (Ritter, 1991).

Ritter (1991) discusses several arguments about the interest of the long-run

performance of IPOs. The first argument states that price patterns are of interest for investors. Ritter (1991) says that these price patterns may produce opportunities for active trading strategies to generate superior returns. The second argument states that nonzero returns in the long-run reject the informational efficiency of the IPO market. Shiller (1990) hypothesize that IPO markets are sensitive to trends that influence market prices. The third argument states that there is a large variation in the volume of IPOs over time. Ritter (1991) suggests the appearance of high volume periods corresponding to poor long-run performances, in this case issuers can take advantage of these “windows of opportunity” by timing new issues. The fourth argument states that the returns of investors in the aftermarket affect the cost of external equity capital for IPOs. The cost of external equity capital is lower, if returns in the aftermarket are lower.

In his research, Ritter (1991) uses 3-year holding period returns and a relative wealth ratio to measure the performance of IPOs. His sample consists of 1526 IPOs that went public in the United States in the period 1975 to 1984. He finds that the average holding period return is 34.47% in the three years after going public. However, a control sample of 1526 listed stocks produces an average total return of 61.86% over this same 3-year holding period. Therefore, IPOs underperform in the long-run. He claims that there is a significant variation in the underperformance year-to-year and across industries. He also supposes that the three main explanations for this underperformance are bad luck, risk measurement or over optimism. However, Brav and Gompers (1997) state that underperformance is not necessarily an IPO effect. They investigate a sample 4341 IPOs from 1922 to 1972 and use equal weighted returns. They argue that firms with similar size and book-to-market all perform poorly, regardless of whether they went public or not.

(12)

2.4. Venture capital and private equity

If venture capital and private equity firms go public, they are called VC-backed and PE-backed IPOs, respectively. Venture capital firms invest financial capital in risky enterprises. These enterprises are too risky in order to be eligible for bank lending (Berk and DeMarzo, 2011, pp. 771 – 772). A private equity firm invests in the private equity of operating companies. They use investment strategies that may include a leveraged buyout, venture capital and growth capital (Berk and DeMarzo, 2011, p. 773). The distinction between these firms is important, because they are different in terms of first-day returns, industry classification, operating characteristics and market size (Levis, 2011).

2.4.1. Venture capital

A private company must seek sources that can provide capital to finance a growing business. One way to attract funds is by venture capital firms; they can provide substantial capital for start-ups. A venture capital firm is a limited partnership that specializes in raising money to invest in the private equity of these young firms (Berk and DeMarzo, 2011, pp. 771 – 772). Venture capitalists have become more familiar since the successes of numerous VC-backed IPOs, such as Apple Computer, Genentech, Lotus Development, Microsoft, Intel and Federal Express (Jain and Kini, 1995). They suggest that venture capitalists are closely monitoring the development of their start-ups and find that the market reflects this monitoring-value in a higher valuation at the time of the IPO.

Lee and Wahal (2004) investigate a sample of 6413 IPOs and analyze the level of underpricing of VC-backed IPOs. They find that VC-backed IPOs have larger first-day returns than comparable non VC-backed IPOs. The average return difference is statistically and

economically significant and ranges from 5.01 percentage points to 10.32 percentage points in their sample period 1980 to 2000. This return difference is particularly pronounced in the bubble period (Lee and Wahal, 2004). However, Megginson and Weiss (1991) match 640 IPOs between 1983 and 1987 by industry and offering size and find significantly lower first-day returns for VC-backed IPOs compared to non VC-backed IPOs. They argue that venture capitalists demonstrate the true value of the firm and hence reduce underpricing. Barry et al. (1990) also find evidence for a negative relation between VC-backed IPOs and the level of underpricing. They focus on the monitoring role of venture capitalists in 1556 IPOs between 1978 and 1987 and conclude that IPOs have a lower level of underpricing if they are better monitored.

Brav and Gompers (1997) examine the long-run underperformance of IPOs and find that VC-backed IPOs outperform non VC-backed IPOs using equal weighted returns. Their sample consists of 934 VC-backed IPOs and 3407 non VC-backed IPOs during the period 1972 to 1992.

(13)

Krishnan et al. (2009) confirm this evidence, because they argue that VC-firms with better reputations invest in portfolio companies with better long-run post-IPO performance.

2.4.2. Private equity

Another way to attract funds to finance a growing business is to raise capital by private equity firms. These firms are organized much like a venture capital firm, but they invest in the equity of existing privately held firms rather than start-up companies. As a result, private equity firms tend to be larger and more established than venture capital firms (Berk and DeMarzo, 2011, p. 773). Many private equity firms use the IPO as an exit strategy, for example to exit buyout-backed firms (Bergström et al., 2006).

Unfortunately, there is no study that examines the level of underpricing of PE-backed IPOs. However, Muscarella and Vetsuypens (1989) indicate that a second IPO after a leveraged buyout (LBO) reduces information asymmetry and they argue that investors are more informed about the company’s value. Although this thesis is not about secondary LBOs, Muscarella and Vetsuypens’ research suggest that the level of underpricing will be lower.

Levis (2011) investigates the aftermarket performance of PE-backed IPOs compared to non PE-backed IPOs by using 3-year buy-and-hold returns. All 1595 IPOs in his sample are listed on the London Stock Exchange during the period 1992 to 2005. He finds that, in the three years after the public offering, PE-backed IPOs show better operating and market performance than otherwise similar non PE-backed IPOs and the market as a whole.

2.5. Conclusions of the literature

The main reason to go public is to attract outside capital and the desire to raise equity capital for the firm (Ritter and Welch, 2002). Ritter and Welch (2002) and Beatty and Ritter (1986) argue that there appears a short-run underpricing puzzle. They find an average first-day return of 18.8% and 19.8% respectively. In addition to this evidence, Loughran and Ritter (2004) find that the level of underpricing changes over time. In order to explain the underpricing of IPOs, the literature offers three general theories. The first theory assumes asymmetric information, the second theory assumes symmetric information and the final theory is about underwriters.

In recent years, academics pay special attention to the stock price performance of IPOs in the years after the offering (Ritter and Welch, 2002). Ritter (1991) concludes that IPO firms significantly underperform in the three to five years after going public. He finds that the average holding period return is 34.47% in the three years after going public. However, he also finds that a control sample of some listed stocks produces an average holding period return of 61.86% over this same 3-year holding period. The three main explanations for this underperformance are bad luck, risk measurement and over optimism (Ritter, 1991).

(14)

Lee and Wahal (2004) find that VC-backed IPOs have larger first-day returns than comparable non VC-backed IPOs. However, Megginson and Weiss (1991) find significantly lower first-day returns for VC-backed IPOs compared to non VC-backed IPOs. Barry et al. (1990) also find evidence for a negative relation between VC-backed IPOs and the level of underpricing. Hence, the literature shows some contradictions with respect to the underpricing of VC-backed IPOs. Brav and Gompers (1997) examine the long-run performance of IPOs and find that VC-backed IPOs outperform non VC-VC-backed IPOs using equal weighted returns. Unfortunately, there is no study that examines the level of underpricing of PE-backed IPOs. Nevertheless, Levis (2011) finds that, in the three years after the public offering, PE-backed IPOs show better operating and market performance than otherwise similar non PE-backed IPOs and the market as a whole.

These conclusions suggest some implications for the research in this thesis, because theories and empirical researches show some puzzles regarding the pricing of IPOs. The underpricing in the short-run and the underperformance in the long-run submit a proposal for a new research. The research in this thesis focuses on the impact of the underpricing on the aftermarket

performance. It studies the implications of the underpricing on the subsequent developments of the IPO firm. The findings in this thesis will be relevant for investors, issuers and underwriters, because they are the main stakeholders who are involved in the IPO. The results will show if the uncertainties about a low offer price will be compensated.

This is interesting, because if the IPO market is not efficient in the short-run, there is no reason to assume that it is efficient in the long-run. The market is not efficient in the short-run, because several researchers find that IPOs are underpriced. Therefore, it is not reasonable to argue that the aftermarket stock price of an IPO firm should be the same as any other stock price. In addition, it is interesting to see what the effect of the underpricing at the beginning of the IPO is on the aftermarket performance of the IPO at a later point in the future. Although this is an interesting question, researchers haven’t investigated this yet.

An addition to this research is the distinction between VC-backed, PE-backed and non-sponsored IPOs. This is interesting, because Levis (2011) finds obvious differences across these three groups. They are different in terms of first-day returns, industry classification, operating characteristics and market size.

Overall, based on the previous studies and existing theories, it is predicted that the level of underpricing has a positive impact on the aftermarket performance. This means that if an IPO has a higher level of underpricing, the aftermarket performance will be higher. This is a result of the higher potential to grow from a low price at the first day of trading (Ritter, 1991). It is also predicted that the level of underpricing of VC-backed and PE-backed IPOs have a smaller impact

(15)

on the aftermarket performance than the level of underpricing of otherwise similar

non-sponsored IPOs. This is a result of the lower initial underpricing of these IPOs, so that they have a lower potential to grow from a higher price at the first day of trading (Levis, 2011; Barry et al., 1990).

3. Methodology, Hypothesis and Data

This section explains the model used to test the hypothesis, discusses the hypothesis and lists the data sources.

3.1. Methodology

A multiple regression model will be used to test whether the level of underpricing has a significant impact on the aftermarket performance of the IPO. The following two models are used to analyze this.

(1) 𝐴𝑓𝑡𝑒𝑟𝑚𝑎𝑟𝑘𝑒𝑡 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒_𝑖 = 𝛽₀+ 𝛽₁ 𝑈𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔_𝑖+ 𝛽₂ 𝑃𝐸 𝑏𝑎𝑐𝑘𝑒𝑑_𝑖+ 𝛽₃ 𝑉𝐶 𝑏𝑎𝑐𝑘𝑒𝑑_𝑖+ 𝛽₄ 𝑃𝐸 𝑏𝑎𝑐𝑘𝑒𝑑_𝑖 ∗ 𝑈𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔_𝑖+ 𝛽₅ 𝑉𝐶 𝑏𝑎𝑐𝑘𝑒𝑑_𝑖∗ 𝑈𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔_𝑖+ 𝛽₆ 𝑂𝑓𝑓𝑒𝑟 𝑃𝑟𝑖𝑐𝑒_𝑖 + 𝛽₇ 𝐿𝑛(𝐴𝑔𝑒)_𝑖+ 𝛽₈ 𝐿𝑛(𝐴𝑠𝑠𝑒𝑡𝑠)_𝑖+ 𝛽₉ 𝑃𝑟𝑜𝑐𝑒𝑒𝑑𝑠_𝑖+ 𝛽₁₀ 𝑁𝑌𝑆𝐸_𝑖+ 𝛽₁₁ 𝑁𝐴𝑆𝐷𝐴𝑄_𝑖 + 𝛽₁₂ 𝐶𝑟𝑖𝑠𝑖𝑠_𝑖+ 𝑢_𝑖 (2) 𝐴𝑓𝑡𝑒𝑟𝑚𝑎𝑟𝑘𝑒𝑡 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒_𝑖 = 𝛽₀+ 𝛽₁ 𝑈𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔_𝑖+ 𝛽₂ 𝑃𝐸 𝑏𝑎𝑐𝑘𝑒𝑑_𝑖+ 𝛽₃ 𝑉𝐶 𝑏𝑎𝑐𝑘𝑒𝑑_𝑖+ 𝛽₄ 𝑃𝐸 𝑏𝑎𝑐𝑘𝑒𝑑_𝑖 ∗ 𝑈𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔_𝑖+ 𝛽₅ 𝑉𝐶 𝑏𝑎𝑐𝑘𝑒𝑑_𝑖∗ 𝑈𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔_𝑖+ 𝛽₆ 𝑂𝑓𝑓𝑒𝑟 𝑃𝑟𝑖𝑐𝑒_𝑖 + 𝛽₇ 𝐿𝑛(𝐴𝑔𝑒)_𝑖+ 𝛽₈ 𝐿𝑛(𝐴𝑠𝑠𝑒𝑡𝑠)_𝑖+ 𝛽₉ 𝑃𝑟𝑜𝑐𝑒𝑒𝑑𝑠_𝑖+ 𝛽₁₀ 𝑁𝑌𝑆𝐸_𝑖+ 𝛽₁₁ 𝑁𝐴𝑆𝐷𝐴𝑄_𝑖 + ∑ 𝛾_𝑗𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦_𝑗 47 𝑗=1 + ∑ 𝛿_𝑡𝑌𝑒𝑎𝑟_𝑡 𝑇−1 𝑡=1 + 𝑢_𝑖

where 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝑗 is the industry fixed effect for industry j, 𝑌𝑒𝑎𝑟𝑡 is the year effect for year t and

𝑢_𝑖 is the residual for IPO i. The dependent variable in the regression is the aftermarket

performance. The aftermarket performance is measured by the 3-year buy-and-hold return of

IPOs (Ritter, 1991). The returns are calculated for two intervals. The first interval is the initial return period, month 0, defined as the offering date to the first closing price. The second interval is the aftermarket period, the following 36 months, defined as the three years after the IPO exclusive of the initial return period (Ritter, 1991). The formula for the 3-year holding period return is the following.

(16)

(3) 𝑅_𝑖 = ∏36 (1 + 𝑟_𝑖,𝑡) 𝑡=1

where 𝑟𝑖,𝑡 is the raw return on firm i in event month t. This measures the total return from a

buy-and-hold strategy where a stock is purchased at the first closing market price after going public and held until the earlier of either its 3-year anniversary or its delisting (Ritter, 1991).

The main independent variable is underpricing. This variable is defined as the

percentage difference between the offer price and the first day close price (Ritter and Welch, 2002). The formula for the first day return is the following.

(4) 𝑈𝑛𝑑𝑒𝑟𝑝𝑟𝑖𝑐𝑖𝑛𝑔_𝑖 =𝑓𝑖𝑟𝑠𝑡 𝑑𝑎𝑦 𝑐𝑙𝑜𝑠𝑒 𝑝𝑟𝑖𝑐𝑒𝑖− 𝑜𝑓𝑓𝑒𝑟 𝑝𝑟𝑖𝑐𝑒𝑖

𝑜𝑓𝑓𝑒𝑟 𝑝𝑟𝑖𝑐𝑒𝑖 ∗ 100%

The sign on this coefficient will be positive, because of the higher potential to grow from a low price at the first day of trading, as explained in the literature review (Ritter, 1991).

The remaining variables are control variables. The variables PE-backed and VC-backed equal one (zero otherwise) if the IPO is backed by private equity or venture capital respectively. If the IPO is non-sponsored, both dummy variables equal zero. These variables control for the effect of PE-backing and VC-backing of IPOs on the aftermarket performance (Lee and Wahal, 2004). These control variables are crucial, because Levis (2011) finds obvious differences across these groups. They are different in terms of first-day returns, industry classification, operating characteristics and market size. The signs on these coefficients will be both positive, because venture capitalists are closely monitoring the development of the start-up firm and private equity firms invest in older and more established firms, both resulting in a better aftermarket performance (Berk and DeMarzo, 2011, pp. 771 – 773).

In addition to the main independent variable, the two interaction terms PE-backed *

underpricing and VC-backed * underpricing measure the incremental effect of an underpriced

PE-backed or VC-backed IPO on the aftermarket performance respectively. The signs on these coefficients will be both negative, because VC-backed and PE-backed IPOs have lower levels of underpricing than otherwise similar non-sponsored IPOs (Barry et al., 1990 and Levis, 2011). The lower levels of underpricing imply a lower potential to grow in the long-run.

The variable offer price indicates the offer price in dollars of the IPO. This variable is added to the model, because there could be a relation between the price setting and long-run returns (Booth and Chua, 1996). The sign on this variable will be negative, because smaller offerings are more speculative (Beatty and Ritter, 1986).

The variable age controls for the company age and is defined by the natural logarithm of the difference between the year of the IPO and the year the company is founded (Ritter, 1984).

(17)

Ritter (1991) states that older firms are less risky and have more certainties. Therefore, the sign on this coefficient will be negative, because these firms will be more stable in the aftermarket with a lower grow potential in the long-run. They are already developed to some degree.

The variable assets is measured by the natural logarithm of the total book value of assets before the IPO and is expressed in millions of dollars. This variable is used as a proxy for the firm’s size (Beatty and Ritter, 1986). The sign on this coefficient will be negative, because uncertainty is reduced if the firm is larger and becomes publicly available. Therefore, there is more information publicly available and they will have a lower potential to generate high returns, because they are already developed to some degree (Ritter, 1991).

The variable proceeds measures the offer size of the IPO. This variable is defined as the offer price times the number of shares offered and is expressed in millions of dollars (Beatty and Ritter, 1986). The sign on this coefficient will be negative, because they are usually the more established firms that have larger IPOs. As a result, they will have fewer growth opportunities and a lower aftermarket performance (Ritter, 1991).

Finally, the variables NYSE and NASDAQ equal one (zero otherwise) if the firm is listed on the NYSE or NASDAQ respectively. If the firm is listed on AMEX, both dummy variables equal zero. These variables control for the different aftermarket performances between stock

exchanges. The sign on the variable NASDAQ will be negative, because, according to Ritter (1991), smaller, higher risk and technology firms list on NASDAQ. Hence, the aftermarket performance will be lower.

Quantitative data of 48 Fama-French industries in the Unites States are collected. Fama-French industries are defined on the basis of four-digit SIC codes (Fama and French, 1997). Table A1 in appendix A summarizes the 48 Fama-French industries. The period that will be analyzed is 2001 – 2012. This period includes the financial crisis during the period 2007 – 2012. It only contains this financial crisis, because it is after the dot-com bubble.

The IPO data are collected for 48 different industries observed in 12 years. However, only in the particular year of the IPO data is collected for every firm and is called cross-sectional data. Although the 3-year buy-and-hold return is also per IPO, the data on the monthly stock prices are time-series data (Stock and Watson, 2012, pp. 50 - 52). A method for analyzing cross-sectional data is the ordinary least squares (OLS) regression. There are four least squares assumptions in the regression model (Stock and Watson, 2012, pp. 238 – 240). The first

assumption is that the conditional distribution of 𝑢𝑖 has a mean of zero. The second assumption

that Stock and Watson (2012, p. 238) make is that the variables are independently and identically distributed (i.i.d.) random variables. The third assumption they argue is that large outliers are unlikely and the fourth assumption states that there is no perfect multicollinearity.

(18)

The model uses robust, clustered standard errors. The clustered standard errors allow the regression errors for heteroscedasticity and to have an arbitrary correlation within a cluster, or grouping, but assume that the regression errors are uncorrelated across clusters (Stock and Watson, 2012, pp. 406 – 407).

In order to control for industry and year fixed effects the dummy variables

∑47𝑗=1𝛾𝑗𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝑗 and ∑𝑇−1𝑡=1𝛿𝑡𝑌𝑒𝑎𝑟𝑡 are added to the model. The OLS regression model has 48-1

different industry intercepts and T-1 different time intercepts, one for each industry and year respectively. These intercepts are represented by a set of binary variables. These binary

variables absorb the influences of all omitted variables that differ from one industry and time to the next. The combined industry and year fixed effects eliminate omitted variables bias arising both from unobserved variables that are constant across industries and from unobserved variables that are constant over time (Stock and Watson, 2012, p. 402). These are all the

variables that determine the aftermarket performance of IPO i, but do not change in a particular industry or over time.

The differences between model (1) and (2) are the crisis dummy in the first model and the industry and year dummies in the second model. The crisis dummy is a different way to control for year fixed effects. The dummy variable is equal to one (zero otherwise) if the time period t is during the financial crisis (2007 – 2012). The coefficient on the crisis dummy will be negative, because the crisis has a negative impact on the long-run performance of IPOs (Ritter, 1991). The crisis dummy and the year dummy are mutually exclusive. Therefore, the first model uses the crisis dummy and the second model uses both the year and the industry dummies.

3.2. Hypothesis

The hypothesis of this research is that the level of underpricing has a positive significant impact on the aftermarket performance in the United States during the period 2001 to 2012. This means that if an IPO has a higher level of underpricing, the aftermarket performance will be higher. The level of underpricing of PE-backed and VC-backed IPOs have a smaller impact on the aftermarket performance than the level of underpricing of otherwise similar non-sponsored IPOs.

The hypothesis suggests that the coefficient on the variable underpricing is positive. This is a result of the higher potential to grow from a low price at the first day of trading (Ritter, 1991). The hypothesis also suggests that the coefficients on the interaction terms PE-backed *

underpricing and VC-backed * underpricing are negative. This is a result of the lower initial

underpricing of these IPOs, so that they have a lower potential to grow from a higher price at the first day of trading (Barry et al., 1990 and Levis, 2011).

(19)

3.3. Data and descriptive statistics

The primary database for IPOs is the Thomson One database. Thomson One supplies a broad range of financial content including information on IPOs. The full sample is composed of 1368 IPOs in the United States from 2001 to 2012 and consists of 374 PE-backed IPOs, 510 VC-backed IPOs and 485 non-sponsored IPOs. All IPOs are listed on NYSE, NASDAQ or AMEX. The sample meets criteria that are common in the empirical IPO literature. IPOs with an offer price below $1.00 per share, unit offers, closed-end-funds, REITs and ADRs are excluded from the sample (Ritter, 1991).

The second database for post-IPO information is Datastream. Datastream is a financial database and supplies company information, stock prices and macroeconomics data. The stock prices after the IPO are collected from this database.

First of all, the dependent variable will be discussed. The dependent variable is the aftermarket

performance and is measured by the 3-year buy-and-hold return (Ritter, 1991). In order to

calculate the 3-year buy-and-hold return, the monthly stock prices are collected. The data on monthly stock prices are collected from the database Datastream (Datastream, 2016).

Second, the main independent variable underpricing is discussed. This variable refers to the money left on the table (Ritter, 1991). In order to calculate underpricing, the first day close price and the offer price are collected. The data are collected from the Thomson One database (Thomson One, 2016). These offer prices are also included as a control variable.

Third, the data on the control variables PE-backed, VC-backed, age, total assets, proceeds,

NASDAQ and NYSE are also collected from the Thomson One database (Thomson One, 2016).

The tables 1, 2, 3 and 4 present the descriptive statistics. Table 1 contains the descriptive statistics of all IPOs. The tables 2, 3 and 4 contain the descriptive statistics of VC-backed, PE-backed and non-sponsored IPOs respectively. In table 2 the variable PE-PE-backed is not always 0, because some IPOs are both backed and PE-backed. The same holds for the variable

VC-backed in table 3.

(20)

Table 1.

Descriptive statistics of all IPOs for the period 2001 – 2012.

Aftermarket Performance is measured by the 3-year buy-and-hold return. Underpricing is defined as the difference between the first day return and the offer price divided by the offer price. PE-backed and VC-PE-backed indicate whether an IPO is private equity PE-backed or venture capital PE-backed respectively. The interaction terms VC-backed * Underpricing and PE-backed * Underpricing indicate the incremental effect on the aftermarket performance of either a VC-backed IPO or a PE-backed IPO respectively. The offer price is the price offered for the IPO. Age is the difference between the year of the IPO and the year the company is founded. Total assets are the total assets of the company prior to the IPO. Proceeds indicate the offer size of the IPO. NASDAQ and NYSE indicate whether the IPO is listed on NASDAQ or NYSE. Crisis indicates whether the IPO is in the financial crisis.

Variables Obs. Mean SD Min. Max.

Industry 1,368 31.70 12.20 1 48 Year 1,368 2,007 3.197 2,001 2,012 Aftermarket Performance 1,367 0.197 1.074 -1.000 8.434 Underpricing 1,368 0.126 0.268 -0.909 3.539 PE-backed 1,368 0.273 0.446 0 1 VC-backed 1,368 0.373 0.484 0 1 VC-backed * Underpricing 1,368 0.0659 0.201 -0.643 3.539 PE-backed * Underpricing 1,368 0.0334 0.154 -0.909 3.237 Offer Price 1,368 14.91 6.297 1 85 Age 1,257 1.777 1.405 -3.502 4.639 Total Assets 1,235 5.217 1.889 -2.303 12.52 Proceeds 1,368 240.5 743.2 1.400 16,007 NASDAQ 1,368 0.565 0.496 0 1 NYSE 1,368 0.412 0.492 0 1 Crisis 1,368 0.465 0.499 0 1

The second column in table 1 shows the total number of observations per variable. Therefore, in the initial sample period there are 1368 IPOs. The third column shows the mean of the variables. The average aftermarket performance for all IPOs is 19.7% and the average underpricing for all IPOs is 12.6%. The third column reports the standard deviation and shows that it varies between 0.154 and 743.2. The fourth and fifth columns present the smallest and largest observations per variable, indicated by the minimum and maximum. The table shows that the lowest offer price is $1 and the highest offer price is $85. It also shows that the lowest 3-year buy-and-hold return is -100% and the highest 3-year buy-and-hold return is 843.4%.

Table 2 shows that there are 510 VC-backed IPOs in the sample period. The average aftermarket performance for backed IPOs is 23% and the average underpricing for VC-backed IPOs is 7.5%. The standard deviation varies between 0.154 and 717.573. This table shows that the lowest offer price is $4.60 and the highest offer price is $85. It also shows that the lowest 3-year buy-and-hold return is -99.9% and the highest 3-year buy-and-hold return is 843.4%.

(21)

Table 2.

Descriptive statistics of VC-backed IPOs for the period 2001 – 2012.

Industry 510 28.182 11.491 1 48 Year 510 2006.710 3.264 2001 2012 Aftermarket Performance 510 0.230 1.185 -0.999 8.434 Underpricing 510 0.075 0.298 -0.643 3.539 PE-backed 510 0.002 0.044 0 1 VC-backed 510 1.000 0 1 1 VC-backed * Underpricing 510 0.075 0.298 -0.643 3.539 PE-backed * Underpricing 510 0.000 0.002 -0.035 0 Offer Price 510 13.281 5.866 4.6 85 Age 476 1.991 0.693 -1.505 3.990 Total Assets 478 4.346 1.139 0.693 8.833 Proceeds 510 147.606 717.573 7.650 16006.880 NASDAQ 510 0.790 0.408 0 1 NYSE 510 0.202 0.402 0 1 Crisis 510 0.494 0.500 0 1

Table 3 shows that there are 374 PE-backed IPOs in the sample period. The average aftermarket performance for backed IPOs is 25.5% and the average underpricing for PE-backed IPOs is 12.2%. The standard deviation varies between 0 and 322.993. This table shows that the lowest offer price is $5.50 and the highest offer price is $65. It also shows that the lowest 3-year buy-and-hold return is -99.9% and the highest 3-year buy-and-hold return is 830.2%.

Table 4 shows that there are 485 non-sponsored IPOs in the sample period. The average aftermarket performance for non-sponsored IPOs is 11.7% and the average underpricing of non-sponsored IPOs is 17.7%. The standard deviation varies between 0 and 959.955. This table shows that the lowest offer price is $1 and the highest offer price is $70.41. It also shows that the lowest 3-year buy-and-hold return is -100% and the highest 3-year buy-and-hold return is 731.2%.

(22)

Table 3.

Descriptive statistics of PE-backed IPOs for the period 2001 – 2012.

Industry 374 31.864 11.666 2 48 Year 374 2006.666 3.124 2001 2012 Aftermarket Performance 373 0.255 1.080 -0.999 8.302 Underpricing 374 0.122 0.275 -0.909 3.2367 PE-backed 374 1.000 0 1 1 VC-backed 374 0.003 0.052 0 1 VC-backed * Underpricing 374 0 0.002 -0.035 0 PE-backed * Underpricing 374 0.122 0.275 -0.909 3.2367 Offer Price 374 16.275 5.937 5.5 65 Age 349 2.136 1.413 -2.642 4.6331 Total Assets 345 6.191 1.481 1.1314 10.27 Proceeds 374 257.126 322.933 15 3786 NASDAQ 374 0.455 0.499 0 1 NYSE 374 0.540 0.499 0 1 Crisis 374 0.428 0.495 0 1

The tables 2, 3 and 4 show that VC-backed IPOs have the lowest average underpricing and sponsored IPOs have the highest average underpricing. They also show that non-sponsored IPOs have the lowest average aftermarket performance and PE-backed IPOs have the highest average aftermarket performance. Private equity firms have the oldest age when they go public and non-sponsored firms have the youngest age when they go public. PE-backed IPOs show the smallest range for the offer price and non-sponsored IPOs show the largest range for

the offer price.

Table 5 reports the cross-correlations of the dependent and independent variables. The table shows that the smallest correlation is between the variables PE-backed and year, this

correlation is negative and is -0.0003. This indicates a weak negative relation. The largest correlation is between the variables NYSE and NASDAQ, this correlation is negative and is -0.9532. This indicates a strong negative relation. The correlation between the variables

(23)

Table 4.

Descriptive statistics of non-sponsored IPOs for the period 2001 – 2012.

Aftermarket Performance is measured by the 3-year buy-and-hold return. Underpricing is defined as the difference between the first day return and the offer price divided by the offer price. PE-backed and VC-PE-backed indicate whether an IPO is private equity PE-backed or venture capital PE-backed respectively. The interaction terms VC-backed * Underpricing and PE-backed * Underpricing indicate the incremental effect on the aftermarket performance of either a VC-backed IPO or a PE-backed IPO respectively. The offer price is the price offered for the IPO. Age is the difference between the year of the IPO and the year the company is founded. Total assets are the total assets of the company prior to the IPO. Proceeds indicate the offer size of the IPO. NASDAQ and NYSE indicate whether the IPO is listed on NASDAQ or NYSE. Crisis indicates whether the IPO is in the financial crisis.

Industry 485 35.231 12.328 1 48 Year 485 2006.625 3.186 2001 2012 Aftermarket Performance 485 0.117 0.932 -1.000 7.312 Underpricing 485 0.177 0.216 -0.813 2.9852 PE-backed 485 0 0 0 0 VC-backed 485 0 0 0 0 VC-backed * Underpricing 485 0 0 0 0 PE-backed * Underpricing 485 0 0 0 0 Offer Price 485 15.584 6.630 1 70.41 Age 433 1.257 1.784 -3.502 4.6388 Total Assets 413 5.411 2.376 -2.303 12.523 Proceeds 485 325.176 959.955 1.4 15774 NASDAQ 485 0.414 0.493 0 1 NYSE 485 0.532 0.499 0 1 Crisis 485 0.462 0.499 0 1

Figure B1 in appendix B shows the distribution of the number of IPOs over time. The number of IPOs per year is volatile. In 2004, 198 firms decided to go public, while in 2008 only 26 firms undertook an IPO. The number of IPOs increases from 2002 until 2007 and then decreases in 2008 during the financial crisis. After 2009 the number of IPOs starts to increase. This pattern is shown for both VC-backed and PE-backed IPOs. In most years the number of VC-backed IPOs is higher than the number of PE-backed IPOs.

Figures C1 and D1 in appendices C and D show the average underpricing and average aftermarket performance per year respectively. Figure C1 shows that the average underpricing is 16.84% in 2001, but only 1.05% in 2008. It also shows that the average underpricing

increases from 2002 till 2007 and decreases during the financial crisis. After the financial crisis the average underpricing increases. Figure D1 shows that the average aftermarket performance is as high as 72.09% in 2002, but decreases to -24.60% in the financial crisis. However, after the financial crisis the average aftermarket performance increases.

(24)

Table 5.

Cross-correlations.

Aft. Perf. refers to the variable aftermarket performance. UP refers to the variable underpricing. PE and VC refer to the variables PE-backed and VC-backed respectively. *, ** and *** indicate the significance level at 10%, 5% and 1% respectively.

Variables Industry Year Aft. Perf. UP PE VC PE * UP VC * UP Offer Price Age Assets Total Proceeds NYSE NASDAQ Crisis Industry 1.0000 Year 0.0072 1.0000 Aft. Perf. -0.1405 *** -0.0012 1.0000 UP 0.0149 -0.0236 0.0329 1.0000 PE 0.0083 -0.0003 0.0028 -0.0088 1.0000 VC -0.2223 *** 0.0102 0.1222 *** 0.1463 *** -0.4695 *** 1.0000 PE * UP 0.0027 -0.0109 0.0023 0.4974 *** 0.3545 *** -0.1680 *** 1.0000 VC * UP -0.0399 0.0288 0.0400 0.5293 *** -0.2809 *** 0.5930 *** -0.0994 *** 1.0000 Offer Price 0.2101 *** 0.0070 -0.1635 *** 0.1505 *** 0.1602 *** -0.2592 *** 0.1165 *** 0.0203 1.0000 Age -0.0932 *** -0.0818 *** 0.1798 *** 0.1010 *** 0.1573 *** 0.1281 *** 0.0840 *** 0.1038 *** -0.0992 *** 1.0000 Total Assets 0.2277 *** 0.0230 -0.0907 ** -0.0668 ** 0.3511 *** -0.4332 *** 0.0780 *** -0.2235 *** 0.4963 *** -0.0031 1.0000 Proceeds 0.1284 *** 0.0665 ** -0.0545 0.0178 0.1867 *** -0.2901 *** 0.0675 ** -0.1073 *** 0.5454 *** -0.0761 *** 0.6346 *** 1.0000 NYSE 0.1559 *** 0.1697 *** -0.0460 -0.0243 0.1602 *** -0.3284 *** 0.0586 ** -0.1283 *** 0.3501 *** -0.1304 *** 0.4955 *** 0.4885 *** 1.0000 NASDAQ -0.1378 *** -0.1591 *** 0.0228 0.0386 -0.1367 *** 0.3501 *** -0.0494 * 0.1475 *** -0.2916 *** 0.1376 *** -0.4381 *** -0.4515 *** -0.9532 *** 1.0000 Crisis -0.0078 0.8453 *** -0.0410 -0.0217 -0.0447 * 0.0463 * -0.0303 0.0558 * 0.0041 -0.1008 *** 0.0042 0.0722 *** 0.1628 *** -0.1532 *** 1.0000

(25)

4. Empirical results

This section presents the main results, does some robustness checks and shows additional results.

4.1. Empirical results

First of all, column 1 of table 6 presents the baseline regression results without industry and year fixed effects. This is the regression of model (1) and uses the crisis dummy to control for year fixed effects. In addition to the crisis dummy in column 1, the regression in column 2 adds

the industry fixed effects. This increases the R2_{by 6.4% from 9.5% to 15.9%. Finally, column 3}

uses year fixed effects instead of the crisis dummy. This increases the R2 to 23.5%. Columns 1, 2

and 3 use all IPOs in the sample.

In column 2, the crisis dummy has a negative coefficient and is statistically significant at the 10% level. This means that the financial crisis has a negative impact on the aftermarket performance. This is also economically interesting, because it suggests that in an economic downturn, such as the financial crisis, the holding period return is lower.

In column 3, the main variable of interest underpricing has a coefficient of 0.531 and is statistical significant at the 1% level. This implies that when the level of underpricing increases with 1%, the aftermarket performance increases with 0.531%. This is consistent with the model implication that a low price at the first day of trading has a higher potential to grow in the long-run.

The two interaction terms PE-backed * underpricing and VC-backed * underpricing both have a negative impact on the aftermarket performance. However, they are not statistically significant. When the IPO is backed by private equity (venture capital), a 1% increase in the level of underpricing has a 0.798% (0.854%) smaller impact on the aftermarket performance than the level of underpricing of an otherwise similar non-sponsored IPO. These results suggest that the level of underpricing of PE-backed and VC-backed IPOs have a smaller impact on the aftermarket performance than the level of underpricing of otherwise similar non-sponsored IPOs. This is economically interesting, because this is a result of the lower initial underpricing of these IPOs compared to non-sponsored IPOs. They have a lower potential to grow form a higher price at the first day of trading.

The coefficients on the control variables PE-backed and VC-backed are positive, but not statistically significant. When the IPO is backed by private equity (venture capital), the

aftermarket performance is 0.121% (0.166%) higher than an otherwise similar non-sponsored IPO. However, this is economically interesting, because private equity firms invest in older more established firms and venture capital firms closely monitor the development of the firm. In this case the financial expertise is enhanced, resulting in a higher holding period return compared to non-sponsored IPOs.

(26)

Table 6.

IPO 3-year buy-and-hold return regressions for the period 2001 – 2012.

This table reports the link between underpricing and aftermarket performance. The dependent variable in all regressions is the aftermarket performance, measured by the 3-year buy-and-hold return. Columns 1, 2 and 3 use all IPOs in the sample. Column 4 uses only PE-backed IPOs. Column 5 uses only VC-backed IPOs. Column 6 uses only non-sponsored IPOs.

Year fixed effects based on the IPO year and industry fixed effects based on the 48 Fama-French industries are included, where the coefficients are not reported for brevity. The t-statistics (in parentheses) are computed using heteroscedasticity-consistent standard errors that are corrected for clustering across year and industry. *** p<0.01, ** p<0.05, * p<0.1

(1) (2) (3) (4) (5) (6)

Aftermarket

Performance All IPOs All IPOs All IPOs PE-backed IPOs VC-backed IPOs Non-sponsored IPOs

Underpricing 0.375* 0.351* 0.531*** 0.336 0.174 0.489 (1.696) (1.727) (2.823) (0.558) (0.684) (0.845) PE-backed 0.118 0.101 0.121 (0.956) (0.744) (0.892) VC-backed 0.301** 0.199 0.166 (2.060) (1.110) (0.974) PE-backed * Underpricing -0.283 -0.475 -0.798 (-0.514) (-0.826) (-1.392) VC-backed * Underpricing -0.718 -0.757 -0.854 (-1.144) (-1.142) (-1.348) Offer Price -0.0420*** -0.0375*** -0.0377*** -0.0393 -0.0354 -0.0417** (-3.462) (-2.899) (-2.868) (-1.376) (-1.308) (-2.210) Ln(Age) 0.105*** 0.108*** 0.124*** 0.200** 0.137 0.0520 (3.076) (3.016) (3.513) (2.380) (1.171) (1.056) Ln(Total Assets) -0.00441** -0.00289 -0.00248 0.000574 0.00434 -0.00316 (-2.221) (-1.322) (-1.154) (0.0960) (1.149) (-0.979) Proceeds 0.000606* 0.000533 0.000585* -0.000511 0.000883 0.00130*** (1.915) (1.636) (1.843) (-0.723) (0.971) (2.674) NYSE -1.306*** -1.323*** -1.240*** -0.161 -1.022*** -1.329** (-6.712) (-4.682) (-3.894) (-0.686) (-2.750) (-2.433) NASDAQ -1.441*** -1.475*** -1.326*** -1.350*** -1.407*** (-7.831) (-5.358) (-4.277) (-4.324) (-2.622) Crisis -0.142 -0.170* (-1.520) (-1.680) Constant 1.137*** 1.394*** 1.200** 0.143 -0.353 2.603*** (5.013) (2.884) (2.033) (0.215) (-0.605) (3.218) Observations 1219 1219 1219 339 465 416 R-squared 0.095 0.159 0.235 0.440 0.250 0.437

Year fixed effects No No Yes Yes Yes Yes

(27)

The control variable offer price is statistically significant at the 1% level and has a negative impact on the aftermarket performance. When the offer price increases with $1, the aftermarket performance decreases with 0.0377%. This suggests that smaller offerings are more speculative. Although this could cause more fluctuations in the first days, the holding period return in the long-run is lower.

The control variable age has a positive coefficient and is statistically significant at the 1% level. When the age of the firm increases with 1 year, the aftermarket performance

increases with 0.124%. Therefore, an older firm has some benefits with respect to a young firm to undertake an IPO. They have more financial expertise and could have a better overview of their positive NPV projects and growth opportunities.

The control variable total assets has a negative coefficient and is not statistically

significant. When the total assets before the IPO increase with 1 million dollars, the aftermarket performance decreases with 0.00248%. This is economically interesting, because uncertainty is reduced if the firm is larger and becomes publicly available. There is more information publicly available and they have a lower potential to generate high returns.

The control variable proceeds has a positive impact on the aftermarket performance and is statistically significant at the 10% level. When the proceeds increase with 1 million dollars, the aftermarket performance increases with 0.000585%. Therefore, the holding period return increases, if there are a higher number of shares offered in the IPO.

The coefficients on the control variables NYSE and NASDAQ are both negative and statistically significant at the 1% level. This means that if IPOs are first listed on these exchanges, they have a lower aftermarket performance with respect to IPOs first listed on AMEX. This is a result of the smaller, higher risk and technology firms that list on these exchanges.

In columns 4, 5 and 6 of table 6 the sample is divided in three subsamples. Column 4 uses only PE-backed IPOs, column 5 uses only VC-backed IPOs and column 6 uses only non-sponsored

IPOs. The R2_{’s are 44%, 25% and 43.7% respectively. However, in these subsamples the main}

variable of interest underpricing is not significant. The coefficients indicate that the level of underpricing of VC-backed IPOs have the smallest effect and the level of underpricing of non-sponsored IPOs the largest effect on the aftermarket performance. This is consistent with the results in column 3, because the coefficients on the interaction terms are negative. The coefficients on the interaction terms are -0.798 and -0.854 for PE-backed and VC-backed IPOs respectively. These coefficients indicate that their levels of underpricing have a lower impact on the aftermarket performance than the level of underpricing of non-sponsored IPOs. However, VC-backed IPOs show the lowest impact.