Essays on the Venture Capital Market A Thesis Submitted to the Faculty of Drexel University by Suting Hong in partial fulfillment of the requirements for the degree of Doctor of Philosophy in June 2014
c Copyright June 2014 Suting Hong. All Rights Reserved.
ii Dedications To my loved grandparents and parents
iii Acknowledgements I would like to thank the members in my committee: Konstantinos Serfes Ph.D. (chair), Mian Dai Ph.D., Mazhar Islam Ph.D., Irina Murtazashvili Ph.D., and Vadake Narayanan Ph.D.. I would not have finished my pursuit for doctorate without their support and encouragement. I would like to give special thanks to Dr. Konstantinos Serfes for his exceptional supervision and advice during my Ph.D. studies. This dissertation has also benefited from comments of seminar participants at Drexel University, Federal Reserve Bank of Philadelphia, and 2013 IIOC meetings, and discussion with Mitchell Berlin, Teresa Harrison, Thomas Hellmann, Sharon Hua, Robert M. Hunt, Samuel Kortum, Tengdong Liu, and Slava Mikhed.
iv Table of Contents LIST OF TABLES... v LIST OF FIGURES... vi ABSTRACT... vii 1. Introduction... 1 2. Literature Review... 8 2.1 The Literature on Entry... 8 2.2 The Literature on Matching... 9 2.3 The Literature on Venture Capital (VC) Market Competition... 10 2.4 The Literature on VC Syndication... 12 3. The VC Market... 13 3.1 History... 13 3.2 Structure... 15 3.3 Data and Data Source... 16 3.4 Patterns of VC Firms Entry and Syndication... 18 3.4.1 Entry... 18 3.4.2 Syndication... 19 4. Competition and Syndication in the VC Market... 23 4.1 Model Set-Up... 23 4.2 Matching in the VC Market... 25 4.2.1 Optimal VC Contracts... 25 4.2.2 Matching Equilibrium... 27 4.2.3 Competitive Effect of Entry... 30 4.3 Syndication of VC Firms... 31 5. Empirical Evidence... 36 5.1 Variables... 36 5.2 The Effect of Entry on Valuation... 38 5.3 The Effect of Entry on Survival... 41 5.4 Endogeneity in Entry and Syndication... 43 5.5 Syndication Between Entrants and Incumbents... 48 5.6 Robustness Check... 49 6. Conclusion... 52 BIBLIOGRAPHY... 54 Appendix A: Proofs... 58 Appendix B: Tables... 71 Appendix C: Figures... 85 Curriculum Vitae... 89
v List of Tables B.1 Entry in the VC Market... 72 B.2 Experience of Incumbents v.s. Entrants in the VC Market... 72 B.3 Syndication in the VC Market... 73 B.4 Syndication and Equity Shares Allocations... 74 B.5 Summary Statistics for VC Financing in Portfolio Companies... 75 B.6 Effect of Entry on Valuation... 76 B.7 Effect of Entry on Valuation (Sample selection corrected)... 77 B.8 Effect of Entry on Survival... 78 B.9 First-Stage Regression... 79 B.10 Effect of Entry on Valuations, Second-Stage Regression... 80 B.11 Effect of Entry on Survival, Second-Stage Regression... 81 B.12 Incumbents Decision to Syndicate with Entrants... 82 B.13 Effect on Pre-money Valuations of Entry of Zero Experience VC Firms... 83 B.14 Effect of Entry of Zero Experience VC Firms (First Stage Results)... 83 B.15 Effect on Survival of Entry of Zero Experience VC Firms... 84
vi List of Figures C.1 Total Number of Deals Involving Entrant VC Firms in the U.S. (Source: VentureXpert)... 86 C.2 Distribution of Experience of Incumbents v.s. Entrants in the VC Market... 86 C.3 Syndication and Equity Shares Received by VC Firms... 87 C.4 One-to-One Financing Patterns... 88 C.5 Syndication between an Incumbent and an Entrant... 88
vii Abstract Essays on the Venture Capital Market Suting Hong Advisor: Konstantinos Serfes, Ph.D. There are two ways for a venture capital (VC) firm to enter a new market: initiate a new deal or form a syndicate with an incumbent. Both types of entry events are extensively observed in the data. This dissertation examines (i) the causes of syndication between entrant and incumbent VC firms, (ii) the impact of entry on VC contract terms and survival rates of VC-backed start-up companies, and (iii) the effect of syndication between entrant and incumbent VC firms on the competition in the VC market and the outcomes of incumbentbacked ventures. This dissertation contributes to the existing literature by developing an analytical framework to examine the strategic interactions between entrant and incumbent VC firms, and to investigate the impact of VC firms entry and syndication on the survival of VC-backed start-ups. The baseline model characterizes the endogenous matching between entrepreneurs and VC firms. In particular, the model captures the following salient features of the VC market: (i) bilateral negotiations between VC firms and entrepreneurs, and (ii) interdependent negotiations of different pairs of VC firms and entrepreneurs. By extending the baseline model to allow for coalition formation among VC firms, this dissertation explores the motive of VC firms to syndicate. Consistent with the empirical observation, the model shows an incumbent VC firm may strategically form syndicates with entrants to maintain its bargaining power. The theoretical analysis provides the following predictions. First, an incumbent VC firm is less likely to syndicate with entrants as the incumbent s expertise increases. Second, entry of VC firms increases the likelihood of survival for incumbent-backed start-up companies, while syndication between entrants and incumbents dampens the competitive effect of entry. Using a data set of VC-backed investments in the U.S. between year 1990 and 2006, this
viii dissertation provides empirical evidence that is consistent with the theoretical predictions. To control for the endogeneity of VC firms entry and syndication, this dissertation follows instrumental variable approach, and the estimation results remain robust.
1 1. Introduction The venture capital (VC) market plays a significant role in financing and nurturing innovative and promising start-up companies. Many highly successful companies received VC funding in the early stages of their development. 1 In 2010, VC-backed companies revenues accounted for 21% of the U.S. GDP and their headcount made up 11% of private sector jobs. 2 Previous study suggests venture capital stimulates the creation of more firms than it funds and positively affects local employment and aggregate income 3. The VC market witnessed substantial growth over the past two decades. In the U.S., VC investment increased more than ten fold from $2.5 billion in 1990 to $28.68 billion in 2011. Furthermore, while only 408 VC firms in the U.S. actively invested in new ventures in 1991, a total of 1,585 VC firms provided start-up financing in 2011, an increase of 288% (VentureXpert). Despite the significance of the VC industry and its rapid growth, little is known about the effect of entry in the VC market. Three empirical regularities about entry and syndication in the VC market motivate this dissertation. First, entrant VC firms are much less experienced than incumbents in a typical local VC market. The median entrant in a given market went through only 10 prior rounds of financing (in other markets), while the median incumbent VC firm experienced 80 financing rounds. One may wonder whether these low-experience entrants have a significant impact on the investment decisions of high-experience incumbent VC firms. Second, along with substantial entry in the VC market, there is extensive syndication between entrants and incumbents. In a median market-year, 57% of the entrants form syndicates with incumbents when investing in a local market for the first time. Syndication may complicate the effect of entry. A natural question is whether syndication softens the competition between entrants and incumbents. Third, VC firms consistently receive more equity shares of start-up companies in syndicate-backed deals than in single-vc-firm-backed deals. In particular, such pattern shows up in VC financing rounds of different investment 1 For example, Facebook, Google, Apple, and FedEx 2 National Venture Capital Association 2012 Yearbook 3 Samila and Sorensen (2011)
2 amounts, and is robust to controlling for the development stages of the VC-backed start-up companies. As a result, one cannot use the existing theories to explain the positive correlation between syndication and VC firm s equity stakes in a company. 4 Instead, a positive correlation between syndication and VC firms equity shares suggests syndication may grant VC firms higher bargaining power against entrepreneurs. In this dissertation, I build a theoretical framework to explore the motivation of VC firms using syndication to increase their bargaining power. This dissertation answers the following important questions regarding entry in the VC market: (i) why do incumbent VC firms syndicate with entrants? (ii) how does entry affect the likelihood of success of incumbent-backed start-up companies? and (iii) what is the effect of syndication among entrant and incumbent VC firms on the investment outcomes of incumbent-backed ventures? I develop a two-sided matching model, which also allows for coalition formation among VC firms to jointly finance a start-up company. I further investigate how entry affects VC contract terms and the likelihood of the success of incumbent-backed start-up companies, while taking into account the syndication between entrants and incumbents. I test the theoretical predictions using a data set extracted from VentureXpert, which covers all U.S. VC investments made between 1990 and 2006. My theoretical model characterizes the endogenous matching between VC firms and entrepreneurs, as well as the syndication of entrants and incumbents. VC firms are heterogenous with respect to their expertise in advising and adding value to start-up companies. Entrepreneurs differ in their business idea quality. VC firms provide capital to wealthconstrained entrepreneurs in exchange for equity shares in the ventures. Furthermore, the model features a moral hazard set-up: By allocating capital and equity share to an entrepreneur, a VC firm has to appropriately motivate an entrepreneur to exert effort. An entrepreneur s effort decides the probability of the success of a start-up company. Consis- 4 Considerable research has studied the motives of VC firms syndication. Some suggest VC firms form syndicates to acquire a second opinion about the quality of entrepreneurial projects (Casamatta and Haritchabalet (2007), Cestone et al. (2006)), but such theory cannot explain why VC syndicates consistently receive more equity shares in financing rounds at various development stages that are associated with different degrees of uncertainty. Brander et al. (2002) and Tian (2011) provide empirical evidence that VC syndication adds values to entrepreneurial ventures. But it remains unclear why VC firms obtain higher equity shares in syndicates than in stand-alone projects.
3 tent with the empirical evidence, I consider entrant VC firms to be less experienced than incumbent VC firms. My model predicts that the less expertise an incumbent VC firm has, the more likely it is to form a syndicate with entrants. This is because a less experienced incumbent faces tougher competition from entrants and develops a stronger incentive to syndicate with entrants to reduce the threat of competition and retain its bargaining power against entrepreneurs. The model also sheds light on the impact entry has on the likelihood of success of an incumbent-backed start-up company. Despite the lower expertise level of entrants, entry of VC firms exerts a positive externality on an incumbent-backed start-up company and leads to a higher success rate of the incumbent-backed venture. This result follows because an entrepreneur receives better contract terms (i.e. more equity/capital) and is better motivated to exert effort upon entry of new VC firms. An incumbent-backed entrepreneur acquires a higher outside value upon entry, and this forces the incumbent VC firm to give more favorable contract terms to the entrepreneur. When an entrant and an incumbent syndicate, however, the competition between entrants and incumbents decreases. An incumbent VC firm may use syndication to maintain its bargaining power and remove the competition effect of entry. In other words, syndication between entrants and incumbents reduces the positive externality of entry. I find empirical evidence consistent with the theoretical predictions. My estimation shows that as an incumbent accumulates more expertise, it is significantly less likely to syndicate with entrants. There is empirical evidence for the positive externality of entry of VC firms: (i) an increase in the entry of VC firms is associated with significantly higher valuations received by an incumbent-backed start-up company and (ii) an increase in the entry of VC firms leads to a higher likelihood of survival of an incumbent-backed start-up company. Furthermore, syndication between entrants and incumbents reduces the positive effect of entry on the likelihood of survival of an incumbent-backed venture, exactly as predicted by the theoretical model that syndication dampens the effect of entry. While the theoretical model does not examine the direct effect of syndication on incumbent-backed
4 start-up companies, 5 the empirical analysis suggests, ceteris paribus, the presence of an entrant VC firm in an incumbent-originated syndicate is associated with a higher likelihood of survival of the start-up company. While these results are highly suggestive, they may be driven by spurious correlations. First, entry of VC firms in a local market may be highly correlated with venture characteristics that are not observable to econometricians. Furthermore, the positive association between entry and VC investment success may be subject to the reverse causality problem, as the presence of promising deals may be likely to attract more VC firms into a new market. In controlling for the endogenous concerns of entry, I construct an instrumental variable using the investment return of limited partners (LPs). Typical LPs include pension funds, university endowments, and insurance companies. In general, LPs keep a balanced investment portfolio and constantly reallocate capital across assets. An increase in LPs return leads to more capital inflows to the VC market and thus a higher level of VC firm entry. At the same time, the LPs investment returns are arguably exogenous to the latent quality of VC-backed ventures. Second, some omitted variables may be correlated with an incumbent s tendency to syndicate with entrants. To address the endogeneity of syndication, I use the number of VC firms that are located within 100 miles of an incumbent as an instrument. Geographic proximity may facilitate the collaboration among VC firms but is arguably exogenous to the investment prospect of an incumbent-backed venture. The estimation results remain robust after I use instruments for entry and syndication: Syndicating with entrants improves the survival rate of an incumbent-backed start-up; entry exerts positive externality on the survival rate of incumbent-backed ventures while syndication with entrants reduces such effect. This dissertation contributes to existing literature in several ways. First, I develop a theoretical model that captures the following salient features of the VC market: (i) bilateral negotiations between VC firms and entrepreneurs, (ii) interdependent negotiations of 5 The theoretical analysis sheds lights on how an incumbent VC firm can use syndication to reduce the degree of competition with entrants, and thus, syndication dampens the positive effect of entry on the survival of the incumbent-backed venture. Such strategic interaction between incumbent and entrant VC firms results in a negative indirect effect of syndication on the venture s survival rate.
5 different pairs of VC firms and entrepreneurs, and (iii) extensive syndication among VC firms. Second, I provide theoretical explanations for the motivation of syndication between entrants and incumbents in the VC market. Third, I examine theoretically and empirically how entry and syndication (between entrant and incumbent VC firms) affect the likelihood of survival of incumbent-backed start-up companies. My results show that entry of VC firms improves the investment outcomes of incumbent-backed ventures while syndication between entrants and incumbents reduces the competitive effect of entry. However, one should not interpret these results as meaning that syndication between entrants and incumbents impairs the efficiency of VC investment. In fact, this dissertation also empirically documents a positive correlation between the presence of entrant syndicate members and the success of an incumbent-backed venture. However, there is no definite conclusion regarding the overall impact of syndication between entrants and incumbents on the success of a given start-up company, as it depends on the relative competitiveness of the incumbent investor in the local market. 6 The remainder of the dissertation proceeds as follows. Chapter 2 reviews the related literature. I begin Chapter 2 with the survey of theoretical literature on entry. There are many theoretical works that examine the effect of entry in various market settings. However, the VC market is different from other oligopolistic markets, and this dissertation provides a theoretical framework to analyze the entry in the VC market or other markets that are characterized by inter-dependent negotiations. Following the review on entry literature, I provide a survey of the matching literature. There are theoretical works examining matching in both transferable and non-transferable utility environments. This dissertation characterizes the matching between VC firms and entrepreneurs in a non-transferable utility environment. I review empirical works on competition in the VC market. Previous literature shows that VC firms compete locally and specialize in a given industry. In addition, there is extensive networking among VC firms, which significantly 6 It can be shown, in a set-up featuring a continuum of incumbents in a market, that as an incumbent gains more expertise and becomes more established in a market, entry of VC firms has a reduced impact on the incumbent-backed ventures (See Hong et al. (2013)). Accordingly, the dampening effect of syndication, if it exists, becomes limited.
6 affects the competitive conducts in the VC industry. Another related stream of literature examines the syndication of VC firms. Previous literature discusses multiple motivations for VC firms to form syndicates and empirically documents that syndication creates values for entrepreneurial firms. In contrast to previous literature, this dissertation explores another cause of VC syndication: strategic cooperation. As VC firms band together, they may possess greater bargaining power when negotiating with entrepreneurs. In Chapter 3, I review the VC market history, structure, and data. I examine the two prominent phenomena prevailing in the US VC market: entry and syndication. The US VC market experienced substantial entry in recent two decades. One may expect entry of VC firms to increase the competition in the VC market. However, along with entry of VC firms, there is extensive syndication between entrant and incumbent VC firms. Therefore, in understanding the impact of VC firms entry, one needs to take into account the coordination among entrants and incumbents through syndication. Furthermore, I show empirical suggestive evidence that forming syndicates with entrants gives incumbents more bargaining power in the negotiation with entrepreneurs. Such observation suggests the imperative to build theoretical model explaining the strategic motivation of incumbent VC firms syndicating with entrants. In Chapter 4, I build a theoretical framework for analyzing the effect of entry and syndication in the VC market. I begin by focusing on the effect of entry without considering VC firms syndication. I develop an endogenous matching model for the VC market consisting of heterogeneous VC firms and entrepreneurs. The VC contracting features a moral hazard setting in which entrepreneurs exert unobservable effort, and VC firms have to provide sufficient incentives. I establish a competitive equilibrium for the matching in the VC market. The main conclusion is that competition of VC firms improves the outside option for entrepreneurs and is associated with better contract terms to entrepreneurs. In Section 4.3, I extend the baseline model and examine the incentives of VC firms to syndicate. Incumbent VC firms may use syndication to dampen the competitive effect of entry of VC firms. However, as an incumbent VC firm gains experience, it is less likely to syndicate with
7 entrants. In Chapter 5, I test the theoretical predictions against the data. I find supportive evidence that entry of VC firms intensifies competition in the market, and thus, benefits entrepreneurs. In the meantime, when an incumbent VC firm syndicates with entrants in a funding round, entry has a reduced effect on the contract terms received by the entrepreneur, and thus, the competition effect of entry on the venture s survival declines. The empirical results are robust to controlling for the endogeneity of entry and syndication. I also present results of robustness checks. Chapter 6 summarizes the key insights and concludes.
8 2. Literature Review This dissertation builds on four streams of literature: the industrial organization literature on entry, the literature on matching, the literature on VC market competition, and the literature on VC syndication. 2.1 The Literature on Entry There is a long-standing interest in examining the relationship between entry and market efficiency. There are many theoretical works providing explanation about the effect of entry on prices; see e.g., Cournot, Bertrand and Hotelling-types of models, monopolistic competition models as in à la Dixit and Stiglitz (1977), and discrete choice models as in Perloff and Salop (1985). Shaked and Sutton (1982) demonstrate that free entry does not dissipate profits in models of vertical differentiation due to the finiteness property. In these models there can be at most a limited number of firms with positive market shares in the industry. The competition among the high-quality firms drives down prices to a level at which there is no room for low-quality products. The finiteness property holds regardless of the entry costs, and is conditional on the feature that the marginal cost of improving quality does not increase too quickly with quality. Therefore, the finiteness property is more likely to hold, if the cost of quality improvement is fixed rather than variable. In these model high-quality incumbents are therefore shielded from competition from lower-quality entrants. Some oligopoly models predict excessive market entry because of the business stealing effect. For example, Mankiw and Whinston (1986) show that free entry can lead to socially excessive investment, if firms produce homogeneous goods. Under free entry, the business stealing effect of entry dominates the gains to consumer surplus. However, entry in the VC market is distinct from entry in other product/service markets. The VC market is characterized by bilateral negotiations between VC firms and en-
9 trepreneurs, and there is extensive syndication among VC firms. The analysis from the existing entry literature cannot be readily applied in the context of the VC market. To examine entry in the VC market, I develop a two-sided matching model, which also takes into account coalition formation among VC firms. 2.2 The Literature on Matching Studying the matching between entrepreneurs and VC firms follows the line of research dating back to Shapley and Shubik (1972) and Becker (1973). In a Becker-type model, utility is transferable among partners, and the complementarity of partners types guarantees monotone matching. Legros and Newman (2007) suggest that, in a non-transferable utility environment, a second condition is needed: the complementarity between an agent s type and his partner s payoff. This second complementarity implies that the degree of transferability is monotone in types. Terviö (2008) devises an assignment model between heterogenous CEOs and firms to explain the diversity of CEO compensations. 1 Both the CEOs and firms are of heterogenous qualities and jointly make contribution to the economic surplus of a firm. The production function is assumed to be strictly increasing in both CEO s ability and firm s size. This supermodularity condition leads to a positive assortative matching (meaning the matching of the most capable managers and the largest firms). In equilibrium the pay levels of CEOs depend on the distribution of firm size and CEO ability in the economy. The model is used to study the effects of CEO ability on profits and CEO pay. The model developed in this dissertation shares a number of features with Terviö (2008), most notably two-sided matching between heterogenous principals and agents. However, this dissertation differs in the following aspects: first, the model features the moral hazard set-up in VC contracts; second, utility cannot be perfectly transferred between the two sides of a VC contract, which is consistent with the common observation of limited liability of entrepreneurs. In this set-up, I am able to analyze the effect of VC firms competition on 1 For other matching models in principal-agent settings, see e.g., Besley and Ghatak (2005), Serfes (2005, 2008), and Dam and Pérez-Castrillo (2006).
10 the contract terms received by entrepreneurs. Dam (2007) analyzes a model of two-sided matching and incentive contracts between VC firms and entrepreneurial firms. VC firms are of heterogeneous monitoring capacities, and are matched with firms with different levels of initial wealth. In equilibrium, VCs with higher monitoring ability invest in firms with lower initial wealth, and thus the matching pattern is negatively assortative. Different from Dam s set-up, my model features the endogenous matching between VC firms of heterogeneous expertise and entrepreneurs of different quality business ideas. Furthermore, in my model, the equilibrium matching between VC firms and entrepreneurs are shown as positive assortative. My model sheds lights on how competition among heterogenous VC firms affect the bilateral negotiations between VC firms and entrepreneurs. Sorensen (2007) examines the matching between heterogenous VC firms and entrepreneurs to separate selection from treatment effect in VC markets. Data show that experienced investors are more likely to go public. This could be attributed to two effects: influence and sorting. Influence effect refers to the value added by experienced VC firms, while sorting effect means experienced VCs are matched to entrepreneurs with better qualifications. Sorensen (2007) empirically separates the two effects by structural approach and employing Bayesian techniques. 2.3 The Literature on Venture Capital (VC) Market Competition This paper examines the competition between incumbents and entrants in the VC market and thus is related to the literature on VC market competition. Hochberg et al. (2010) find that networking among incumbent VC firms helps to deter entry by other firms, and a VC firm has an increased likelihood of entering a new market if it has cooperated with the incumbents before. In addition, a densely networked VC market would lead to lower valuations for entrepreneurial firms. Using a market structural model approach, Hochberg et al. (2011) examine the competition among heterogeneous VC firms, with respect to industry specialization, and they
11 suggest that the presence of strong cooperative ties between VC firms dampens the competitive effects of entry in a local VC market. My paper differs in the following three aspects: (i) I develop a theoretical framework and provide empirical evidence in examining the impact of entry on the contract terms and success rates of VC-backed deals; (ii) I explicitly investigate the response of incumbents to entry of new firms; and (iii) I analyze the interaction effect of entry and syndication on start-ups financed by incumbents. By showing that syndication with entrants dampens the positive externality of entry, I provide supportive evidence to the findings in Hochberg et al. (2011). Gompers and Lerner (2008) examine how inflows to venture capital funds affect the valuation received by entrepreneurs. Growth in venture capital commitments increases the valuations of new investments. The findings are consistent with suggestions that competition for a limited number of good investments may be responsible for rising prices. In this dissertation, I push the line of inquiry forward by working on the following aspects: first, I build an endogenous matching model that enables me to examine the mechanism through which competition affects VC contract terms; second, I take into account the possibility that VC firms can strategically form syndicates to reduce the degree of competition in the VC market; third, I examine how VC firms expertise levels affect the motivation of VC firms using syndication to dampen competition. Inderst and Mueller (2004) and Silviera and Wright (2006) develop search models for the VC market. The search equilibria in their models are characterized by the ratio of the number of VC firms to the number of entrepreneurs in the market, commonly referred to as market thickness in the search theory literature. Using a matching framework, however, this dissertation shows that it is not just the market thickness that matters; the qualities of entrepreneurial ideas and the investment expertise of VC firms are also critical for the optimal design of VC contracts, especially because these determine the outside options of entrepreneurs and the incentives of VC firms to form syndicates. Furthermore, Inderst and Mueller (2004) consider homogenous VC firms and entrepreneurs, while Silviera and Wright (2006) introduce ex-post heterogeneity, i.e., the heterogeneity arises after a VC
12 firm is matched with an entrepreneur. In contrast, we explicitly account for different match qualities by allowing entrepreneurs to be ex-ante heterogenous with respect to their business ideas, and VC firms with respect to their expertise. 2.4 The Literature on VC Syndication This paper investigates the motivation of an incumbent to syndicate with entrants. Considerable research has explored the motives of VC firms to form syndicates and the impact of syndication on start-up companies. Casamatta and Haritchabalet (2007) and Cestone et al. (2006) theoretically propose that VC firms syndicate in order to acquire a second assessment on a project. Brander et al. (2002) test for two causes of VC syndication: the second opinion and the value-added hypothesis. According to the value-added hypothesis, lead VC firms form syndicates to tap the expertise of additional VC firms, and this would add value to the ventures. Brander et al. (2002) find supportive evidence for the value-added hypothesis. Tian (2012) shows that VC syndication nurtures the innovation of portfolio companies and is associated with a higher likelihood of success as well as a better post-ipo operating performance. Du (2009) examines VC firms preferences for syndication partners and shows that VC firms are less likely to syndicate with partners that are different from them. Hochberg et al. (2011) examine the relationship between VC firms tie-formation and four types of VC firm resources: experience, network access, available capital, and investment scope. They find that VC firms tend to form ties with the best available partner in terms of investment scope and network access, but they form ties with partners with dissimilar levels of experience. VC firms also build ties to trade capital for other value-added resources. In contrast to previous literature, my paper explores another cause of VC syndication: strategic cooperation. As VC firms band together, they may possess greater bargaining power when negotiating with entrepreneurs. Specifically, my analysis suggests that an incumbent can benefit from syndicating with entrants since syndication reduces the competitive threat of entry of new firms.
13 3. The VC Market 3.1 History The first true VC firm was American Research and Development Corporation (ARDC), established in 1946 by MIT professor Karl Compton, Harvard Business School professor General Georges F. Doriot, and Ralph Flanders. ARDC made investment businesses run by soldiers who were returning from World War II. The success of ARDC inspired the launch of other venture organizations. In 1978, $424 million was invested in new VC funds, among which 32% share was contributed by individuals and only 15% was supplied by institutional investors. It was not until 1978 that VC experienced its first major fund raising year. With the passage of the Employee Retirement Income Security Act (ERISA) in 1974, corporate pension funds were prohibited from holding certain risky investments including many investments in privately held companies. In 1978, the US Labor Department relaxed certain of the ERISA restrictions, under the prudent man rule, thus explicitly allowing pension managers to invest in high-risk assets, including VC. The annual flow of money into venture funds reached approximately $750 million in 1979 1. 1980s was trying times for VC industry. Many VC-backed companies received great success 2 and this led to major proliferation of venture capital investment firms. The number of VC firms dramatically increased, and the capital managed by these firms increased from $3 billion to $31 billion over the course of the decade 3. However, VC fund return became relatively low by the end of the 1980s. This was due to the increased competition for promising projects, lack of experience of newly-established VC firms, and the crash of the stock market in 1987. The VC industry experienced limited growth throughout the 1980s and the first half of the 1990s. From the mid-1990s to 2000, VC fundraising soared in the U.S.. This was in response 1 National Venture Capital Association 2000 Yearbook 2 For example, Apple Computer, Genentech, Digital Equipment, etc. 3 Statistics of fundraising and investment activities in the VC industry are released in National Venture Capital Association Yearbook.
14 to the success of many VC-backed technology start-up companies in late 1990s. As inexperienced VC firms exited the market and initial public offerings of stock for technology companies rapidly exploded, many VC firms witnessed large returns. Capital commitment to VC funds continued to increase since 1995 and spiked $101 billion in 2000. The number of VC firms was 1,109 in 2000, a two-fold increase since 1995. However, the stock market crash in 2000 shook the entire VC market. Many VC firms were forced to write off the portfolio companies or exit the market in the early-2000s. The number of actively investing VC firms dropped from 1,053 in 2000 to 505 in 2003. From 2003 through 2007, the fundraising and investment activities in the VC market remained stable. Measuring industry activity with the total dollars invested in a given year shows that the industry has remained generally in the $20 billion to $30 billion range since 2002. In 2012, $26.7 billion was invested in 3,143 companies. The outbreak of the 2007-2008 financial crisis caused increased difficulty for VC firms to raise money. In 2012, commitments totaling $20.1 billion were made to 183 funds. This is roughly two-thirds of the annual levels seen in 2005-2007, and approximately one-fifth of the annual amount raised at the bubble peak. This is the sixth consecutive year in which more money was invested by the industry than raised in new commitments. VC firms were not able to distribute sufficient payoffs to the investors and it has been very difficult to raise new funds. VC firms invest extensively in high-technology companies. The leading VC investment sector is software, which accounted for 31% of the total VC investment amount in 2012, followed by Biotechnology 15%, Industrial/Energy 10%, and Medical Devices and Equipment 9%. Since the 1960s, it has been a feature of the VC industry that investments concentrated in California and Massachusetts. In 2012, California companies received 53% of the total VC investment amount and Massachusetts-based companies received 12%. Over the past two decades, VC firms have gradually shifted attention to the two ends of the spectrum of development stages. 51.8% of total deals in 2012 were seed- and early-stage deals, the highest percentage since 1985. In the meantime, 22.4% of deals went to later-stage
15 companies in 2012, which is also toward the top end of the historical range. The rest of the deals were made in the expansion-stage companies. 3.2 Structure Most VC funds take the organizational form of limited partnership. Limited partners (LPs) supply capital to the funds and have limited liability for the investment. Typical LPs include wealthy individuals and institutional investors, such as pension funds, insurance companies, and university endowment. The general partners (GPs), also known as venture capitalists, actively manage the fund identifying promising investments and then advising the portfolio companies that they choose to invest in. The lifespan of a fund is typically ten years. The disbursement for GPs generally consists of two parts: annual management fees (with a classic figure of 2%) and carry which occurs in the event of a successful exit (i.e. IPO or M&A deals). Carry aligns the interests of LPs and GPs by ensuring that both parties share in the gains. VC firms are actively involved in the management of portfolio companies. The GPs help set strategies, develop market, and build networks for a portfolio company, which leads to value creation. In most cases, GPs sit in the board of a portfolio company and play important role in the governance decision. Previous literature finds evidence for the value-added role of VC firms. Hellmann and Puri (2000) find VC funding significantly reduces the time for a portfolio company to bring its product to market. Hellmann and Puri (2002) find that VC funding leads to professionalization of portfolio companies: VC funding is associated with faster hiring of vice presidents of marketing & sales, as well as faster replacement of founder with professional CEO. Most of the VC deals are carried out in multiple rounds of financing. In some cases, subsequent funding is possible only when a company reaches certain performance milestones. A large body of literature studies the incentives and consequences of staging VC investments. Kaplan and Stromberg (2003) provide empirical evidence that staging finance increases the ability of VC firms to liquidate the venture if performance is unsatisfactory. In a world
16 of incomplete contracting, Hart and Moore (1994) suggest that original founders skills can be critical to the development of ventures, and an entrepreneur cannot contractually commit to staying with a firm, which creates hold-up threats for investors. Neher (1997) suggests that staged financing helps mitigate this commitment problem, as the gradual embodiment of the entrepreneurs human capital in the ventures physical capital is equivalent to building up collateral. This reduces an entrepreneur s incentive to leave the venture in early rounds. However, Tian (2010) finds evidence inconsistent with the hold-up hypothesis of implementing staged financing, but provides supportive evidence that staging of capital infusions is a substitute for intensive monitoring of entrepreneurs by VC investors. Once successful portfolio companies mature, venture funds generally exit their positions in those companies by taking them public through an initial public offering (IPO) or by selling them to another company (merger & acquisition). After a successful exit, GPs can distribute the proceeds to the LPs. In 2012, the number of venture-backed companies going public was 49 and there were 449 acquisitions. 3.3 Data and Data Source I extract the VC investment data from the VentureXpert database provided by Thomson Financial. This comprehensive database has been extensively used in VC research (see, e.g., Kaplan and Schoar (2005), Srensen (2007), and Hochberg et al. (2010)). VentureXpert provides detailed information on firms that have received venture capital financing. The database includes information on the dates and investment amounts of different financing rounds, the identities of investing VC firms, the development stage and industry groups of portfolio companies, and the dates and types of exit (e.g., IPO, acquisition, or liquidation). The whole sample contains all venture capital investments made in U.S. companies from 1975 to 2006, with a total of 91, 762 funding rounds in 31, 943 portfolio companies by 6, 275 VC firms. It is well known that VC firms specialize in specific industries and tend to invest in local start-up companies (e.g., Hochberg et al. (2010)). I therefore define VC markets as follows:
17 First, I differentiate among the six main industry groups in the VentureXpert database. These include Communications and Media, Computer Related, Semiconductors and Other Electronics, Biotechnology, Medical, Health and Life Sciences, and Non-High- Technology. Second, for each industry, I group all companies located in the same state. For example, Computer Related in California is a different market from Biotechnology in Massachusetts. To improve the explanatory power of my regression analysis, I exclude all observations for inactive market periods. This concerns markets with either fewer than five deals in the current year, or fewer than 25 deals in the past five years. This results in a total of 1, 378 market observations (over the span of 17 years). The primary sample includes all VC investments made between 1990 and 2006, in companies that received their first round of VC funding beginning in 1990. Furthermore, I exclude observations of investments in mature companies that are at the buyout stage. I also exclude investments by corporate VC firms, which usually make strategic investments and generally have different objectives than traditional VC firms. VC syndication entails joint investment by two or more VC firms in an entrepreneurial firm. Empirically, I define the VC syndicate as a group of VC firms sharing a given round of financing in an entrepreneurial firm. A syndicate is lead managed by a single VC firm that usually originates the deal and acts as the most active investor in advising and professionalizing the company. Following the literature, I identify the lead VC firm as the investor that participated the earliest and made the highest investment in the portfolio company. To examine the effect of entry on incumbent-backed projects, I consider VC funding rounds that are lead-managed by incumbents and exclude observations of financing rounds with entrants as lead investors. In controlling for the endogeneity issue of entry, I explore the shock from the VC funding supply and construct an instrumental variable that reflects limited partners investment returns. I extract the university endowment returns from the annual surveys conducted by the National Association of College and University Business Officers (NACUBO). Membership organizations of NACUBO represent more than 2,500 colleges, universities, and higher edu-
18 cation service providers. NACUBO conducts voluntary surveys of member schools regarding endowment returns and investment characteristics. Such an annual survey provides comprehensive information on the university endowment and has been used in previous research (Lerner et al. (2008)) I use the data from 1987 to 2006 in constructing the instrumental variable. In addition, I track VC investments made by U.S. LPs using the VentureXpert LP data. 3.4 Patterns of VC Firms Entry and Syndication 3.4.1 Entry I define an entrant as a VC firm that invests in a given market for the first time. Notice that an entrant may have accumulated experience investing in other industry/geographic locations before entering the new market of interest. In measuring the level of entry in the VC market, following Hochberg et al. (2010), I construct the following variables: (i) the number/fraction of deals entrants are involved in, (ii) the number/fraction of deals led by entrants, (iii) the total number of entrants in the market, and (iv) the total number of entrants leading a deal. Table B.1 reports the descriptive statistics of the entry measures. In a median market year, six (40%) deals involve entrants while four (25%) deals have an entrant as the leader; four of the eight entrants lead at least one deal. Figure C.1 illustrates the number of deals for each industry group in which at least one entrant was involved. The different industries have witnessed a similar trend in the entry of new VC firms: Entry peaked in 2000 and remained relatively stable after a sharp decline in the years 2001 and 2002. figure C.1 It is well documented that VC firms differ with respect to their expertise and investment experience (e.g., Sorensen (2007)). Following Nahata (2008), I construct three reputation measures for each VC firm at each financing round date. The first reputation measure is based on the cumulative market capitalization of IPOs backed by the VC firm in the
19 IPO market. For each VC firm, I cumulate the dollar market value of all companies taken public by the VC firm since year 1987 until a given calendar year and then normalize it by the total market value of all VC-backed IPO companies within the same time period 4. The second reputation measure is the cumulative dollar investment by a VC firm since 1980 until a given calendar year, normalized by the overall aggregate investment in the VC industry in those years. The third measure of the VC firm s expertise is the number of prior financing rounds since 1975. Table B.2 reports the statistics of the three expertise measures for entrant and incumbent VC firms, respectively. The median entrant went through 10 financing rounds in other markets, garnered 0.005% share of the aggregate VC industry investment, and experienced zero IPO exit prior to its entry into the new market. The incumbent VC firms are better established than the entrants with a median IPO share of 0.33%, VC investment share of 0.059%, and experience going through 80 rounds of financing. Interestingly, most entrants already have some prior investment experience when entering a new market; however, their experience tends to be substantially less than the investment experience of incumbent VC firms. Figure C.2 shows the distribution of entrants experience and incumbents experience using kernel density estimation based on the financing round experience measure. The distribution of entrants experience is more skewed to the right than the distribution of incumbents experience. 3.4.2 Syndication Cooperation among VC investors is an enduring and striking feature of the VC industry. Of the 45, 926 VC funding rounds between year 1990 and 2006 covered by the sample 5, approximately 51% are backed by more than one VC firm. Table B.3 reports the summary statistics of market level syndication patterns and further differentiates among syndicates consisting of both incumbent and entrant VC firms, incumbents only, or entrants only. In a median market year, 44% of the VC funding rounds receive money from syndicates; 4 The market value of a VC-backed IPO company is based on the initial-day closing price and shares outstanding on that date. The Center for Research in Security Prices (CRSP) is the source of the information. 5 The sample does not include VC funding rounds at the buyout stage or observations of inactive markets with less than 5 VC funding rounds in a given year or less than 25 funding rounds in the past five years.
20 about 45% of syndicates consist of both incumbent and entrant VC firms, 43% syndicates of incumbent VC firms only, and 0% syndicates of entrant VC firms only. There is an extensive amount of syndication between entrants and incumbents. As shown in Table B.3, in a median market year, 57% of entrants syndicate with incumbents, and 57% of incumbents form syndicates with entrants. In particular, the syndication between entrants and incumbents has an asymmetric pattern. In a median market year, 33% of the entrants join an incumbent-led syndicate; in the meantime, only about 9% of incumbents join an entrant-led syndicate. It is more frequent for an entrant to take part in a deal initiated by incumbents rather than the other way around. This also suggests the significance of understanding the entry through syndication in the VC market. The data suggest VC firms receive higher equity shares of start-up companies in syndicates than in stand-alone projects. Furthermore, syndicates consisting of both entrant and incumbent VC firms obtain higher equity shares than incumbent- or entrant-only syndicates. Due to the lack of contract details in the VentureXpert data, I use the following ratio to quantify the negotiation results between entrepreneurs and VC firms in a given funding round: R I-to-Premoney =(Current Round Investment Amounts)/(Pre-money Valuations+ Current Round Investment Amounts). This constructed variable is positively correlated with the equity shares received by current round participating investors in the venture 6. I compare R I-to-Premoney for three types of VC-funding deals: (i) funding rounds backed by a single VC firm, (ii) by a syndicate consisting of incumbent- or entrant-only, and (iii) by a syndicate consisting of both entrant and incumbent VC firms. Many factors may influence the contracting between VC firms and entrepreneurs, including the investment amounts and the development stages of the company. Therefore, I examine the patterns of R I-to-Premoney not just by syndicate types, but also by investment sizes and development stages. I set up nine bins of investment amounts of a funding round, ranging from investment sizes of smaller than $0.5 million to above $20 millions. Early stage companies are in the development stage of Seed/Start-up or Early Stage, while later 6 R I-to-Premoney does not reflect the overall equity shares owned by VC firms that invested in the start-up company in a previous funding round. Nevertheless, this measure serves well our purpose of quantifying the newly acquired equity shares by VC firms in the current funding round.