0 Thesis Finance
MSc Business Economics Nina Dréau (6149642) Supervisor: Dr. J.E. Ligterink
July 7th, 2016
Corporate Venture Capital in a multiple agency framework
and its effect on IPO underpricing
Abstract
This empirical study investigates whether two different VC-investor types, corporate and traditional, differ in ability and motivation to mitigate the agency problems arising in a multiple agency framework upon the event of an IPO, resulting in a differential underpricing level. Although there is a growing body of literature on the superior value-adding ability of corporate VCs (CVCs) upon exit events, research on the downside costs of an exit event is lacking. Using a sample of 1339 US IPO firms going public between 1995 and 2015, the findings of this study suggest the level of underpricing to be 15.6% higher for CVC-backed IPO firm, compared to traditional VC-backed (TVC-backed) firms. Controlling for different firm characteristics, the difference in underpricing mainly results from the two VC-investor types picking portfolio firms differing in profitability, size, and expected growth. Furthermore, the difference in underpricing level is almost entirely driven by the bubble period and, although insignificant, the financial crisis. CVC- and TVC-backed firms show no significant difference in valuation of All-star analyst coverage as proxy for one of the non-price dimensions of an underwriter. The results imply CVC investors are worse monitors of the underwriter compared to TVC investors, especially when they are investing for strategic reasons.
1
Statement of originality
This document is written by student Nina Dréau who declares to take full responsibility for the content 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.
Acknowledgement
Hereby I want to thank the people who made it possible to create these 45 pages of my thesis. Thank you Dr. J.E. Ligterink for the useful feedback moments. Thank you Martijn, Ottie and JP for listening to my complaints and always being supportive. Thank you Else, Chloé, Milou, Yoeri, and all my other study mates for drinking the many coffees at the Krater and making writing my thesis enjoyable.
2
List of abbreviations
CVC – Corporate Venture Capital
“Established firms investing minority equity in an entrepreneurial venture” (Dushnitsky, & Shapira, 2010)
TVC – Traditional Venture Capital
“Dedicated, independent, and professionally managed pools of capital investing equity in privately held entrepreneurial ventures”
IPO – Initial Public Offering PE – Private Equity VC – Venture Capital
VCs – Venture Capital investors
CVCs – Corporate Venture Capital investors TVCs – Traditional Venture Capital investors R&D – Research & Development
3 Table of Contents Statement of originality ... 1 Acknowledgement ... 1 List of abbreviations ... 2 1. Introduction ... 4 2. Literature review ... 7
Venture Capital funds ... 7
Venture Capital history ... 8
Corporate vs. Traditional Venture Capital ... 9
Underpricing ... 11
The Multiple Agency Framework ... 13
Non-price dimensions ... 16
3. Methodology ... 17
Part I ... 17
Part II ... 19
4. Data & Descriptive Statistics ... 19
Sample Selection ... 19
Descriptive statistics ... 21
5. Results ... 22
Testing control variables ... 22
Propensity Score Matching ... 23
All-Star analyst coverage & Underpricing ... 26
Corporate Venturing & lust for All-Star analyst coverage ... 28
6. Robustness Checks ... 29
7. Conclusion & Discussion ... 33
8. References ... 35
4
1. Introduction
In 2015 Corporate Venturing accounted for 13% of the total dollar amount invested in the entrepreneurial ecosystem, which is equal to 21% of all venture capital deals (National Venture Capital Association, 2016). From the mid-90s onwards, corporate venture capital investments – an established firms investing minority equity in an entrepreneurial venture (Dushnitsky, & Shapira, 2010) – substantially increased (Gompers & Lerner, 2001). Corporate venture capital investors (CVCs) differ from purely financial focused traditional capital investors (TVCs) in terms of their investment rationale (Benson & Ziedonis, 2010), incentive schemes (Dushnitsky & Shapira, 2010), and organizational structure (Luukkonen, Deschryvere & Bertoni, 2013). Although there is a growing body of literature on CVCs being able to obtain similar or higher market valuations upon exit events (e.g. Gompers & Lerner, 2001; Ivanov & Xie, 2010), evidence on the costs of an exit event is lacking. This empirical study aims at filling part of this gap by investigating the effect of CVC-backing on the level of underpricing when an entrepreneurial firm decides to go public.
In order to scale the business entrepreneurial ventures are in need of external funding, but due to high levels of asymmetric information, high uncertainty levels, and the lack of historical data on firm performance, attracting financing in traditional ways is difficult. Debt is not easily available to these firms, making other external sources of funding such as venture capital investments an important intermediary in financial markets (Gompers & Lerner, 2001). Venture Capitalists (VCs) are private equity (PE) investors who, unlike regular PE investors, invest in young, privately held companies. These young companies often face high-risks on the one hand, and potentially high-returns on the other. Receiving venture capital (VC) funding yield important benefits for an entrepreneurial firm, such as strategic advice, monitoring, certification, and corporate governance, but comes at the cost of losing control (Hellmann & Puri, 2000).
At some point in the entrepreneurial lifecycle there are two main exit strategies for a venture: going to the market via an Initial Public Offering (IPO) or search for an acquirer willing to take over the company. Either of these exit strategies enable the initial shareholders to cash out and gain a return on the investment. The high information asymmetry concerning the venture value allows the underwriter to underprice, making a profit at expense of the venture (Arthurs, Hoskinsson, Busenitz & Johnson, 2008). On the supply side underwriters want to maximize profit from an IPO process via underpricing – the difference between the offer and after-market price (Chahine, Filatotchev & Wright, 2007), but on the demand side issuers want to avoid excessive underpricing leading to a wealth transfer from old shareholders to new ones (Lee & Wahal, 2004). Gompers & Lerner (2001) suggest VCs are able to mitigate the agency problems arising from information asymmetry, but
5 simultaneously new agency issues arise, including multiple actors. The complicated ownership structure in a (corporate) VC-backed venture going public raises conflicts of interests between the multiple actors when bringing the venture firm public. The multiple agent theory extents the original agency theory, examining conflicts of interests among more than one principal-agent group for which at least one agent is connected to a different principal (Arthurs et al., 2008). In light of the underwriting process of VC-backed IPO firms one can distinguish between three main actors in a multiple agency relationship: the founder (the entrepreneur), the lead venture capitalist, and the underwriter.
To the question why some IPO firms are more underpriced than other, Habib & Ljungqvist (2001) suggest besides differences in information asymmetry, the valuation uncertainty and the risk of lawsuit, a possible explanation for underpricing differences is found in the degree the issuer cares about this underpricing. The aim of this study is to investigate whether corporate and traditional venture capitalists (TVCs) – “independent, professionally managed, dedicated pools of capital that focus on equity or equity-linked investments in privately held, high-growth companies” (Gompers & Lerner, 2001, pp.146) – differ in ability and motivation to mitigate the agency problems arising in a multiple agency framework upon the event of an IPO, resulting in a differential underpricing level. In the past, controlling for a selection bias, multiple studies found a higher level of underpricing for VC-backed IPO firms compared to non-VC-VC-backed IPO firms (e.g. Lee & Wahal, 2004; Arthurs et al., 2008). Whereas previous studies assumed all VCs are traditional venture capital funds with similar characteristics, CVCs differ from TVCs in organizational structure, investment rationale, and incentive scheme. This study differ from previous research by distinguishing two VC investor types and study their separate behaviors in the underwriting process investigating the following research question:
RQ: What is the effect of Corporate Venture Capital investments on the level of underpricing of an IPO firm?
Due to the incentive schemes of CVCs being less performance based (Dushnitsky & Shapira, 2010), the rationale for investing being strategic besides solely financial (Benson & Ziedonis, 2010), and the organizational structure of CVCs lacking limited partners as principals (Luukkonen et al., 2013), it is relevant to investigate the consequences of corporate venturing in the multiple agency framework. The time-horizon for example differs for both investor types due to the organization structure (Luukkonen et al., 2013), which according to Arthurs et al., (2008) affect the ability to monitor in multiple agency relationships and protect against underpricing.
6 Besides the conflicts of interest arising from the multiple agency theory due to the degree the actors care about underpricing, another factor contributing to excessive underpricing is found in the non-price dimensions of an underwriter (Liu & Ritter, 2011). Liu & Ritter (2011) suggest issuers not only care about the price-dimensions, such as the IPO proceeds, but also about the non-price dimensions such as All-star analyst coverage. The authors find that especially VC-investors are willing to pay for these non-price dimensions via the acceptance of higher underpricing levels. Based on the different characteristics of corporate and traditional VCs, this study does not only investigate whether the two investor types differ in monitoring behavior towards the underwriter, but as well whether this behavior could (partly) be explained by their differing valuation of one of the non-price dimensions – All-star analyst coverage.
To control for unobserved factors, such as the investment rationale, the propensity score matching method is used to compare the level of underpricing for a CVC-backed IPO firm to that of a TVC-backed firm with similar firm characteristics. This method is based on previous studies on venture financing suggesting venture financing is not randomly assigned but due the ex-ante analyses conducted by the VC and the design of contracts to mitigate the agency- and information asymmetry problems receiving VC-backing is endogenous (e.g. Lee & Wahal, 2004; Chemannur, Louskina & Tian, 2014). Also, additional robustness checks are conducted to test whether the differential results for underpricing levels between the two investor types hold in an extreme economic conditions. Due to the need to fasten the speed of innovation, in a more competitive environment nurturing external innovation grow in importance, leading to an increase in CVC investments (Fulghieri & Sevilir, 2009). This suggests the difference in investment rationale, being strategic or solely focused on gaining a financial return, is not consistent over time.
The findings of this study have important implications for the entrepreneur and the choice among different venture capital investors. Although limited, alternative sources of external funding other than debt are available to grow a business. It is important to understand what distinct the different external funding alternatives in both, positive and negative consequences such as losing ownership share and costs arising in a multiple agency framework. Besides new insights for the entrepreneur, from an academic perspective this study extent the earlier research on the effect of VC-backing on underpricing (e.g. Lee & Wahal, 2004; Liu & Ritter, 2011) by distinguishing between the two different investor types, corporate and traditional VCs. Also additional insight in the multiple agency theory upon the event of an initial public offering is provided, extending earlier findings by Arthurs et al. (2008).
7 The remainder of this paper is organized as follows: based on previous literature chapter 2 elaborates on (the history of) VC, the differences between corporate and traditional VC investors and the underpricing phenomenon. Also an elaboration is given on the multiple agency theory and the non-price dimensions in the underwriting process, ending with the formulation of the hypotheses. Thereafter, chapter 3 describes the methodology, chapter 4 include the data and descriptive statistics, followed by a discussion of the results in chapter 5 and robustness checks in chapter 6. The final chapter gives a concluding remark on this empirical study.
2. Literature review
This chapter elaborates on the history and rationale behind (corporate) venture capital and the differences between traditional and corporate VC investors. Furthermore, the concept of underpricing is discussed in terms of the multiple agency theory and non-price dimensions. Based on previous literature the research gap is introduced and finally hypothesis are formulated
Venture Capital funds
The entrepreneurial business cycle start with a business idea, developing this into a viable product, finding a market for the product, scale the business, and in some cases get ready for an initial public offering (IPO) or acquisition. The different phases an entrepreneurial firm pass can be split into the start-up phase, the scaling-phase and the going-to-the-market phase. In order to reach the second or third phase a crucial aspect is raising sufficient funding to scale the business, a challenging task for the entrepreneur (e.g. Hellmann, 2002; Benson & Ziedonis, 2010). One of the reasons for the turnover of innovative ideas into new world-class companies is the size and sophistication of the venture capital industry in the US (Maula, Autio & Murray, 2005). Whereas innovative start-ups are often found by managers who are highly technically skilled, venture capital investments enable these start-ups to grow and reach the full commercial potential.
Small, young firms often face higher levels of asymmetric information, high uncertainty levels, and lack historical data on firm performance, making it difficult to attract financing in traditional ways. Since debt is not easily available to these firms, external sources of funding such as venture capital investments form an important intermediary in financial markets (Gompers & Lerner, 2001). Venture Capitalists (VCs) are private equity (PE) investors who, unlike regular PE investors, invest in young, privately held companies. These young companies often face high-risks on the one hand, and potentially high-returns on the other.
VC investors can add value to a venture via assisting in the screening process of new employees, in coaching and mentoring, and providing access to a large network of contracts including investment
8 bankers (Croce, Marti & Murtinu, 2013). Hellmann & Puri (2002) empirically test whether VCs play a role in professionalizing a start-up company and the findings suggest VCs are more involved with their portfolio companies than traditional financial intermediaries such as banks. VCs also have a better ability to nurture human capital and professionalize the start-up company. Whereas traditional financial intermediaries primarily used to have a monitoring role, dealing with information asymmetry and corporate governance issues, venture capitalists have additional roles such as supporting teambuilding or CEO turnover (Hellmann & Puri, 2002). On the other hand, VC funding also comes at costs such as significant loss of cash flow and control rights (Hellmann & Puri, 2002) and underpricing (Lee & Wahal, 2004; Liu & Ritter, 2011).
Bettignies & Brander (2007) examine the choice of an entrepreneur between the traditional external finance opportunities such as banks, and venture capital. Besides the limited accessibility to debt, this choice includes a tradeoff between keeping full control over the company for the entrepreneur and the ability of the venture investor to provide substantial managerial contribution to the venture. In the first case, the entrepreneur is fully incentivized to exert effort, but in case of venture capital investments a two-sided moral hazard problem arise. Both, the VC as well as the entrepreneur, must have incentives, such as equity ownership, to exert effort in the start-up company. When increasing the equity stake for the VC the entrepreneur’s incentive decrease and should be compensated for by high-value managerial input from the VC (Bettignies & Brander, 2007). One consequence for an entrepreneur losing ownership share at the expense of a (corporate) VC investor, who might have a different objectives with the venture, is the rise of new agency issues.
Venture Capital history
Although the first true VC firm was raised in 1946, the venture industry boomingly increased only in the late 1970s and early 1980s. Since mid-1990 nearly 40% of the total number of IPOs was venture capital backed (Bradley, Kim & Krigman, 2015). The investor type changed from being mainly individual investors to pension funds accounting for the largest part of the investments (Gompers & Lerner, 2001). Not only did the traditional VC investments – those investments made mainly with a financial objective – increased extensively, also a new form of venture investments emerged, venture investments made by corporates.
For many, the first phenomenon one gets familiarized with when studying economics is that of market competition. In order to remain competitive in the continuously changing environment, exploring growth opportunities, making technological progress and being innovative is of high importance. Whereas traditionally Research and Development (R&D) departments used to be the main source of innovative output, increased competition and shorter product life cycles incentivize
9 corporate firms to fasten the speed of innovation (Enkel, Gassmann & Chesbrough, 2009; Fulghieri & Sevilir, 2009). Due to an increasingly competitive environment and a decreasing product lifecycle, corporates nowadays are more and more forced to look for outside innovation opportunities complementary to internal oriented, centralized R&D centers (Vanhaverbeke, van de Vrande & Chesbrough, 2008). Throughout the years a remarkable shift has taken place from mainly internally oriented innovation to using knowledge and expertise from outside the firm to innovate, a phenomenon referred to as open innovation. Open innovation – defined as “the use of purposive inflows and outflows of knowledge to accelerate internal innovation, and expand the markets for external use of innovation, respectively” (Chesbrough, 2006, p. 2) – has gained more and more attention in both the academic world and in practice. One effective type of open innovation for corporations is establishing corporate venture capital (CVC) funds and nurture external innovation (Chemmanur, Loutkina & Tian (2014).
Corporate vs. Traditional Venture Capital
In the venture capital literature one can distinguish between two types of venture capitalists based on the value-adding ability to the portfolio firms, organizational structure, investment rationale, and incentive scheme: traditional venture capitalists (TVCs) – “independent, professionally managed, dedicated pools of capital that focus on equity or equity-linked investments in privately held, high-growth companies” (Gompers & Lerner, 2001, pp.146) – and corporate venture capitalists (CVCs) - an established firms investing minority equity in an entrepreneurial venture (Dushnitsky, & Shapira, 2010).
In the past, the differential value-adding ability of corporate VCs to the portfolio ventures has been extensively studied. Different VCs can be distinguished based on their ability to provide nonfinancial resources besides the provision of capital funding. These nonfinancial resources, such as expertise in product development, can lead to substantial performances (Park & Steensma, 2012). The value-adding ability of corporate VC-investors differs from that of traditional VC-investors, both in terms of origin and consequences (Maula & Murray, 2001; Maula, Autio & Murray, 2005). Traditional VCs add value via enterprise nurturing, including activities such as fund raising, key employee recruitment, and professionalizing the company. Corporate VCs additionally help venture firms build commercial credibility, are able to provide technological support (Maula & Murray, 2001; Maula, Autio & Murray, 2005), and are able to offer more industry experience than TVCs (Chemmanur, Loutskina & Tian, 2014). Besides additional value-adding abilities, receiving corporate venture capital backing is subject to several important drawbacks. Firstly, CVCs lack the incentive schemes of TVCs to maximize the venture firm value, build on performance-based compensation (Gompers & Lerner, 2000;
10 Chemannur et al., 2014). A second drawback of receiving CVC investments is the restriction on accessing complementary assets from other parties than the corporate firm in the open market (Park & Steensma, 2012). Moreover, Hellman (2002) suggests the success of the venture can either complement or cannibalize the assets of the corporate firm, depending on the venture characteristics. If the ventures’ assets complement the corporate assets, synergies can be achieved and choosing a strategic investor is fruitful. If the ventures’ asset is a treat to the corporate the strategic investor cares about equity holding without providing support of holding or board seats. Due to the equity stake of the CVC investor, the stake of traditional VCs, as well as their incentive to provide support, is reduced (Hellmann, 2002). A last negative consequence of CVC-backing involves the desire to maximize firm value of the corporate parent via corporate venturing instead of maximizing the portfolio firm value (Chesbrough, 2002).
Next to the value-adding ability, a second important difference between CVCs and TVCs enhances the investment rationale. Hellmann (2002) suggests TVCs mainly invest for financial reasons – maximizing financial return for the limited partners, whereas CVCs often have an additional strategic motive to invest (Chemmanur et al., 2014). Park & Steensma (2012) suggest the main objective of a traditional VC is to realize a maximized capital gain at an exit event such as an IPO or acquisition, whereas additional strategic investment motives for the corporate investor include nurturing external innovation or follow technological trends. Other strategic reasons to invest are the identification of potential acquisition opportunities (Benson & Ziedonis, 2010) or achieving synergies between the venture and the corporate business (Hellmann, 2002). Although the investment rationale of CVCs is more strategic in nature, previous literature shows no consensus on different exit strategies between the two investor types. Because CVC-investors often have strategic reasons to invest, such as developing technologies, which are relevant for the corporate itself, the portfolio company is more likely to be acquired by the corporate firm (Hellmann, 2002; Cumming, 2008). On the other hand, Guo, Lou & Perez-Castrillo (2015) suggest potentially more successful ventures are more likely to exit through an IPO, whereas the venture with a lower probability of being successful are more likely to be acquired. Further findings suggest CVCs invest larger amounts of capital in the portfolio ventures, increasing the probability of the venture being successful and thus the likelihood of the venture exiting via an IPO. Furthermore, Park & Steensma (2012) examined the performance of venture capital backed companies in terms of ability to exit trough an IPO and concluded CVC-backed ventures are more likely to go public and are less likely to fail compared to ventures without CVC backing. Whereas the investment rationale of corporate investors is often strategic, according to Chesbrough (2002), CVCs investing mainly for financial reasons aims at gaining a return on
11 investment above that of TVCs due to superior industry experience, the strong balance sheet, and the longer time horizon.
Thirdly, corporate and traditional VCs differ in terms of incentive schemes for investment managers. Dushnitsky & Shapira (2010) investigate the effect of compensation of CVC investors. The authors study the principal-agent framework, analyzing how incentives shape managerial investment behavior leading to certain performance. In light of the principal-agent framework, a tradeoff exists between the risk sharing from the agents’ perspective and the motivation to exert effort from the principal perspective (Dushnitsky & Shapira, 2010). The compensation schemes are thus shaping managerial behavior and its consequences. The findings suggest the investment practices between CVC and TVC investors differ mainly due to the lack of performance pay for CVC investors. Increasing the performance pay incentive lead to more aggressive investment practices (Dushnitsky & Shapira, 2010). Furthermore, one of the consequences of the aggressive behavior, the succession rate of portfolio exits, increase when CVC investors are incentivized by performance pay. Additionally, Fulghieri & Sevilir (2009) find that CVCs can limit the dilution of ownership in an entrepreneurial venture by increasing the financial funding, leading to better effort incentives for the entrepreneur. Besides different incentive schemes and investment rationale, one of the most important differences between the two VC investor types is the organizational structure, affecting the multiple agency framework at the IPO event and the time-horizon of the investors. CVC funds have corporate subsidiaries whereas TVC funds are structured as general partners raising capital from limited partners, which they invest in a portfolio of ventures (Luukkonen et al., 2013). The differences in organizational structure have important implications for the behavior of the investment managers, especially in event of an initial public offering. Not only implies the differential organizational structure between the two investor types a different multiple agency setting, it also affects the time-horizon of the investor. The investment time-horizon of traditional VCs is finite and at the end of the fund lifespan the VC is pressured to return a profit to the limited partners. Due to the lack of contractually enforces lifespans Chemmanur, Louskina & Tian (2014) suggest CVCs have a longer investment horizon compared to TVCs.
Underpricing
At some point in the entrepreneurial lifecycle, a venture might decide to raise new capital via an initial public offering (IPO) – for the first time issuing equity to the public. Concerning the underwriting process there is a tradeoff between old shareholders’ wealth maximization and profit maximization for the underwriter. On the supply side underwriters want to maximize profit from an IPO process via underpricing – the difference between the offer and after-market price (Chahine,
12 Filatotchev & Wright, 2007), but on the demand side issuers want to avoid excessive underpricing leading to a wealth transfer from old shareholders to new ones. With regard to this tradeoff, underwriters should be careful not to lose IPO mandates due to exerting too much underpricing (Liu & Ritter, 2011).
Although on first sight the underwriter market seems to be one of competition due to the large number of underwriters, likely to result in a minimized profit, Goldstein, Irvine & Puckett (2011) estimate still 45% of the first day return is captured by underwriters. Underpricing can be viewed as wealth transfer from old shareholders to new ones, since the new shareholders can buy the initial shares publicly offered at a discount. If sold at the first day closing price instead of the offering price total proceeds would have been higher, or if less shares would have been issued, the dilution of ownership of initial shareholders lower (Loughran & Ritter, 2002). Academics have extensively studied the underwriting puzzle – why so many underwriters co-exist with high levels of underpricing – and offer several possible explanations:
Firstly, high underpricing is found at IPOs for which the market price is higher than originally expected, balancing out the cost of underpricing via the unexpected extra wealth (Loughran & Ritter, 2002). Second, due to the presence of hot and cold issue periods together with the given that offer prices partly adjust to public information lead to predictable first-day returns can based on lagged market returns (Loughran & Ritter, 2002).
Past research conclude the level of underpricing increase when the IPO firm is backed by a venture capital investor (e.g. Arthurs et al., 2008; Liu & Ritter, 2011). The next set of explanations for the existence of underpricing in the competitive underwriter market is based on VC-backing. Gompers & Lerner (2001) claimed VCs are able to mitigate the agency problems arising from information asymmetry, but simultaneously new agency issues arise including multiple actors.
The quid pro quo agreement between the underwriter and VCs is a third explanation for the acceptance of underpricing (Loughran & Ritter, 2001; Lee & Wahal, 2004). In order to gain access to ‘hot IPOs’ in the future, investors are willing to accept higher levels of underpricing. According to Lee & Wahal (2004) a more likely explanation for the acceptance of high underpricing levels by VCs is the grandstanding hypothesis. The grandstanding hypothesis builds on the fact VCs act as agents towards the limited partners, emphasizing the importance of establishing a reputation, taking portfolio companies public and returning money to the original investors, in order to raise capital in the future (Gompers, 1996). Both these possible explanations for the acceptance of underpricing by the issuer involve a conflict of interest between the entrepreneur and the VC, the first having no
13 incentive to accept the underpricing, whereas the second might have different reasons to bear the costs.
Fourth, to the question why some IPO firms are more underpriced than other, Habib & Ljungqvist (2001) suggest besides differences in information asymmetry, the valuation uncertainty and the risk of lawsuits a possible explanation for underpricing differences is found in the degree the issuer cares about this underpricing. Lastly, a more recent explanation for the underwriting puzzle and acceptance of underpricing costs by the issuer is given by the issuer’s willingness to pay for the non-price dimensions. Liu & Ritter (2011) find the level of underpricing to be larger for issuers valuing the non-price dimensions instead of solely focusing on maximizing IPO proceeds. The underlying explanation is the focus of VCs on maximizing the market price at lockup period expiration, at which the invested shares can be sold and returns distributed to the limited partners, via non-price dimensions such as All-star analyst coverage.
The Multiple Agency Framework
One of the possible explanations for the existence of underpricing is discussed in the multiple agency framework. The traditional agency theory examines conflicts of interest between a principal (the owner or shareholder) and an agent (the manager). The multiple agent theory is an extension of this principal agency framework, examining conflicts of interests between different principal-agent groups when at least one agent is connected to a different principal (Arthurs et al., 2008). Arthurs et al. (2008) study the multiple agency theory using entrepreneurial ventures going public and how the multiple agency conflicts affect IPO underpricing. In light of the underwriting process of VC-backed IPO firms one can distinguish between three main actors in a multiple agency relationship: the founder (the entrepreneur), the lead venture capitalist, and the underwriter.
VC-backed venture TVC CVC Corporate parent (Principal) Underwriter (Agent) Limited partners (Principal) In centiv e sc h em e VC-Underwriter ties Figure 1. The Multiple Agency Framework
Entrepreneur (Agent)
14 As shown in figure 1, the entrepreneur and VC syndicate (together the VC-backed venture) act as principal towards the underwriter. Furthermore, the VC-investors act as principal towards the entrepreneur and simultaneously as agent towards either the limited partners (TVC) or the corporate parent (CVC) and are subject to incentive schemes. Not distinguishing between the two different VC types, Arthurs et al. (2008) find the dual identity of the insiders (VC and entrepreneur) in an IPO setting increase the effectivity as monitors, but due to the limited partners, monitoring the VCs, the efficiency is diminished. Also the ties between the VC and underwriter possibly raise new agency conflicts. On the one side, the entrepreneur is concerned about the long-term viability of the venture and the own future employment, whereas the external VC values the long-term VC-underwriter ties more than the short-term relation with the entrepreneur (Arthurs et al., 2008). CVCs differ from purely financial focused TVCs in terms of their incentive schemes (Dushnitsky & Shapira, 2010), and organizational structure (Luukkonen, Deschryvere & Bertoni, 2013). Da Rin, Hellmann & Puri (2011) highlight the importance to distinguish between different VC investors based on their ownership and governance structures and incentive-based compensation, which is founded to be lower for CVC investors. TVCs invest as agents on behalf of the limited partners, whereby achieving financial returns during the lifespan of the fund is the main goal. In order to ensure future financing by limited partners and due to the performance-based compensation, signaling success via exit events of the portfolio company and achieving reputation is important for traditional VC investors – the grandstanding hypothesis (Lee & Wahal, 2004). Furthermore, Chemmanur et al. (2014) suggest CVCs have a longer time-horizon due to the lack of contractually enforced lifespans and TVC investors are more concerned with quick exits compared to CVCs.
The shorter time-horizon and pressure to achieve financial returns due to the organizational structure and incentive schemes suggest the interests of a TVC-investor and the entrepreneur are conflicting. On the one side, TVCs have a reason to accept the underwriter to gain a quick profit and to underprice the venture going public, on the other side, the entrepreneur cares about the long term viability of the venture (Arthurs et al., 2008). CVC-investors on the other hand are affiliates of established corporate firms, leading to an increase in time-horizon, a decreased need for quick exits and most importantly, a lower need to gain a return on investment due to the lack of performance-based compensation. This suggests the interests of a CVC-investor are better aligned to those of the entrepreneur, and CVC-backed issuers are better able to monitor the underwriter, reducing the acceptance level of underpricing.
Arthurs et al. (2008) suggest the long-term VC-underwriter ties are more important than the short-term VC-entrepreneur and underwriter-entrepreneur relation, increasing the willingness to accept
15 higher underpricing levels. The quid pro quo agreements between the underwriter and the investor (Gompers, 1996; Lee & Wahal, 2004) are based on the desire to build an enduring relationship with the underwriters to ensure participation in ‘hot’ IPOs in the future. The importance of the ties to the underwriter most likely primarily exist for TVC-investors, whose sole investment objective is to gain financial profits and return the money to its limited partners, and is lacking for CVC-investors. Due to the longer time-horizon, reducing the pressure to gain financial return during a certain life span, and more strategic investment rationale, the need to enhance quid pro quo agreements is less for corporate investors. Based on the grandstanding hypothesis – emphasizing the importance of signaling success via exit events such as offering the portfolio firm to the public, and the quid pro quo agreement argument – emphasizing the importance of the underwriter-VC ties, corporate investors are hypothesized to be better monitors of the underwriter and thus better able to mitigate the agency costs arising in the multiple agency framework.
On the other hand, the lack of performance-based compensation for the CVC investors suggests the investors are less aggressive (Dushnitsky & Shapira, 2010), reducing the motivation to monitor the other agents and mitigate costs arising from information asymmetry. Also, Hellmann (2002) suggests receiving a corporate VC-investment can be fruitful if the venture’s asset complements the corporate firm’s assets, but when the corporate buisiness is threatened by the venture, strategic CVC-investors only hold equity stake to reduce the other VC-investors to provide support to the venture. This suggests the corporate governance mechanism to be less efficient, leaving the underwriter more opportunity to maximize profit via underpricing. Excessive underpricing due to information asymmetry, and offering newly issued stock to the long-term clients at a discount, is one way for underwriters to favor their long-term clients (Loughran & Ritter, 2003).
Although CVC-investors have less reason to accept excessive underpricing costs due to the lack of need for grandstanding and establishing pro quo agreements, the incentive to monitor the other agents in the IPO process is reduced for corporate investors due to the lack of performance-based compensation and investment rationale. In the multiple agency setting the lowered incentive to monitor is expected to outweigh the ability to better mitigate the agency costs, leading to the first hypothesis formulated as follows:
H1: CVC-backed IPO firms experience higher levels of underpricing than TVC-backed firms
16
Non-price dimensions
Another explanation for the underwriting puzzle is found in the separation between price-dimensions and non-price price-dimensions such as the underwriter rank or All-star analyst coverage (He & Tian, 2013; Liu & Ritter, 2011). According to Liu & Ritter (2011) the existence of excessive underpricing is found in the elasticity of the demand and the market power underwriters are facing. Underwriters employing an All-star analyst or achieve a high underwriter rank have more market power and thus the underwriter market is one of oligopolies. Further findings suggest especially venture capitalists value the non-price dimensions, such as receive All-star analyst coverage, leading to higher underpricing levels for VC-backed IPO firms (Liu & Ritter, 2011).
A recent study by Bradley, Kim & Krignan (2015) find top VC-backed IPO firms are more likely to receive All-star analyst coverage, leading to more underpricing and higher negative return upon lockup expiration. One possible explanation can be found in the fact that underwriters benefit from underpricing and VCs from information momentum – the VC accepts a higher level of underpricing and expect the All-star analyst to add value to the company enabling the VC to cash out at a higher price at lockup expiration (Bradley et al., 2015). Loughran & Ritter (2003) are the first to introduce the analyst lust hypothesis – issuers value analyst coverage and are likely to hire an underwriter with an influential analyst. Whereas previous literature has proven All-star analysts to be advantageous, exhibiting superior performance relative to non-All-star analysts (Bradley, Jordan, & Ritter, 2008), Liu & Ritter (2011) find that IPOs are much more underpriced when they have coverage from An All-star analyst. Based on the study by Liu & Ritter (2011) the second hypothesis state the following:
H2: An VC-backed IPO firm facing All-star analyst coverage is subject to higher underpricing
The analyst lust hypothesis, as proposed by Liu & Ritter (2011), state VCs have a greater lust for All-star analyst coverage because venture capitalists value the price on the day limited partners receive their shares, which is after the lockup period rather than at time of the IPO. In their study on underwriter oligopolies, Liu & Ritter (2011) assume all venture capitalists value the non-price dimensions by the same amount, but due to the different organizational structure and investment rational the lust for All-star analyst coverage is likely to differ between CVC-backed and TVC-backed issuers. Whereas the investment rationale of corporate investors is often strategic, Chesbrough (2002) find CVCs investing mainly for financial reasons aim at gaining a return on investment above that of TVCs due to superior industry experience, the strong balance sheet, and the longer time horizon. Due to the organizational structure, resulting in a longer time horizon, the pressure for a quick exit is lowered (Chemmanur et al., 2014). This, together with the strong balance sheet,
17 enabling the CVC to accept higher underpricing levels to profit from the information momentum (Bradley et al., 2008), suggest the corporate investors are likely to value to the non-price dimensions more. This suggest the lust for All-star analyst coverage is larger for corporate investors relative to traditional investors, leading to the third hypothesis:
H3: CVC-backed IPO firms are more likely to be covered by an All-star analyst than TVC-backed IPO firms
3. Methodology
To investigate the effect of CVC-backing on underpricing this study is split into two parts. Part I tests for a differential effect between CVC-backing – at least one of the investing firms is a corporate venture capital firm, and TVC-backing – none of the investing firms are CVC firms. Making use of propensity score matching (Dehejia & Wahba, 2002; Caliendo & Kopeinig, 2008), the first hypothesis is tested controlling for the endogeneity issue of selection bias among CVC and TVC investors. Part II of this study investigates whether the potential difference in underpricing can be explained by one of the non-price dimensions – All-star analyst coverage.
Part I
In order to receive venture capital funding, contracting takes place between the entrepreneur and the VC covering the allocation of cash flow rights, ownership, and control rights, voting rights or board seats (e.g. Gompers & Lerner, 1996). These contracts are designed to mitigate information and agency problems, suggesting VCs are careful in the selection of portfolio firms. The current study will follow previous studies on corporate venture capital and match CVC-backed companies with comparable TVC-backed companies (e.g. Bruton, Filatotchev, Chahine &, 2010; Lee & Wahal, 2004; Ivanov & Xie, 2010). First the probability of receiving corporate venture capital backing is estimated, where after each CVC-backed company is matched to a TVC-backed counterpart based on this propensity score. A test for the average treatment effect on the treated (ATT) is conducted using the matched pairs, to test for a difference in underpricing level. Following Liu & Ritter (2011), a firm is defined as being CVC-backed if at least one of the investing firms is a corporate affiliate based on private equity information from the Thomson One database.
Ideally the level of underpricing of a CVC-backed firm would be compared to the underpricing level in case the same firm did not receive CVC-backing. In such and experimental setting an OLS regression with a dummy for CVC-backing could be conducted to test for a difference in underpricing, but due to the nonrandom distribution of Venture backing such a regression yield a selectivity bias. Previous studies on venture financing suggest that venture financing is not randomly
18 assigned but the ex-ante analyses conducted by the VC and the design of contracts to mitigate the agency- and information asymmetry problems suggest the receiving of VC-backing is endogenous (e.g. Lee & Wahal, 2004; Chemannur et al., 2014). Chemmanur et al. (2014) suggest CVC-backed and TVC-backed IPO companies differ in age, revenues and profitability for example, suggesting receiving CVC-backing is likely to be endogenous and this selection bias should be controlled for. To account for this bias, instead of an ordinary OLS regression, a matching method is used.
To control for the selection bias firstly the propensity scores of all VC-backed IPO firms, the probability of an entrepreneurial firm receiving CVC-backing, is calculated based on the following control variables significantly differing for both VC-investor types (Lee & Wahal, 2004): a bubble dummy (to control for the large number of IPO firms in the year 1999 and 2000), an internet dummy, the lead underwriter rank (to distinguish between the effect of VC backing and underwriter quality), venture age, a headquarter-state dummies since ventures headquartered in California or Massachusetts might have easier access to corporate venture capital (Lerner 1995 in Ivanov & Xie, 2010), the post IPO book value per share scaled by the offering price, the post IPO total assets per share scaled by the offering price (as proxy for size), the revenues per share scales by the offering price (as proxy for profitability), and a dummy equaling 1 for positive earnings per share in the post IPO fiscal year (as proxy for growth).1 Furthermore, industry and year fixed effects are included.
( | ) ( )
Where Pr is the probability a venture receives CVC-backing based on the vector of the regressors X, is the cumulative distribution function of the standard normal distribution, the vector of the coefficients of X.
Once the propensity scores have been calculated each CVC is matched to a comparable TVC based on propensity score (Ivanov & Xie, 2010), whereby the 2-digit SIC industry is required to be similar to control for differences in operating risk. Using the matched pairs, a test for the difference in the level of underpricing is conducted. To test the effect of CVC-backing on the level of underpricing, the average treatment effect on the treated (ATT) is used, allowing to compare the level of underpricing of a CVC-backed venture to the underpricing level the venture would most likely experience when not receiving the CVC-backing (Caliendo & Kopeinig, 2008).
( | ) [ ( )| ] [ ( )| ]
Where Y is the level of Underpricing, D is a dummy equaling 1 of the venture received CVC-backing, and τATT is
the average treatment effect of the treated.
1
Due to scarcity of financial information on the pre-IPO fiscal year of the IPO firms, all financials from the IPO fiscal year are used.
19
Part II
To test for a potential explanation for the effect found under part I, the next set of tests will investigate whether the effect of CVC-backing on the level of underpricing can be explained by one of the non-price dimensions – All-star analyst coverage. Firstly an OLS regression on underpricing is conducted including a dummy for All-star analyst coverage as variable of interest. Thereafter, a second test is conducted to test whether a CVC-backed IPO firm is equally likely to be covered by an All-star analyst as a TVC-backed company.
The second hypothesis will be tested using an OLS regression on the full sample including a dummy for All-star analyst coverage. The All-star analyst dummy, equaling 1 for a top three analysts defined by the Institutional Investor magazine (Liu & Ritter, 2011), being the independent variable of interest, and underpricing the outcome variable. Control variables included are the logarithm of net proceeds, age, underwriter rank, a dummy indicating an internet company, a dummy indicating a CVC-backed venture, a dummy for the bubble period, a dummy for positive earnings per share, scaled assets-, book value-, and revenue per share, and industry and year fixed effects:
The final hypothesis tests for a difference in All-star analyst lust between CVC-backed US IPO firms and TVC-backed US IPO firms using test for equality of proportions. The null-hypothesis states the proportion of CVC-backed firms receiving All-star analyst coverage equals the proportion TVC-backed firms. Additionally to the test for equal proportions between two investor types, an interaction term for All-star analyst coverage and CVC-backing is included in the regression tests on underpricing. If significant differences are found for the probability of receiving All-star analyst coverage between CVC-backed and TVC-backed IPO firms, tests will be rerun using the propensity score method to control for endogeneity issues.
4. Data & Descriptive Statistics
Sample Selection
Financial data on US IPO firms that went public between 1995 and 2015 is collected from the CRSP/Compustat merged database, and venture capital data is collected from Thomson One’s Private Equity database. Furthermore, due to limited availability in the previous named databases, data on Underwriter rankings and All-star analyst coverage is gathered from Jay R. Ritter’s
20 Warrington website2. Following Liu & Ritter (2011), best efforts offers, ADRs, closed-end funds, REITs, financial institutions, utilities companies, partnerships, reverse LBOs, issues with an offer price below $5, and offers with missing financials for the post-IPO year are excluded. Furthermore, the type and identity of the venture investor should not be missing and the lead underwriter and the underwriter rank should be known. In case of co-managed underwriting processes following Lee & Wahal (2004) the first underwriter mentioned in Thomson Reuters’ database is defined as the lead underwriter.
Data on (Corporate) Venture Capital backing is collected from Thomson One’s Private Equity database. In total 2266 PE-backed companies went public in the US between 1995-2015, of which 1598 were venture capital backed. After correcting for missing underpricing information, a total sample of 1339 VC-backed IPO companies is remained, of which 623 were subject to corporate venturing. For each company information on the industry (SIC code), net proceeds, the number of shares offered, the age of the company, the state the company is headquartered, and the lead underwriter is collected. To control for Underwriter rank, data on the IPO Underwriter Reputation Ranking is used from the earlier paper by Loughran & Ritter (2003).
For the propensity score matching financial data is needed to match each CVC with a TVC based on different matching methods. Financial information on the post-IPO fiscal year is collected from the merged CRSP/Compustat database for all (C)VC-backed IPO companies. Data is collected on total assets, book value per share, earnings per share, revenue per share and total common shares outstanding. The financial data is matched to the VC data based on Ticker symbol.
Following Liu & Ritter (2011), an IPO company is classified as being covered by an All-star analyst if the analyst from the lead underwriter is classified as an All-star analyst (top 3) in the annual ranking of the Institutional Investor (II) magazine’s October issue. Due to limited access to the magazine, the secondary data from Liu & Ritter (2011) is used to determine whether an IPO company has received star analyst coverage in the year following the IPO. Due to the limited availability of data on All-star analyst ranking, the sample for which the analyst lust hypothesis can be tested only covers US IPOs going public between 1993 and 2009. This sample includes 839 IPOs in total of, which 159 are subject to All-star analyst coverage. The sample of CVC-backed IPOs for which All-star analyst coverage is known includes 383 firms, of which only 76 experiences All-star analyst coverage.
21
Descriptive statistics
The number of US venture backed companies is graphed in Figure 2.1. A large peek is shown for the bubble period, followed by a steep decline after the bubble burst. In line with previous literature (e.g. Loughran & Ritter, 2002; Lee & Wahal, 2004) the graph suggests the number of IPOs is cyclical. However, over the twenty years a downward trend is visible in traditional venture backed companies, whereas the number of CVC-backed IPOs seems to be almost equal to the level before the dot.com bubble.
Figure 2.2 shows the difference in average underpricing over the year between CVC- and TVC-backed IPO companies. Again, the bubble period shows a remarkable increase in level of underpricing, and the cyclical character of the number of IPOs seems applicable to the level of underpricing as well. The scatterplot suggests a slightly higher level of underpricing for CVC-backed IPO companies compared to the TVC-backed companies.
Table 1 shows summary statistics for both the TVC-backed sample, as well as for the CVC-backed sample. The table suggests CVC-backed IPO firms are subject to higher underpricing levels and, in line with previous research (e.g. Ivanov & Xie, 2010; Chemmanur et al., 2014), are smaller in terms of total assets per share, have a smaller book value per share, are less profitable, and are less likely to earn positive earnings per share in the post IPO fiscal year. Whereas previous research (e.g. Ivanov & Xie, 2010; Chemmanur et al., 2014) suggest backed firms to be younger, table 1 suggest CVC-backed firms from the current sample to be older when going public. Furthermore, CVC-CVC-backed companies are more likely to be located in California or Massachusetts, and the number of CVC-backed companies was relatively large during the bubble period.
0 50 1 0 0 1995 2000 2005 2010 20151995 2000 2005 2010 2015
TVC-backed IPOs CVC-backed IPOs
N u mb e r o f IPO s Year Graphs by CVC-BACKED
The number of US IPOs between 1995-2015
-2 0 0 0 2 0 0 4 0 0 6 0 0 1995 2000 2005 2010 20151995 2000 2005 2010 2015
TVT-backed IPOs CVC-backed IPOs
U n d e rp rici n g IPO year Graphs by CVC-BACKED
The level of underpricing in the US 1995-2015
Figure 2.2 comparison CVC-backed and TVC-backed IPOs and levels of underpricing
Figure 2.1 comparison CVC-backed and TVC-backed IPOs between 1995-2015
22
Table 1. Characteristics of TVC- and CVC-backed IPO companies
This table compares the firm characteristics of the TVC- and CVC-backed IPO companies going public in the US between 1995-2015. Underpricing is measured as the percent change of the first-day closing price and the offer price, LocationDummy is a dummy variable equaling one if the firm is located in either Massachusetts or California, BubbleDummy is a dummy variable equaling one if the company went public during 1999 or 2000, InternetDummy is a dummy variable equaling one if the company is in the internet business, Age indicated the number of years between the IPO date and the founding date, Proceeds_log indicated the logarithm of net proceeds given in dollar millions, Underwriterrank indicates the rank of the underwriter in the IPO year (Jay R. Ritter’s Warrington website3), EPSDummy is a dummy variable equaling one if the company has positive earnings per share in the post-IPO year, and the assets per share (AT_share), book value per share (BV_share) and revenues per share (REV_share) are given in dollar millions and are scaled by the offer price per share.
Variable CVC-backed TVC-backed Difference p-value
# CVC-backed # TVC-backed N Underpricing 51.46 35.82 15.64*** 0.00 623 716 1339 LocationDummy 0.62 0.50 0.12*** 0.00 623 716 1339 BubbleDummy 0.37 0.25 0.12*** 0.00 623 716 1339 InternetDummy 0.39 0.37 0.02 0.52 623 716 1339 Age 5.41 4.65 0.76*** 0.00 623 714 1337 Proceeds_log 4.20 4.10 0.10* 0.02 623 716 1339 Underwriterrank 7.39 7.45 -0.06 0.73 552 651 1203 AT_share 0.38 0.41 -0.03* 0.04 439 411 850 BV_share 0.28 0.30 -0.02 0.12 439 411 850 REV_share 0.15 0.25 -0.10*** 0.00 439 411 850 EPSDummy 0.22 0.40 -0.18*** 0.00 439 411 850 AllStar 0.24 0.23 0.01 0.72 456 383 839 * p<0.05, ** p<0.01, *** p<0.001
5. Results
This chapter discusses the results of the regressions and appropriate tests conducted to investigate the difference in underpricing level for corporate venture capital backed and traditional venture capital backed US IPO firms.
Testing control variables
After mean-comparison testing for significant differences for all independent variables listed in table1, the earnings per share dummy, age, location dummy, bubble dummy, and the scaled assets per share and revenue per share are indeed significantly different for both investor types. Surprisingly, and contradicting the findings by Ivanov & Xie (2010), no significant difference is found for the underwriter rank, suggesting corporate investors do not value this non-price dimension differently than traditional investors. Also no significant difference is found for the likelihood of the CVC- and TVC-backed firms being an internet company and for the coefficient for the scaled book value per share is not significant, which is possibly due to the high correlation with the scales assets
23 per share (Appendix A). Corporate investors often face a more strategic investment objective of the CVCs, whereas TVCs aim at picking the portfolio companies to maximize the financial return (Chemmanur et al., 2014). In line with Chemmanur et al. (2014) our results from table 1 suggest CVC-backed IPO firms are less likely to have positive earnings per share, revenues are lower, and the IPO firms are smaller in terms of total assets. Whereas Chemmanur et al. (2014) explain the less promising financials of the IPO being subject to corporate venturing by the age at which the issuer takes the venture to the market, results in table 1 suggest CVC-backed firms being older at the event of an IPO. The net proceeds are significantly larger for CVC-backed IPO firms, which is in line with previous studies suggesting CVC-backing leads to higher returns at exit events (e.g. Gompers & Lerner, 2001; Ivanov & Xie, 2010). Furthermore, when testing for equal distribution of the CVC-backed IPOs the significant p-value signals the corporate VC CVC-backed firms are unequally distributed among state, IPO year, and 2-digit SIC industry classification (Appendix C). These findings are in line with Lee & Wahal (2004), who test for equal distribution of VC-backed ventures among state and industry.
Propensity Score Matching
Although the simple mean-comparison test for the variable of interest suggests the level of underpricing is 15.6% higher for corporate venture capital backed IPO companies compared to non CVC-backed companies (Appendix B), no conclusions can be drawn up to this point. The test is subject to a selection bias, due the probability of an entrepreneurial firm receiving CVC-backing being endogenous. A CVC investor might have a bias in picking entrepreneurial firms with certain firm characteristics. To control for this selection bias, propensity score matching is used to rule out the possibility of the differential underpricing levels being explained by the incentive of the CVC to choose firms with specific characteristics (Chemmanur et al., 2014).
The first stage in the propensity score matching is running a regression to determine to probability for each VC-backed IPO firm receiving CVC-backing (Lee & Wahal, 2004). The instrumental variables used to determine the propensity scores are the ones which show significant different means between the two investor types: net proceeds, age, total assets per share, revenue per share, and earnings per share. To control for the unequal distribution of CVC-backed firms between US states, IPO year, a location dummy equaling 1 if the IPO firm is located in either Massachusetts or California and zero otherwise, and a bubble dummy equaling 1 if the firm went public during the bubble period are additionally included as controls. Furthermore, year an industry fixed effects are included to control for year and industry specific heterogeneity (Chemmanur et al., 2014). The number of observations for which financial information from the post-IPO fiscal year is available considerably
24 declines the sample size so two propensity scores are determined. The propensity score for the 1339 US IPO firms from panel A is estimated based on the age, the logarithm of net proceeds, a location dummy and a dummy for the bubble period. Panel B exists of 850 IPO firms and additionally includes a dummy for positive earnings per share and the scaled assets per share, book value per share, and revenue per share as instrumental variables. The table in appendix D shows the probability regressions of the instrumental variables on the treatment, receiving CVC-backing. The predictive power of CVC-backing (pseudo-R2) is 3.7% and 6.2% for panel A and panel B respectively, suggesting panel B is a better predictor for the IPO firm receive CVC-backing based on the firm characteristics, reducing the selection bias. The reduction in selection bias is also shown by the balancing condition – the treatment being independent of the instrumental variables (Appendix E).
Since the variable of interest is the level of underpricing for CVC-backed US IPOs, the second stage regression tests for a significant average treatment effect of receiving CVC-backing, controlling for the selection bias. This method allows comparison of the underpricing level of a CVC-backed IPO firm to that of a similar firm based on specific characteristics without CVC-backing. Using the propensity score matching method, the selection bias is reduced by 15.8% and 19.9% for panel A and panel B respectively (Appendix E).
Following Lee & Wahal (2004), the nearest-neighborhood matching is conducted with replacement, allowing each control item (TVC-backed IPO firm) to be matched to multiple CVC-backed counterparts based on their propensity score. Furthermore, bootstrapped standard errors are used to conduct statistical inference. The first column of table 2 shows the results for the selection-bias adjusted difference in underpricing between the CVC-backed and the TVC-backed US IPOs using nearest-neighborhood matching.
Results from the nearest neighborhood matching for panel A support the first hypotheses, suggesting the level of underpricing is almost 17% higher if an entrepreneurial firm is backed by a corporate venture capital investor (significant at a 1% level). This result is similar to the earlier suggested 15.6% difference from the mean-comparison test between the levels of underpricing for a CVC-backed compared to a TVC-backed firm. The similar results are not surprising since the explanatory power of the instruments used in panel A is small, only 3.7%, suggesting the level of underpricing does not differ due to the ability to obtain higher net proceeds (Ivanov & Xie, 2010) or the choice of CVC firms bringing portfolio firms public at older age. Also, the differential level of underpricing between the VC-investor types cannot be explained by the selection bias based on the IPO firm being headquartered in either California or Massachusetts.
25 Table 2. Matched level of Underpricing
This table shows the output for the average treatment effect of receiving a corporate venture capital investment on the level of underpricing after adjusting for the selection bias. Panel A test the ATT controlling for the age in number of years between the IPO date and the founding date, the logarithm of net proceeds, the firm going public during the bubble period, and a location dummy indicating whether the IPO company is headquartered in either Massachusetts or California. Panel B estimate the ATT additionally controlling for the revenues per share, assets per share, and book value per share (all three scaled by the offer price and given in dollar millions), and a dummy for positive earnings per share. Both regressions include year an industry fixed probability scores. Standard errors are given in parentheses and the stars indicate significance levels.
Panel A - ATT
NN matching Radius matching Kernel matching
16.993** 13.261*** 14.250*** (5.17) (3.74) (4.33) Panel B - ATT 4.577 6.087 6.694 (5.66) (5.47) (4.70) * p<0.05, ** p<0.01, *** p<0.001
Panel B includes additional controls for selection bias based on profitability, size, and growth, leading to a strong decrease in effect of CVC-backing on underpricing. The highly insignificant results for the nearest neighborhood method on panel B show there is not enough evidence for a difference in underpricing level between a CVC-backed IPO firm and a TVC-backed IPO firm. Past research found evidence for CVC investors investing in less profitable and riskier ventures (e.g. Ivanov & Xie, 2010; Chemmanur et al., 2014). The result of the current study suggest the difference in underpricing level between CVC-backed and TVC-backed IPO firms can be explained by the fact that corporate investors have a selection bias and pick portfolio firms with different financial characteristics to invest in. Besides a larger tolerance of failure, a possible explanation for the less promising operating performance of CVC-backed IPO firms given by Chemannur et al. (2014) is the younger age when going public. Results from table 1 show the average age of CVC-backed firms in the current sample is older, violating this explanation and suggesting the difference in operating performance rather being due to different investment objectives. Corporate investors often face a more strategic investment objective compared to the traditional investors, such as nurturing external innovation or increasing the corporate firm value or reputation. When investing for strategic reasons, CVC-investors have a higher tolerance for failure (Chemmanur et al., 2014). Another explanation is given by Hellmann (2002), suggesting corporate investors investing for strategic reasons resume to provide support to the venture and even cannibalize the support provided by other investors if the venture assets form a treat to the corporate business. Assuming the less promising financials to be a proxy for a higher tolerance of failure or the consequence of
26 cannibalization, the findings suggest the differential monitoring behavior between the two VC investor types in the multiple agency framework, leading to a differential underpricing level, is mainly driven by the strategic investment rationale of the corporate.
According to Dehija & Wahba (2002) using different matching methods yield approximate similar results when there is sufficient overlap between the propensity scores of the treatment and the control group. To control for the lack of sufficient overlap, instead of using the one-on-one matching method, the Radius matching method and Kernel matching method are additionally used to test for significant differences in underpricing based on one-to-many matching (table 2, column 2 and 3). The two additional matching methods yield similar results to the nearest neighborhood method, confirming the propensity scores for both CVC-backed firms and TVC-backed counterparts have sufficient overlap. On average the matched difference in underpricing level is 14.8% (Panel A) and 5.8% (Panel B).
The results suggest the incentive to monitor the underwriter is weakened in case of a strategic investment rationale, leading to a weaker corporate governance framework. Controlling for differences in financial characteristics, the underpricing difference disappears, suggesting both investor types have similar incentives to monitor when the portfolio firms are equal in profitability, size and growth perspective. Arthurs et al. (2008) conclude agents with a longer time horizon are better able to monitor other agents in the multiple agency framework. Also, agents with shorter time horizons align their goals better to principals and the agents with the longer time horizon, leading to pre-existing VC-underwriter ties being a possible explanation for IPO underpricing. Whereas Arthurs et al. (2008) suggest the longer time-horizon and organizational structure of CVC-funds are a reason for excessive underpricing, results of this study suggest another explanation is found in the strategic investment rationale influencing the monitoring behavior towards the underwriter.
All-Star analyst coverage & Underpricing
One possible explanation for the difference in underpricing, different than the ability to mitigate the costs arising from the multiple agency framework, is the difference in preference for non-price dimensions between a CVC-backed issuer and a TVC-backed issuer. Firstly, based on Liu & Ritter (2011) one such non-price dimension – All-star analyst coverage, has been tested on the full sample. Next a mean-comparison test is conducted to test for a difference in lust for All-star analyst coverage between the two VC-investor types.
