U NDERPRICING V ENTURE C APITAL Q UALITY AND

(1)

AND

U

NDERPRICING

An empirical investigation on the relationship between

Venture Capital firm quality and the level of underpricing

Author:

A.M. Jacobs

Supervisor:

dr. J.H. von Eije

January 2008

UNIVERSITY OF GRONINGEN

(2)

VENTURE CAPITAL QUALITY AND UNDERPRICING

An empirical investigation on the relationship between

Venture Capital firm quality and the level of underpricing

Abstract

Using a sample of 106 Venture Capital (VC) backed Initial Public Offerings (IPO) in Europe we study the influence of VC firm quality on the level of underpricing. Seven quality measures based IPO involvement and European Private Equity & Venture Capital Association (EVCA) membership are constructed to proxy for VC firm quality. We regress the quality proxies and a number of independent offering, firm and market characteristics on underpricing and find by means of Ordinary Least Squares that all proxies show the expected negative relationship with the level of underpricing as found by Lin and Smith (1998). Of the seven proxies, only one based on total offer proceeds as lead VC firm appears to be statistically significant on a 10 percent confidence level. In conclusion, we find marginal evidence that IPOs backed by VC firms of higher quality are less underpriced than those backed by lower quality VC firms.

Allard M. Jacobs

allardmjacobs@hotmail.com

(3)

PREFACE

This thesis is the conclusion of my Msc Business Administration (Finance) at the University of Groningen. Over the last three months I have been studying the relationship between Venture Capital firm quality and the level of underpricing. The result of that work is found in this paper.

(4)

LIST OF TABLES

Table 1. Average initial returns for 39 countries ... 12

Table 2. Results of PE and VC backed IPO underpricing studies ... 27

Table 3. Results of VC backed IPO underpricing studies with respect to firm quality ... 30

Table 4. Overview of independent variables... 41

Table 5. Descriptive statistics (n = 106)... 43

Table 6. IPO volume over time (n = 106) ... 44

Table 7. Country distribution of IPOs (n = 106)... 44

Table 8. Industry distribution of IPOs (n = 106)... 45

Table 9. Underwriter quality distribution of IPOs (n = 106) ... 46

Table 10. VC firm age descriptive statistics (n = 175)... 46

Table 11. VC firm ranking no. of IPOs (n = 175)... 46

Table 12. VC firm ranking total offer proceeds (n = 175) ... 47

Table 13. VC firm ranking average offer proceeds (n = 175)... 47

Table 14. EVCA membership (n = 175) ... 48

Table 15. OLS regression results (n = 106) ... 50

Table 16. Overview of expected and found signs of independent variables... 52

Table A1. Thomson IPO league tables... 68

Table A2. Ranking table underwriters ... 68

Table B1. Common sample correlation matrix (n = 106) ... 69

Table C1. Ramsey RESET test results ... 70

(7)

LIST OF FIGURES

(8)

1. INTRODUCTION

Shares issued in Initial Public Offerings (IPO) often show high initial returns on their first day of trading. It is beyond doubt that this ‘underpricing’ of the shares is costly to a firm’s initial owners; shares offered in the issue are sold at too low a price. The consequential money that is left on the table increases with every share of the company sold in the offering. Large, early stage investors looking for an exit of their investment – such as Venture Capital (VC) firms – are the obvious losing party in this event. However, it could well be the backing of the IPO by VC firms that reduces this level of underpricing as is shown by pioneering scholars as Barry, Muscarella, Peavy and Vetsuypens (1990) and Megginson and Weiss (1991). The aim of this study is to examine this certification role VC firms provide one step further; we study the influence of the quality of the VC firm on the level of underpricing.

The notion of quality of a firm – whether it is a VC firm or any other – is a rather subjective one. What is of high quality to one person could be considered of no quality at all by the next. It could be due to this subjectivity or due to the private nature of VC itself, but at this point in time research on the influence of VC firm quality on underpricing is limited. Since Lin and Smith (1998), only Ljungqvist (1999) and Dolvin and Pyles (2006) have explicitly studied this relationship. We construct seven quality proxy variables and test them on their significance in an underpricing regression model. Contrary to previously mentioned research, we study European IPOs as opposed to IPOs in the United States.

We find that the level of underpricing is negatively related to one of the quality proxies, i.e. the variable based on the offer proceeds of all IPOs the VC firm has been involved in our sample shows is significant at a 10 percent confidence level. Moreover, all other proxies for quality show the expected negative relationship to the level of underpricing, although these results are not significant. These findings marginally show that IPOs backed by higher quality VC firms tend to be less underpriced than those backed by lower quality VC firms and are in line with the results from Lin and Smith (1998).

(9)

(10)

2. LITERATURE REVIEW

This chapter is divided in four sections. After this introduction, section 1 reviews the concept of Initial Public Offerings (IPO) and the motivation behind the decision to take a company public. Section 2 discusses literature regarding the phenomenon of underpricing in an IPO. Section 3 provides a brief insight into the Venture Capital (VC) firm. Finally, section 4 reviews the influence of VC backing and VC firm quality on the level of underpricing. Furthermore, an overview of independent control variables used by scholars in VC backed underpricing regression models will be presented.

2.1 Initial Public Offering1

Most new companies initially finance their business by raising equity capital among a (small) number of private investors, with the absence of a liquid market where stakes in the respective companies can be traded. With this liquidity problem in mind, it is argued that companies reach a point in their growth cycle where they need additional capital financing, which is achieved through an Initial Public Offering (IPO). “An Initial Public Offering occurs when a security is sold to the general public for the first time, with the expectation that a liquid market will develop. (…) Once the stock is publicly traded, this enhanced liquidity allows the company to raise capital on more favourable terms than if it had to compensate investors for the lack of liquidity associated with a privately-held company” (Ritter, 1998; p. 5). This straightforward reasoning on the going public decision is also provided in Ritter and Welch (2002; p. 1796) as the “desire to raise equity capital for the firm and to create a public market in which the founders and other shareholders can convert some of their wealth into cash at a future date.”

One of the first scholars to come up with a formal theory on the decision to go public was Zingales (1995). He emphasizes aspects of the traditional trade-off between the costs and benefits of going public. “The decision of a firm to go public is the result of a value-maximizing decision made by an initial owner who wants to eventually sell his company” (p.

(11)

426). Zingales finds that it is much easier for potential acquirers to identify potential takeover targets when they are publicly traded. Moreover, initial owners realize they are subject to greater pressure from acquirers on pricing concessions during a potential takeover process than are outside investors. By undertaking an IPO, the initial owners thus increase the likelihood of an acquisition of their company for a higher value than what they would be paid for in an outright sale.

In her 1996 article Roëll provides an overview of the determinants of the decision to go public. She describes the advantages of a stock market flotation as manifold. Most of the determinants of the decision stem from conventional theories as the access to new capital finance; enhanced company image and publicity; motivating management and employees; cashing in; and exploiting mispricing.

Inevitably, with benefits come costs and the decision to go public is often the previously-mentioned trade-off between these costs and benefits of going public. On the cost side, Zingales (1995) and Roëll (1996) shortly summarize these as the registration and underwriting costs; the annual disclosure costs; constraints on the freedom of action in making business decisions; the agency problems generated by a separation between ownership and control; and the underpricing cost.

(12)

2.2 Underpricing in Initial Public Offerings

2.2.1 Underpricing

Among the first scholars to document that when companies go public, the shares they sell tend to be underpriced are Logue (1973) and Ibbotson (1975). This concept is best defined as a significant share price increase on the first day of trading (initial return), i.e. the offer price is substantially lower than the first day closing price. Evidently, underpricing is costly to a firm’s initial owners; shares offered in the issue are sold at too low a price, while the value of shares retained after the IPO is diluted. In monetary terms, firms that go public appear to leave many billions ‘on the table’2 in the IPO market.

2.2.2 Empirical evidence of underpricing

Numerous articles have been devoted on finding empirical evidence that shares in IPOs are initially underpriced. Loughran, Ritter and Rydqvist (1994) present a comprehensive overview on the average first day or initial return (i.e. the excess return over the offer price or underpricing). They provide a country based overview of studies performed on average initial return, of which an updated version can be found in table 1.

Table 1. Average initial returns for 39 countries

Country Source Sample

Size Time Period Average Initial Return

Australia Lee, Taylor and Walter; Woo 381 1976-1995 12.1%

Austria Aussenegg 83 1984-2002 6.3%

Belgium Rogiers, Manigart and Ooghe; Manigart; DuMortier 93 1984-2004 14.2% Brazil Aggarwal, Leal and Hernandez 62 1979-1990 78.5% Canada Jog and Riding; Jog and Srivastava; Kryzanowski and Rakita 500 1971-1999 6.3% Chile Aggarwal, Leal and Hernandez; Celis and Maturana 55 1982-1997 8.8% China Datar and Mao; Gu and Qin (A-shares) 432 1990-2000 256.9% Denmark Jakobsen and Sorensen 117 1984-1998 5.4% Finland Keloharju; Westerholm 99 1984-1997 10.1% France Husson and Jacquillat; Leleux and Muzyka; Paliard and _{Belletante; Derrien and Womack; Chahine} 571 1983-2000 11.6% Germany Ljungqvist; Rocholl 545 1978-2001 31.1% Greece Nounis, Kazantzis and Thomas 363 1976-2005 25.1%

(13)

Country Source Sample Size Time Period Average Initial Return

Hong Kong McGuinness; Zhao and Wu; Ljungqvist and Yu 857 1980-2001 17.3% India Krishnamurti and Kumar 98 1992-1993 35.3% Indonesia Hanafi; Ljungqvist and Yu; Danny 265 1989-2003 20.2%

Iran Bagherzadeh 279 1991-2004 22.4%

Israel Kandel, Sarig and Wohl; Amihud and Hauser 285 1990-1994 12.1% Italy Arosio, Giudici and Paleari; Cassia, Paleari and Redondi 181 1985-2001 21.7% Japan Fukuda; Dawson and Hiraki; Hebner and Hiraki; Pettway and _{Kaneko; Hamao, Packer and Ritter; Kaneko and Pettway} 1,689 1970-2001 28.4% Korea Dhatt, Kim and Lim; Ihm; Choi and Heo 477 1980-1996 74.3% Malaysia Isa; Isa and Yong 401 1980-1998 104.1% Mexico Aggarwal, Leal and Hernandez 37 1987-1990 33.0% Netherlands Wessels; Eijgenhuijsen and Buijs; Jenkinson, Ljungqvist and _Wilhelm 143 1982-1999 10.2% New Zealand Vos and Cheung; Camp and Munro 201 1979-1999 23.0%

Nigeria Ikoku 63 1989-1993 19.1%

Norway Emilsen, Pedersen and Saettem 68 1984-1996 12.5% Philippines Sullivan and Unite 104 1987-1997 22.7%

Poland Jelic and Briston 140 1991-1998 27.4%

Portugal Almeida and Duque 21 1992-1998 10.6% Singapore Lee, Taylor and Walter; Dawson 441 1973-2001 29.6% South Africa Page and Reyneke 118 1980-1991 32.7% Spain Ansotegui and Fabregat 99 1986-1998 10.7% Sweden Rydqvist; Schuster 332 1980-1998 30.5% Switzerland Drobetz, Kammermann and Walchli 120 1983-2000 34.9% Taiwan Lin and Sheu; Liaw, Liu and Wei 293 1986-1998 31.1% Thailand Wethyavivorn and Koo-smith; Lonkani and Tirapat 292 1987-1997 46.7%

Turkey Kiymaz 163 1990-1996 13.1%

United Kingdom Dimson; Levis; Ljungqvist 3,122 1959-2001 17.4% United States Ibbotson, Sindelar and Ritter; Ritter 15,333 1960-2005 18.1%

Source: Loughran, Ritter and Rydqvist (1994 - updated 2006)

(14)

2.2.3 Underpricing rationalized

The significant empirical evidence on the existence of underpricing resulted in a large set of theoretical literature in the 1980s and 1990s that tries to rationalize this underpricing in public offerings. Ritter (1998; 2003) and Ljungqvist (2007) provide an extensive overview of this literature and theoretical aspects of underpricing. Ljungqvist groups the theories under four broad headings: asymmetric information, institutional reasons, control considerations and behavioural approaches. In the upcoming sections an overview of these theories, extended with additional literature, is provided. The theories are not mutually exclusive. Furthermore, one theory can be applied better to one IPO than to another.

2.2.3.1 Information asymmetry

Asymmetric information models stem from the belief that one party is better informed than others. In a public offering the key parties are the issuing firm, the bank that underwrites3 and markets the deal and the investors that will buy the shares issued.

Winner’s curse

One variant of the information asymmetry hypothesis is the winner’s curse model developed by Rock (1986). In his model, Rock assumes that some investors are better informed about the true value of the shares in the offer than are other investors. This causes the informed investors to only bid for attractively priced shares, i.e. shares that are underpriced. These investors crowd out the other investors when good issues are offered and they withdraw from the IPO market in the case of bad issues. Thus, the return uninformed investors earn is below the average underpricing return, since they will receive all the shares they bid for in unattractive offerings and only a modest part in the attractive ones. In this case, uninformed investors may become unwilling to bid for new issues, leaving the IPO market with only equally informed investors. The issuing firm must underprice its shares in order to guarantee that the uninformed investors purchase these shares in the issue.

(15)

Information revelation theory / market feedback hypothesis

The pro rata allocation of shares gives way to the winner’s curse. This can be countered by book building, in which underwriters have the ability to allocate shares at their discretion. Book building is the process where the underwriter attempts to determine the enthusiasm of institutional investors, which is eventually used in setting the offer price. If some investors – as Rock assumes – have superior information over others, their information is most valuable to the underwriter. However, there must be an incentive for these investors to reveal their interest. The allocation of shares through book building can provide such an incentive (Benveniste and Spindt, 1989; Ritter, 1998). Here, the underwriter allocates no, or only a few, shares to investors who bid conservatively and those who bid aggressively are awarded a disproportionately large amount of shares. For the informed investors to be induced to reveal their information the issue must be underpriced.

Principal-Agent model

Mechanisms such as book building highlight the importance of underwriters in the IPO process. This dominant position gives rise to the principal-agent problem between the issuer and the underwriter. As shown before, underpricing is a wealth transfer from issuer to the investor. This can lead to rent seeking behaviour, where investors seek large allocations of underpriced shares through special arrangements with the underwriter behind the issuer’s back (Baron and Holmström, 1980). In addition to that, underwriters might allocate underpriced shares to top executives in the hope to receive future business from their companies. This practice is known as spinning. In both instances, the underwriter gains directly from intentional underpricing.

Investment banker’s monopsony power hypothesis

(16)

Quality signalling theory

When a firm issues its shares underpriced, it will “leave a good taste in investors’ mouths” (Ibbotson, 1975; p. 264). This can be favourable in further equity offerings, where they can offer their issue at a higher price. Assume a low and high quality firm. Low quality firms have the incentive to mimic those with high quality. In this case the signal is the offer price. One can assume that the true type of the firm is revealed before the next financing stage. This is where the high quality firm can reap the benefits from its underpricing. However, it should be noted that evidence on “the hypothesized relation between initial returns and subsequent seasoned new issues is not present, casting doubt on the importance of signalling as a reason for underpricing” (Ritter, 1998; p. 9).

Marketing theory

Closely related to the quality signalling theory is the notion of the IPO as a marketing event. Due to a high first day return a substantial amount of publicity is generated. This publicity could lead to additional investor interest (Chemmanur, 1993) or additional product market revenue due to greater brand awareness (Demers and Lewellen, 2003). However, it should be questioned how costly this advertising is opposed to traditional advertising mechanisms.

2.2.3.2 Institutional reasons

Legal insurance

Although other countries have similar laws, predominantly in the United States underwriters and issuers are exposed to risk of litigation by investors based on misstatements or omissions in the IPO prospectus. Ibbotson (1975) and Tinic (1988) argue that underpricing may be used to act as insurance against such litigation. However, this explanation is rather US focused, since strict liability laws are not a global phenomenon as is underpricing. Empirical evidence shows that the risk of litigation is not economically significant for most European and Asian countries (Ljungqvist, 2007).

Price stabilization

(17)

Price stabilization is intended to diminish price drops in the aftermarket, which are often observed in the first days or weeks after the issue. By setting a lower offer price, the chance of a price fall in the aftermarket is significantly reduced. Surprisingly, this ‘price manipulation’ is legal in many countries. This leads statistically to more instances of underpricing (Ljungqvist, 2007).

Tax arguments

Generally, employment income is taxed heavier than capital gains. This induces firms to reward their employees by allocating them appreciating assets over salary. Underpriced shares, given to the firm’s employees at the public offering, are an example of such an asset. Rydqvist (1997) studied this phenomenon in Sweden, where capital gains were taxed less than employment income. In 1990 the tax regulations were altered in such a manner, that gain from underpricing became subject to income tax. He noted that underpricing fell from 41 percent pre-1990 to 8 percent post-1990.

2.2.3.3 Control considerations

The decision to go public is one of the first steps in the separation of ownership and control. The allocation of ownership can have a large effect on the incentive for managers to make optimal operating and investment decisions for shareholders. Where the separation of ownership and control is incomplete, agency problems can emerge. Instead of pursuing maximum shareholder value, managers try to maximize their personal benefits. Two opposing theories have been developed to explain underpricing by means of an agency cost approach; Brennan and Franks (1997) put forward that managers use underpricing to avoid monitoring by outside shareholders, where Stoughton and Zechner (1998) suggest that underpricing encourages monitoring. Both approaches will be explained in more detail below.

Retaining control

(18)

‘public good’ characteristics – as all shareholders may benefit equally – shareholders will invest in a sub-optimally low level of monitoring; i.e. it may not pay any one of them to monitor the performance of the management (Schleifer and Vishny, 1986). Here, underpricing is used to generate excess demand to facilitate strategic allocation of shares (Smart and Zutter, 2003).

Reducing agency costs

The costs from the agency problem are ultimately borne by the shareholders, which are in most instances the managers as well. If their ownership in the firm is large enough that agency costs outweigh personal benefits, monitoring by an outside investor will be in the manager’s best interest as well. However, monitoring will only occur if this is privately optimal for the shareholder. According to Stoughton and Zechner (1998), managers will try to allocate a large stake to an investor, which should provide him with an incentive to monitor. However, if the proposed allocation of shares is sub-optimally large to that investor (e.g. due to not being properly diversified), underpricing may be used to provide an incentive for the investor to accept the larger than desired allocation.

2.2.3.4 Behavioural theories

The behavioural approach to underpricing focuses on both the issuer and the investor. It assumes either investors that demonstrate ‘irrational’ behaviour and drive the share price beyond their true value, or issuers that are subject to a behavioural bias which causes them to fail to put pressure on the underwriter to reduce underpricing.

Investor sentiment

(19)

issuer from carrying out such a pricing strategy. By allocating large amounts of shares to institutional investors, who subsequently hold part of their shares, issuers get round these regulatory constraints. Because holding shares is risky (particularly in hot issue markets) the shares need to be underpriced to offset this risk (Ljungqvist, Nanda and Singh, 2004).

Prospect theory and mental accounting

Underwriters can benefit from underpriced shares if investors engage in rent-seeking behaviour to increase their chances of being allocated these shares. However, the issuer will not be pleased with leaving money on the table due to underpriced shares. Loughran and Ritter (2002) combine prospect theory (how one evaluates gains and losses) with mental accounting and provide an argument why the issuers might fail to “get upset” (p. 413). They argue that issuers sum the wealth loss due to underpricing with the wealth gain on the retained shares. If this perceived gain exceeds the underpricing loss, the issuer is satisfied with the underwriter’s performance in the IPO and the intentional underpricing by the underwriter will not be chastised.

Informational cascades

When investors make their investment decisions sequentially, i.e. investors are able to base their bid on the information of bids by early investors, informational cascades can develop. A high level of bids by early investors induces later investors to bid for the offering as well. In this case, early investors can either ‘make or brake’ an offering. These early investors can exploit this advantage by demanding that the issuer underprices its shares for committing to the IPO.

2.3 Venture Capital

2.3.1 Venture Capital at a glance

(20)

in family-owned companies or the buy-out and buy-in of a business by experienced managers may be achieved using PE funding.

The different concepts of PE can basically be divided into two different categories: Venture Capital (VC) and buy-out financing. Buy-out financing is the concept where a business, business unit or company is acquired from the current shareholders (public or private). This can be achieved through a management buy-out (MBO), a management buy-in (MBI), an institutional buy-out (IBO) or a leveraged buy-out (LBO).

VC is the second subset of PE and is best described as equity investments made for the launch, early development or expansion of a business. It refers to professional equity co-invested with the entrepreneur to fund an early stage (seed and start-up) or expansion venture. The high risk the investor takes is offset by the expectation of a higher than average return on the investment. VC firms have backed many (high-technology) companies in the past, such as Apple, Intel and Microsoft (Prowse, 1998).

VC firms are “often active investors, monitoring the progress of firms, sitting on boards of directors, and meting out financing based on the attainment of milestones. [They] monitor strategy and investment decisions as well as take an active role in advising the firms” (Gompers and Lerner, 1999a; p. 4). Next to their active involvement in day to day operations and providing of monetary capital, VC firms endow entrepreneurs with intangible capital such as access to consultants, investment bankers and lawyers.

(21)

2.3.2 Exit of a Venture Capital investment

In the course of time, most PE firms gain from its investment through profit taking. However, since VC firms usually invest in young start-up companies that often exhibit losses in that stage, this profit taking is seldom observed. Next tot that, as there is no liquid secondary market, a VC firm can realize its return on the investment only at the exit of it. This exit is therefore of great influence on the performance of the investment. For this reason the exit receives attention even before the company is invested in. The choice for an exit is dependent on the stage of the company; the wish for a continuing exposure after the exit; the costs; the achievable value; the company’s wishes; and the risk the exit poses. In recent years, the exit climate has been favourable due to positive economic prospects and the amount of money in the market.

The amount of money in the market increases acquisition competition. Figure 1 shows that the positive exit valuations caused a significant rise in the number of exits. However, a decline in the number of exits can be noticed from 2001 until mid 2002. This follows the 2000 economic downturn, which decreased earnings outlooks and the money available for acquisitions resulting in lower valuations.

Figure 1: Large European Private Equity (>500m Euro) exit volume

0 10 20 30 40 50 60 70 80 90 2002 2003 2004 2005 2006 N u m b e r o f e x it s

(22)

Gladstone (1989) describes six way of concluding a VC investment: (1) sale of the company’s shares in an IPO; (2) trade sale (sale of shares to another company); (3) repurchase of the shares by the firm itself; (4) sale of shares to another investor (secondary buy-out) or financial institution; (5) reorganization of the company; and (6) liquidation of the company. Here, we investigate the first option. As we can see from figure 2, which depicts the build-up of possible exits over the last five years, the choice for an IPO as an exit route has been fairly stable over this time frame. However, we can see a relatively large increase in the last year, which can be largely attributed to the positive public market environment.

Figure 2: European divestment at type

30,9% 20,4% 23,7% 22,6% 22,7% 6,6% 5,6% 4,8% 4,5% 9,0% 5,3% 6,2% 7,0% 4,5% 7,2% 30,0% 11,6% 9,7% 4,7% 3,8% 3,9% 20,2% _13,1% 18,4% 16,6% 8,4% 15,9% 21,3% 23,3% 17,1% 3,9% 6,0% _2,9% 4,0% _5,4% 11,0% _14,1% 17,5% 18,0% 18,2% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 2002 2003 2004 2005 2006

Trade sale IPO Sale of quoted equity Write-of f Sale to another PE firm Repayment of shares/loans Sale to financial institution Other

Source: European Private Equity & Venture Capital Association (EVCA)

2.4 Underpricing in Venture Capital backed Initial Public Offerings

2.4.1 Venture Capital backing and underpricing: theories

(23)

Certification and monitoring model

Megginson and Weiss (1991) state that “third party certification has value whenever securities are being issued in capital markets where insiders of the issuing firm and outside investors have different information sets concerning the value of the offering firm” (p. 880). This information asymmetry can lead to a stand-off, where outside investors are not willing to purchase securities at the offer price as set by the issuer. A certifying party can stand in for the reliability of the issuing terms. For third-party certification to be credible the certifying party must meet three tests.

Firstly, there must be reputational capital at stake, which would be sacrificed by falsely certifying an offering. As is shown in several studies (Gompers, 1996; Dolvin and Pyles, 2006), over time most VC firms take more than one company public. VC firms, therefore, “have a very strong incentive to establish a trustworthy reputation in order to retain access to the IPO market on favourable terms” (Megginson and Weiss, 1991; p. 881).

Secondly, the value of the certifier’s reputational capital must be larger than the greatest one-time pay-off from falsely certifying an issue. It is shown by Sahlman (1990) that one of the criteria to be a successful VC firm manager is to be able to establish profitable ‘follow-on’ funds and to achieve an enhanced deal flow from entrepreneurs. The investment VC firms make in reputational capital allow them to remain competitive in the VC industry, as well as in the capital markets. This indicates that to be successful as a VC firm reputational capital is of outmost importance. The one-time pay-off will never supersede the profits this reputation will generate in the future. Furthermore, VC firms are large shareholders of the issuing firm. Clearly, one of the most obvious ways they might gain from false certification and take advantage of the accompanying high price is to sell shares in the IPO. Retention by VC firms of their stake in the company after the offer diminishes their profits from certifying falsely and can therefore be an instrument for reliable certification.

(24)

Barry et al. (1990) focus on the monitoring role of VC firms. Although the monitoring role of VC firms is often separately mentioned from the certification role, both theories are closely related. The monitoring role stems from the agency approach that depicts that VC firms should use a number of methods to monitor their companies to control self-fulfilling behaviour of entrepreneurs. This can take the form of stage financing, board membership and detailed legal contracts (Wang, Wang and Lu, 2003). Besides the controlling effect from these methods, in this way VC firms can also add value to the companies on an operational basis. The certifying role of the VC firm is supported by a number of scholars among who Wang et al. (2003), who state that VC firms are better certifiers than underwriters and auditors for two reasons. Firstly and this is where certification and monitoring intertwine, VC firms are much more knowledgeable about the issuing firm, because of their equity holdings, their positions on the board and the longer and closer working relationship with the firm (monitoring). Secondly, they mention the reputational capital – as used in Megginson and Weiss (1991) – that is at stake for false certification.

The role of the VC firm in the IPO goes clearly beyond ‘cashing in’ on their investment. Where the monetary profits may be the incentive to the VC firms for an IPO, it acts as a certifying and monitoring party in the offering. In this way the asymmetrical information problem is decreased, which improves the chances of a successful public placement by the issuing firm and reduces underpricing.

Grandstanding model

The certification and monitoring models depict a positive influence on the offering process and a negative relationship with the level of underpricing. Next to that line of research, another line has been introduced that depicts a positive relationship with underpricing. Within this reasoning we find the grandstanding model.

(25)

A second approach to the grandstanding model builds on the certification hypothesis. Barnes and McCarthy (2002) state that IPOs backed by young VC firms will be more underpriced than those backed by older VC firms, since these veteran firms are better able to certify firm quality at the IPO.

Adverse selection hypothesis

A second hypothesis which causes a positive relationship between underpricing and VC backing is the adverse selection hypothesis problem (Amit, Glosten and Muller, 1990). The adverse selection provides a negative effect of VC on the IPO. Due to the information asymmetry between entrepreneurs and VC firms, adverse selection results in the fact that less capable entrepreneurs will choose to involve VC firms in order to share the risk while more capable entrepreneurs will manage their companies without external involvement. Thus, the quality of VC-backed firms is usually “not the best due to the ante hoc effect of the venture capitalists’ participation” (Wang et al., 2003; p. 2018). The adverse selection hypothesis is diametrically opposite to the screening hypothesis. In the screening hypothesis the belief is that, since VC firms only participate in a small minority of the total stream of firms that seek external financing (Sahlman, 1990), these firms will be of better quality than firms that are not backed by VC (Chemmanur and Loutskina, 2006). This should lead to lower underpricing.

Conflict of interest hypothesis

(26)

Market power model

A more recently developed model is the market power hypothesis proposed by Chemmanur and Loutskina (2006). The market power hypothesis draws upon the notion that VC firms over time develop long term relationships with various key players in the IPO market (e.g. underwriters, auditors and institutional investors). This is caused by their role as a powerful recurring player in that market. These lasting relationships allow them to attract greater participation by these players (especially the investors) in the IPOs of their portfolio firms and obtaining a higher price for the equity of these firms. VC firms may be motivated to obtain higher valuations for the IPOs of their portfolio firms, due to the concern about their reputation with their own fund investors. This reputation is of outmost importance for VC firms’ ability to raise sufficient financing in subsequent funds (Gompers and Lerner, 1999a). This higher price for equity should lead to less severe underpricing and in some extreme examples to overpricing.

2.4.2 Empirical findings on underpricing in Venture Capital backed Initial Public Offerings

Since the studies by Barry et al. (1990) and Megginson and Weiss (1991) a stream of research seeking the relation between VC firms and the creation of public companies has emerged. Numerous articles have been devoted on the difference in underpricing in VC backed firms as opposed to non-VC backed firms. For this study, the literature can be divided into three categories: one group that documents a positive relationship between VC backing and underpricing (i.e. VC backing leads to more underpricing), a second group that finds a negative relationship and a third group that does not find a significant relationship at all. Table 2 on the next two pages provides an overview of the results of these studies. Furthermore, it offers an outline of independent variables that are used in the statistical regression models.

(27)

Da Silva Rosa, Velayuhten and Walter, 2003; Brau, Brown and Osteryoung, 2004; Klaassen and Von Eije, 2007).

Table 2. Results of PE and VC backed IPO underpricing studies

Authors Data Result Independent variables

Barry, Muscarella, Peavy and Vetsuypens (1990)

433 VC backed IPOs and an equal control sample of non VC backed IPOs for the period 1978-1987 (USA)

Lower underpricing Offer proceeds Underwriter rank

Aftermarket standard deviation SIC code

No. of VC investors

Pre IPO equity holding of all VC investors VC selling intensity

Non VC selling intensity

Equity holding of VC firms 1 year after IPO No. of board seats held by VC

Share of board seats held by VC

No. of board seats held by VC 1 year after IPO Duration of board seat lead VC

Age of lead VC

No. of prior IPOs lead VC Funds under management lead VC Megginson and

Weiss (1991)

320 VC backed IPOs and an equal control sample of non VC backed IPOs for the period 1983-1987 (USA)

Lower underpricing Offer proceeds

Market share of lead underwriters Age of the firm

VC dummy Hamao, Packer and

Rittter (1999)

456 VC backed IPOs for the period 1989-1995 (Japan)

Higher underpricing Age of the firm Offer proceeds Book to market ratio Auction results Time period dummies VC indicators

VC dummy VC = underwriter

VC = affiliated with securities firm / bank / SBIC / foreign

Francis and Hasan (2001)

415 VC backed IPOs and 428 non VC backed IPOs for the period 1990-1993 (USA)

Higher underpricing Offer price Firm size Offer proceeds Insider holding VC dummy Underwriter compensation Underwriter ranking Wang, Wang and

Lu (2003)

92 VC backed IPOs for the period 1987-2001 and an equal control sample of non VC backed IPOs (Singapore)

Lower underpricing P/E ratio

Book to market ratio Offer proceeds VC dummy Underwriter quality Auditor quality Age of the firm Ownership percentage Debt ratio

(28)

Authors Data Result Independent variables

Da Silva Rosa, Velayuthen and Walter (2003)

Total sample of 333 industrial IPOs of which 38 are VC backed for the period 1991-1999 (Australia) No significant difference in underpricing No regression model

Franzke (2003) 79 VC backed IPOs, 160 non VC backed IPOs and 61 bridge financed IPOs for the period 1997-2002 (Germany)

Higher underpricing Annualized volatility of first 20 daily returns No. of employees

Exhaustion of book building range Market trend

Bull/bear market dummy Underwriter rank VC dummy VC rank VC stake

Participation ratio (fraction of shares sold in the offering)

Dilution factor (no. of primary shares offered divided by the no. of pre-flotation shares) Non-underwriting costs of the IPO Lee and Wahal

(2004) 2,208 VC backed IPOs and a matched control sample of non VC backed IPOs for the period 1980-2000 (USA)

Higher underpricing Net proceeds SIC dummies Year dummies Underwriter rank

Headquarter-state dummies

Book value per share scaled by offering price Revenue per share scaled by offering price Total assets per share scaled by offering price EPS dummy

VC dummy Brau, Brown and

Osteryoung (2004)

126 VC backed

manufacturing firm IPOs and a control sample of 108 non VC backed IPOs (USA)

No significant difference in underpricing

Total assets prior to IPO No. of employees prior to IPO Offer proceeds

Market to book ratio High tech dummy Underwriter rank VC dummy Frederikslust and

van der Geest (2004)

38 PE backed IPOs and 68 non PE backed IPOs for the period 1985-1998 (Netherlands)

Lower underpricing (not significant)

IPO method (fixed / flexible) Lead manager

Age of the firm VC dummy Klaassen and von

Eije (2007)

16 VC backed IPOs and a control sample of 39 non VC backed IPOs for the period 1994-2005 (Netherlands)

No significant difference in underpricing

Age of the firm IPO year Size dummy

Net earnings growth dummy Technology dummy Underwriter quality Auditor quality VC dummy

VC experience (no. IPOs)

(29)

2.4.3 Influence of Venture Capital firm quality on underpricing

The grandstanding hypothesis developed by Gompers (1996) set off a new stream of research that looks for a relationship between VC firm characteristics and underpricing. As stated before, Gompers found that companies backed by young VC firms are more underpriced than those backed by more veteran VC firms. It could be said that VC firm age can proxy for VC firm quality. More veteran VC firms have been part of more IPOs and should therefore be more knowledgeable. This should increase their ability as a certifier and should lead to lower underpricing. After Gompers, who only looked at VC firm age as a distinguishing characteristic, Lin and Smith (1998) were the first to specifically focus on VC reputation in public offerings. This was followed by Ljungqvist (1999) and Dolvin and Pyles (2006), who perform similar studies with regard to VC firm quality4. Dolvin and Pyles believe that by merely looking at differences between VC backing and non VC backing, one “conceals any possible dependence that certification value may have with respect to the quality of this particular certification agent” (p.354). Next to VC firm age, a number of different measurements can be used for VC firm quality. Table 3 on the next page provides a similar overview as table 2; in this case of the literature on VC firm quality and underpricing. The table is expanded with a column for VC firm quality proxies.

Once more we can see that results from academic studies are ambiguous. However, it can be questioned whether VC firm age can proxy for quality. Age does not automatically bring along quality. If the results with VC firm age as a proxy for quality (Gompers, 1996; Barnes and McCarthy, 2002) are taken out from table 3, we find that apart from Lin and Smith (1998) in the remaining studies higher quality results in higher underpricing. This finding is opposite to what we would expect from the certification hypothesis with regard to quality of the certifier and the level of underpricing. It is not surprising that it are these scholars that criticize whether underpricing is the appropriate measure for wealth loss to the issuer in an IPO. Ljungqvist (1999) argues “that differences in underpricing per se are uninformative and possibly misleading when not controlling for differences in entrepreneurs’ incentives to control underpricing” (p. 2). This theory comes from Habib and Ljungqvist (2001), who state that “owners can affect the level of underpricing through the choices they make in promoting

4_{Both Ljungqvist (1999) and Dolvin and Pyles (2006) focus on wealth losses in IPOs as opposed to}

(30)

Table 3. Results of VC backed IPO underpricing studies with respect to firm quality

Authors Data Result VC quality proxy Independent variables

Gompers (1996) 433 VC backed IPOs for the period 1978-1987 (USA)

Younger VC firms show higher underpricing

VC firm age dummy VC firm age

No. of IPOs in previous 4 months

Offer proceeds

Standard deviation of stock return

Age of the firm

Duration of the VC’s board membership

Underwriter rank Lin and Smith

(1998) 497 VC backed IPOs for the period 1979-1990 (USA) Lower underpricing for more reputable VC firms

Total number of IPOs in which the VC firm was involved

Funds under management by the VC firm before the IPO

Funds under management by the VC firm as a percent of US total venture capital pool VC firm age

Number of deals where lead VC investor is lead

Offer proceeds Market run up Age of the firm

Ratio of revenue to assets Underwriter ranking Ljungqvist (1999) 513 VC backed IPOs 1996-1998 (USA) Higher underpricing for top 20 VC firms

VC league table ranking by no. of IPOs dummy VC league table ranking by

market capitalization of IPOs

Participation ratio (fraction of shares sold in the offering) Dilution factor (no. of primary

shares offered divided by the no. of pre-flotation shares) Age of the firm

Dollar fee for underwriting cover

Barnes and

McCarthy (2002) 85 VC backed IPOs for the period 1992-1999 (UK)

No significant

result VC firm age dummy Market liquidityOffering size

Aftermarket standard deviation Age of the firm

Duration of the lead VC’s board membership to the IPO Underwriter rank

Industry dummy Dolvin and Pyles

(2006)

976 VC backed IPOs for the period 1986-2000 (USA) Higher underpricing for higher quality VC firms Lead of VC syndicate dummy

No. of times lead of VC syndicate in IPO sample No. of time member of VC

syndicate in IPO sample Market share as lead VS

firm in a syndicate Forbes’ magazine ranking

dummy

Offer proceed Age of the firm High tech dummy Internet dummy Underwriter rank Overhang

Return of NASDAQ (15 days) Percentage change in the final

offer price Bubble dummy

(31)

later in instances of underpricing. The results from Ljungqvist (1999) and Dolvin and Pyles (2006) confirm this argument; IPOs backed by high quality VC firms show less wealth loss than those backed by low quality VC firms.

Although this wealth loss approach provides a new insight into underpricing of new issues, the concept of underpricing remains an undeniable wealth loss to initial owners. This is even the case in IPOs with large share retentions. Therefore, we consider underpricing to remain equally important to issuers. For the purpose of this study we draw from the certification theory as developed by Megginson and Weiss (1991) and the findings by Lin and Smith (1998), which predict that IPOs that are backed by more reputable, higher quality VC firms, will show less underpricing. From this theory we derive our testable hypothesis:

(32)

3. METHODOLOGY

This chapter presents an overview of the methodology that will be used to empirically test the hypothesis. Section 1 describes the measure for the concept of underpricing. Next, section 2 discusses the regression analysis and the testing for significance. Section 3 presents an overview of the independent variables and their expected relationship to underpricing used in the regression model. Finally, in section 4 the regression model will be presented.

3.1 Underpricing

The dependent factor this study examines is the level of underpricing. The notion of underpricing is best considered as a considerably lower offer price of a new issue compared to its first day closing price. Therefore, underpricing is defined as the initial return of the share on the first day, calculated as follows:

0 0 1 P P P IR  (1)

where IR is the initial return, i.e. the level of underpricing; P1 is the first day closing price of the share; and P0 is the original offer price.

3.2 Regression analysis

To test the possible relationship between the quality of the VC firm to the level of underpricing a regression model is constructed. A regression model describes the relationship between a dependent variable on the one hand of the equation and one or more independent variables on the other hand. A regression model takes the following form:

(33)

where y is the dependent variable; α is the intercept (“constant” term), βi is the respective parameter of independent variable xi; and ε is the model's error term. The error tem represents the unpredicted or unexplained variation in the response variable. In our model the dependent variable y is the level of underpricing and the independent variables xi are factors from IPO-specific information as described in section 3.3. This regression model will tested by means of the Ordinary Least Squares (OLS) method. OLS regresses underpricing on the independent variables (xi) and estimates the respective parameters (βi) for these variables.

Testing for significance

The outcomes of the regression model will be tested on two aspects of significance. Firstly and most importantly, the parameters of the independent variables in the regression will be tested on their significance. This test will provide an indication of the significance of our independent variables of interest to the level of underpricing. To test for this significance the t-test is conducted: ) ˆ ( ˆ * i i i SE t      (3)

where ˆ_i is the estimated value of the of the parameter; *

i

 is the hypothesized value; and )

ˆ ( _i

SE  is the standard error of the previously mentioned estimate. The critical value t will be assessed on three levels of significance, namely 1 percent, 5 percent and 10 percent.

Secondly, the regression itself will be tested on its explanatory power. Where the t-test tests for a single hypothesis (i.e. involving one coefficient), the F-test is suitable for testing multiple hypotheses. The F-statistic tests the null hypothesis that no relationship exists between the dependent variable and all of the independent variables and is calculated in the following manner: m k T x TSS URSS RRSS F    (4)

(34)

number of independent variables; and m is the number of restrictions. The critical value F is usually assessed on a 5 percent level of significance.

A second approach refers to the goodness of fit of the model. This means we look at how well the independent variables in the model actually explain variations in the dependent variable. In order to answer this question we calculate the R-square (R2) statistic:

TSS RSS

R2  1 ₍₅₎

where TSS is the total sum of squares, i.e. the total variation across all observations of the dependent variable about its mean value; and RSS is the residual sum of squares, i.e. the part of the variation the model was not able to explain. R-square must always lie between zero and one, where a value of one implies that the model has explained all of the variability of y about its mean value. However, R-square will never fall if more independent variables are added to the regression. This makes it impossible to determine whether a given variable should be included in the regression or not (Brooks, 2002). In order to bypass this problem the adjusted R-square is used:





_   _     2 2 ₁ 1 ₁ _R k T T R (6)

where _R2 _{is the adjusted R-square; T is the number of observations; and k is the number of}

independent variables. In case an extra independent variable is added to the model, k increases and unless R-square increases by more than an offsetting amount, the adjusted R-square will actually fall.

Linearity

(35)

other operation of such nature (Brooks, 2002). To test for linearity in our model a Ramsey RESET test – a general test for misspecification of the functional form – will be performed. Heteroskedasticity

Furthermore, when using OLS a number of assumptions are made. One of these is that the error term has a constant variance. This will be the case if the observations of the error term are assumed to be drawn from identical distributions. Heteroskedasticity is a violation of this assumption. Heteroskedasticity does not cause OLS coefficient estimates to be biased. However, the variance – and thus the standard error – of the coefficients tends to be underestimated, inflating the results from t-tests. This can cause insignificant variables appear to be statistically significant. To test for the presence of heteroskedasticity among the errors, White’s general test for heteroskedasticity is performed. Both the Ramsey RESET test and White’s general test will be assessed on their F-statistic on a 5 percent significance level.

3.3 Independent variables

As is shown from the extensive set of literature on underpricing, the level of this underpricing is influenced by a large number of variables. To control for these factors we include a number of them in our regression. We group these variables under four headings: variables related to offering characteristics, issuing firm characteristics, market characteristics and VC firm characteristics. Where the first three groups are included as independent control variables, the last group on VC firm characteristics is of our primary interest. In the upcoming sections these variables are discussed and their expected relationship to the level of underpricing is predicted. The regression term is provided in small caps between brackets.

3.3.1 Offering characteristics

Offer price (PRICE)

(36)

(1987) and Ibbotson, Sindelar and Ritter (1988), who reported a negative relationship between the offer price and underpricing.

Offer proceeds (PROCEEDS)

The offer proceeds (offer size) are defined as the offer price times the number of shares sold to the public. In line with Francis and Hasan (2001), the log of the proceeds is included to control for the level of asymmetry. We follow Tinic (1988), who shows that smaller, more risky firms tend to make small offerings and Beatty and Ritter (1986), who document that large offerings are less underpriced. Therefore, we expect a negative relationship between the offer proceeds and underpricing.

Retention of shares (RETENTION)

As discussed in the literature, retention of shares5 can act as a certifying mechanism in the IPO. By retaining a stake in the company the issuers (i.e. all shareholders prior to the IPO) signal the quality of the firm, since retention diminishes profits from certifying falsely and can therefore be an instrument for reliable certification (Megginson and Weiss, 1990). This should lead to lower underpricing in the offering. However, the concept of retention could be addressed from another point of view. By holding back shares in the offering, the issuers preserve the opportunity to gain from future upward price movements of the shares. In this sense retention could convey information about the future prospects of the firm (Gumanti and Niagara, 2006) and should draw additional interest on the offering from outside investors. This increased interest will drive up the price and the magnitude of the initial return (underpricing). In line with this reasoning, a number of scholars find a positive relationship between retention and the level of underpricing (Clarkson, Dontoh, Richardson and Sefnik, 1991; How and Low, 1993). We follow these findings and expect a positive relationship between retention of shares and the level of underpricing. In our measure for retention we do not take the greenshoe or over-allotment option6 into account.

5 _{Other scholars (Ljungqvist, 1999; Franzke, 2003) refer to the participation ratio of the IPO, i.e. fraction of}

shares sold in the offering. Both approaches basically address the same concept from a different starting point. Where the participation ratio looks at the stake that is sold to the public, the retention approach looks at the remaining stake that is held back by the issuer(s).

(37)

Auditor quality dummy (AUDITOR)

The inclusion of auditor quality stems from the certification hypothesis. Similarly to VC firm backing, companies that are backed by high quality auditors can act as a certifying mechanism for good quality issues (Balvers, McDonald and Miller, 1988). In line with the certification hypothesis, we expect to find a negative relationship between auditor quality and underpricing. To assess quality, we follow the methodology from Balvers et al. (1988) and Feltham, Hughes and Simunic (1991). A dummy variable is constructed that is coded 1 if the auditor is one of the ‘Big Four’ auditors and 0 otherwise. The ‘Big Four’7 auditors are Deloitte, Ernst & Young, KPMG and PricewaterhouseCoopers.

Underwriter quality (UNDERWRITER)

In line with auditors, high quality underwriters can act as a third party certifier in IPOs as well. More prestigious underwriters are often perceived to be associated with high quality offerings. Carter and Manaster (1990) show that IPOs led by these prestigious underwriters are associated with lower underpricing. As a measure for underwriter quality or prestige, we construct a variable based on the underwriter rankings from the Thomson Financial IPO League Table (section C7c) for each year in our sample (2004 to 2007). For every year an underwriter ranks in the top ten of the Thomson league table the variable increases by 1 (i.e. underwriters that are top ten ranked in every year of our sample are awarded a 4; underwriters that are top ten ranked in three of the four years are awarded a 3; etc.). In line with Carter and Manaster, we expect a negative relationship between underwriter quality and the level of underpricing.

3.3.2 Issuing firm characteristics

Firm size

Francis and Hasan (2001), Franzke (2003) and Brau et al. (2004) find that large companies tend to be less underpriced, since their size diminishes the asymmetric information problem. Therefore, firm size and the level of underpricing should show a negative relationship. There are a number of measures that could proxy for firm size, of which the number of employees and total assets at the date of the IPO are most commonly used. However, as one can imagine,

7_{Balvers et al. (1988) use the ‘Big Eight’ as high quality auditors and Feltham et al. (1991) the ‘Big Six’.}

(38)

firm size and offer proceeds tend to show a high level of correlation. This is predominantly the case when total assets are used. In order to prevent multicollinearity8 in the regression, we use the natural log of the number of employees (EMPLOYEES) as a proxy for firm size.

ICB code dummies

Pharmaceuticals, high technology companies as well as internet-related firms are considered to be more risky and are therefore associated with higher levels of underpricing (Dolvin and Pyles, 2006). A dummy variable is constructed to account for issuing firms in these sectors and others of similar increased risk in the regression. We construct one 3-digit Industrial Classification Benchmark (ICB) code dummy and one 2-digit ICB code dummy, which will cover industries expected to show higher levels of underpricing. These are the following: a 457x dummy (ICB457), which covers the Pharmaceuticals & Biotechnology Sector and a 95xx

(ICB95) dummy, which covers the Technology Supersector.

3.3.3 Market characteristics

Market trend (MARKET)

In line with Franzke (2003) we use the market trend in the underpricing regression. She documents that there seem to exist psychological and market factors that lead to a positive relationship between the respective stock market and the degree of underpricing. As does Franzke, the forty day market return prior to the IPO date is used. The corresponding market we use for each IPO is the primary stock market of its respective country of offering, since these markets best reflect trends in investor sentiment. We expect the return to be positively related to underpricing.

IPO year

In order to control for changes in investor climate and sentiment we include a year variable in our regression. Similarly to the market trend, a positive investor climate could lead to a positive relationship with the level of underpricing. Year 2004 will be taken as base year and therefore not be included in the regression. We construct year dummies that account for the

8_{Multicollinearity is a statistical term for the existence of a high degree of linear correlation amongst two or}

(39)

remaining years our sample covers (2005, 2006 and 2007). The resulting coefficients of the dummies show the deviation in the level of underpricing for its respective year as opposed to 2004.

3.3.4 Venture Capital firm characteristics

VC firm age (VCAGE)

The age of the VC firm is not included as a quality measure in this study. However, the variable is examined in the regression to look for support of the grandstanding hypothesis (Gompers, 1996). In line with the hypothesis, we expect the age of the VC firm to be negatively correlated to the level of underpricing. Young VC firms have the incentive to bring a company to the public sooner and underprice their offerings more to show their ability to potential investors. The age of a VC firm is defined as the time span between the year of incorporation and the year of the offering.

VC firm quality

The notion of quality of a firm – whether it is a VC firm or any other – is a rather subjective one. However, there are a number of indicators that could proxy for this quality in the case of VC firms. In the small set of VC quality literature the involvement of the VC firm in IPOs is the most frequently used one. A greater involvement in IPOs – whether in number or in size – should increase the perceived quality by the general public, as it adds to the reputation and recognition of the VC firm. In line with Lin and Smith (1998) and Dolvin and Pyles (2006), we distinguish three different variables that draw upon this IPO involvement proxy for quality. Firstly, we construct a variable based on the number of IPOs in which the VC firm has participated in our sample (VCIPO). Secondly, we include a variable based on the log of

the total offer proceeds of all IPOs in our sample where the VC firm was involved in (VCPROC). Thirdly, a variable based on the log of the average offer proceeds of the VC firm’s

IPOs in our sample is constructed (VCAVGPROC).

(40)

coded one if the VC firm is a member of the EVCA and zero otherwise. The EVCA is the only association that recognizes VC firms on a European scale and unites them as such. Admission to the EVCA is subject to approval. We expect a VC firm with an EVCA membership to be perceived as one of higher quality.

Finally, we construct three interaction dummies between the previously mentioned quality proxies. The dummies are created in the following manner. For the three IPO involvement measures league tables are created, i.e. the VC firms are assigned ranks according to their number of IPOs, total IPO proceeds and average IPO proceeds. The dummy variable takes the value of one in IPOs where the VC firm ranks in the top ten of one of these tables and is a member of the EVCA (VCIPO*EVCA, VCPROC*EVCA and VCAVGPROC*EVCA). These dummies

should provide an even greater and more distinct measure for quality, since they take both proxies of VC quality into account; one based on IPO involvement and one on EVCA membership.

It should be noted that only the lead VC firm is taken into account, i.e. the VC firm with the largest stake in the issuing company at the time of the IPO. Therefore, we consider one VC firm per IPO. Furthermore, in order to prevent multicollinearity, all indicators for VC firm quality will be tested independently on their significance in the model, since they are expected to show a high level of correlation. In line with Lin and Smith (1998), we expect a negative relationship between the quality proxies of the VC firm and the level of underpricing.

3.3.5 Summary of independent variables

(41)

Table 4. Overview of independent variables

Independent variable Regression term Expected sign

Offer price PRICE

-Offer proceeds PROCEEDS

-Retention of shares RETENTION +

Auditor quality AUDITOR

-Underwriter quality UNDERWRITER

-Firm size EMPLOYEES

-ICB code ICB457 +

ICB95 +

Forty day market run up MARKET +

IPO year 2005 0

2006 ₀

2007 0

VC firm age VCAGE

-VC firm quality VCIPO

-VCPROC -VCAVGPROC -VCEVCA -VCIPO*EVCA -VCPROC*EVCA -VCAVGPROC*EVCA -3.4 Regression model

From the information provided above, we construct the following regression to be tested:

                                VCQUALITY VCAGE MARKET ICB ICB EMPLOYEES R UNDERWRITE AUDITOR RETENTION PROCEEDS PRICE IR 14 13 12 11 10 9 8 7 6 5 4 3 2 1 2007 2006 2005 95 457 ] [log ] [log (7)

Equation (7) will be tested on the different measures for quality (VCQUALITY) as described in

(42)

4. DATA

This chapter introduces the data sample. Section 1 describes the way the sample is constructed and the sources of the data. In section 2 descriptive statistics of the sample are presented.

4.1 Sample construction and data sources

As we aim to study the influence of VC quality on the level of underpricing of IPOs, initially all VC backed IPOs in the enlarged European Union9 from after the Internet bubble in 2001 are gathered. We purposely take IPOs from after the bubble to avoid as much noise as possible in our sample. These IPOs are taken from Bureau Van Dijk’s Zephyr database on mergers and acquisitions. A critical requirement in VC backing IPO studies is an appropriate definition of VC; if this is not the case, the results could be distorted by IPOs that are backed by buy-out financing, which are most of the time more notable and larger PE firms (e.g. KKR, Blackstone Group and CVC Capital Partners). Zephyr provides the option to select IPOs that are backed by VC and PE (buy-out financing). This gives us an initial sample of 268 IPOs. From there on we manually excluded all IPOs that were backed with buy-out financing, i.e. firms on which was apparent that they have been part of a buy-out (e.g. MBO or LBO). In addition to that, IPOs on which a reasonable doubt arose over the origin of financing were excluded as well. Furthermore, the database provided just two IPOs in 2002 and 2003 with very limited information, which instigated us to set up our time frame from 2004 to 2007. Having removed the IPOs that did not meet the criteria, we found a sample of 175 VC backed IPOs. On this sample we will base our quality proxies for the VC firms. However, not all required data on the independent control variables was available for the complete sample. The testing of the regression will therefore be conducted on a final sample of 106 VC backed IPOs.

As is mentioned above, the IPOs are taken from the Zephyr database. This database is used to collect the data on offering characteristics as well. Market data is taken from Thomson

U NDERPRICING V ENTURE C APITAL Q UALITY AND

AND

U

NDERPRICING

An empirical investigation on the relationship between

Venture Capital firm quality and the level of underpricing

Author:

A.M. Jacobs

Supervisor:

dr. J.H. von Eije

January 2008

VENTURE CAPITAL QUALITY AND UNDERPRICING

An empirical investigation on the relationship between

Venture Capital firm quality and the level of underpricing

PREFACE

TABLE OF CONTENTS

LIST OF TABLES

LIST OF FIGURES

1. INTRODUCTION

2. LITERATURE REVIEW

3. METHODOLOGY





4. DATA