Venture Capital: The Impact of Asymmetric Information on Optimal Investments, Learning and Exit Outcomes [1 ed.] 9783896447531, 9783896737533


119 100 12MB

German Pages 152 [151] Year 2019

Report DMCA / Copyright

DOWNLOAD PDF FILE

Recommend Papers

Venture Capital: The Impact of Asymmetric Information on Optimal Investments, Learning and Exit Outcomes [1 ed.]
 9783896447531, 9783896737533

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

Venture Capital: The Impact of Asymmetrie Information on Optimal Investments, Learning and Exit Outcomes

Studienreihe der Stiftung Kreditwirtschaft an der Universität Hohenheim Herausgeber: Prof. Dr. Hans-Peter Burghof

Band 57

Julius Tennert

Venture Capital The Impact of Asymmetric Information on Optimal Investments, Learning and Exit Outcomes

Verlag Wissenschaft & Praxis

JBJ

Bibliografische Information der Deutschen Nationalbibliothek Die Deutsche Nationalbibliothek verzeichnet diese Publikation in der Deutschen Nationalbibliografie; detaillierte bibliografische Daten sind im Internet über http://dnb.d-nb.de abrufbar.

Dl 00 ISBN 978-3-89673-753-3 © Verlag Wissenschaft & Praxis Dr. Brauner GmbH 2019 D-75447 Sternenfels, Nußbaumweg 6 Tel. +49 7045 930093 Fax +49 7045 930094 [email protected] www.verlagwp.de

Alle Rechte vorbehalten Das Werk einschließlich aller seiner Teile ist urheberrechtlich geschützt. Jede Verwertung außerhalb der engen Grenzen des Urheberrechtsgesetzes ist ohne Zustimmung des Verlages unzulässig und strafbar. Das gilt insbesondere für Vervielfältigungen, Übersetzungen, Mikroverfilmungen und die Einspeicherung und Verarbeitung in elektronischen Systemen. Druck und Bindung: Esser printSolutions GmbH, Bretten

5

Preface The European venture capital market experienced significant growth in the recent years. This has a variety of reasons. On the one hand, investors increased their demand for risky assets as expected returns in traditional asset classes were strongly impacted by the expansive monetary policy of the ECB. On the other hand, European founders were inspired by the success stories drawn from US start-ups in the Silicon Valley, reaching for more risk capital in Europe. Despite its strong growth, the European venture capital market is rather neglected in scientific research. However, for European politicians and decision makers in the industry it is more important than ever to better understand how an efficient funding process for business innovations looks like and what incremental value venture capitalists can add to this process. Only with an in-depth understanding of venture capitalists' business model, we can take the right steps to keep innovative European entrepreneurs in Europe or even attract the most talented entrepreneurs from all over the world to set-up their business idea in Europe. This way, we can ensure that Europe remains one of the most successful regions in the world. This book addresses the fundamental question of how international venture capitalists do their business with European start-ups and whether success factors significantly differ between Europe and the world's largest venture capital market in the US. The results have impact for both, politicians and decision makers in the industry as they give advanced insight how to increase effectiveness of the funding and exit process. In the long-term,

β

this can make risk capital less expensive, and innovations and economic growth faster. Hohenheim, December 2018 Prof. Dr. Hans-Peter Burghof

7

Contents List of Figures

8

List of Tables

9

List of Abbreviations

13

1 Introduction

15

2 Moral Hazard in V C Finance 2.1 Introduction 2.2 Formal Model 2.2.1 The Entrepreneurial Project 2.2.2 Maximization Function of the VC 2.2.3 Participation Constraint of the Entrepreneur . . . . 2.2.4 Solution without Relevance of Human C a p i t a l . . . . 2.2.5 Solution with Relevance of Human Capital 2.2.6 Opportunity Loss of the VC 2.3 Empirical Approach 2.3.1 Sample 2.3.2 Econometric Method 2.3.3 Measures 2.3.4 Empirical Results 2.3.5 Robustness Checks 2.4 Conclusion

21 21 24 24 26 26 28 29 29 34 34 34 36 39 43 49

3 Learning in Funding Relationships 3.1 Introduction 3.2 Learning and Comprehensiveness 3.3 Frictions in the Learning Process

53 53 56 58

8

Contents

3.4 Formal Hypotheses 3.5 Data and Methodology 3.5.1 Sample 3.5.2 Measures 3.5.3 Econometric Specification 3.6 Results 3.6.1 Summary Statistics and Selection Regressions . . . . 3.7 Conclusion

60 63 63 64 74 77 77 86

4 VCs as Intermediaries in Exits 4.1 Introduction 4.2 Agency Cost in IPO and Trade Sale Exits 4.2.1 Adverse Selection Pre-Exit 4.2.2 Moral Hazard Post-Exit 4.3 Investor Characteristics 4.4 Firm Characteristics 4.5 Sample and Methodology 4.5.1 Sample 4.5.2 Methodology 4.5.3 Control Variables 4.6 Results 4.6.1 Descriptive Statistics 4.6.2 Multivariate Analysis 4.6.3 Classification Tree Analysis 4.7 Conclusion

91 91 94 94 95 97 100 101 101 102 103 105 105 Ill 116 128

5 Conclusion 5.1 Summary of the Main Results 5.2 Outlook

131 131 132

Bibliography

134

9

List of Figures 2.1 Optimal Timing of Investments 2.2 Opportunity Loss of the VC from Optimal Timing of Investments 3.1 Impact of Informational Updates on the Firm Value, Comprehensive and Uncomprehensive Information 3.2 Impact of Informational Updates on the Firm Value, Symmetric and Asymmetric Information 3.3 Average and Aggregate Investments to Projects 3.4 Average Pre-Money Valuations of Entrepreneurial Firms . . 4.1 4.2 4.3 4.4 4.5 4.6

CART: CART: CART: CART: CART: CART:

European Sample Low IPO Rates, European Sample High IPO Rates, European Sample U.S. Sample Low IPO Rates, U.S. Sample High IPO Rates, U.S. Sample

32 33 62 64 79 80 122 123 124 125 126 127

11

List of Tables 2.1 2.2 2.3 2.4

Descriptive Statistics and Correlation Matrix Variable Break-Points Accelerated Failure Time Model: Funding Duration . . . . Accelerated Failure Time Model: Funding Duration, Early and Later Stages Model Fit

41 42 44

3.1 Classification of Financing Rounds I 3.2 Classification of Financing Rounds I I 3.3 Probit Regression (1) and Cox Proportional Hazard Rate Modeling (2): Selection Equations 3.4 GLS Regression: Pre-Money Valuation

67 70

2.5

4.1 4.2 4.3 4.4

Summary Statistics, Summary Statistics, Logistic Regression: Logistic Regression:

European Sample U.S. Sample European Sample U.S. Sample

46 51

83 87 107 109 117 118

13

List of A b b r e v i a t i o n s AIC CART CO CVC DCF E. G. EXP GAMMA GICS GLS IPO 1MB, IRR LLOG LNORM LOG LO G2 M&A MIO MSCI R&D R, 2 RESP. U.K U.S.

Akaike Information Criterion Classification and Regression Trees Contai and O'Quigley (1999) Break-Point Value Corporate Venture Capitalist Discounted Cash Flow Example Given Exponential Distribution Function General-Gamma Distribution Function Global Industry Classification Standard Generalized Least Squares Initial Public Offering Inverse Mills Ratio Internal Rate of Return Log-Logistic Distribution Function Log-Normal Distribution Function Natural Logarithm Logarithm to Base 2 Mergers and Acquisitions Million MSCI Inc. Research and Development R,-Squared Measure Respectively United Kingdom United States of America

14

VC VSTOXX WEIB

Venture Capitalist Volatility Index of the EURO STOXX 50 Index Weibull Distribution Function

15

Chapter 1

Introduction Venture capitalists (VCs) are specialized intermediaries between innovative entrepreneurial firms and opportunity oriented investors. The rational for the co-existence of VCs and other intermediaries, such as banks, is given by the VC's higher ability to evaluate entrepreneurial projects (Ueda, 2004) and to monitor entrepreneur's effort (Hölmstrom and Tirole, 1997). This predicts entrepreneurial projects with less collateral, higher return, higher growth, and higher risk to attract venture capital instead of bank lending (Metrick and Yasuda, 2011). The role of the VC goes beyond the role of a passive screening agent. While a bank only monitors the financial health of its borrowers, the VC provides his portfolio firms managerial advice and access to his network to actively increase the survival and growth of the firms (Bottazzi et al., 2008; Bruton et al., 1997; Hellmann and Puri, 2002; Metrick and Yasuda, 2011; Sapienza et al., 1996). The venture capital set-up offers a proficient natural laboratory to test implications of the theory of the firm. This is because one can observe interactions between the VC (as the principle) and the entrepreneur (as the agent) in an isolated environment. 1 First, interactions between the 1

Also, contracts between the V C (as the agent) and its sponsors (as the principal) are analyzed in the literature (Chung et al., 2012; Gompers and Lerner, 1996, 1999a; Lerner and Schoar, 2004; M e t r i c k and Yasuda, 2010; P h a l i p p o u and Gottschalg, 2008; Sahlman, 1990). T h i s aspect is seized in chapter 3, for the case of information

16

VC and the entrepreneur are not impacted by confounding factors, for example dispersed stock ownership and short term pressure associated with publicly traded firms (Metrick and Yasuda, 2011). Second, the nature of innovative entrepreneurial firms entails that little public information about the firm exist and hence that information asymmetry between the VC and the entrepreneur is great. Finally, the VC directly invests equity in the entrepreneurial firm and thus has a claim in the firm's profit. This entails an incentive problem for the entrepreneur. 2 The funding relationship between the VC and the entrepreneur can be classified into three phases, whereas each phase is appropriately related to one specific kind of agency risk: First, the selection phase. In this phase, the VC screens several potential investment opportunities and acts as a screening agent to overcome adverse selection (Akerlof, 1970; Campello and Matta, 2010; Chan, 1983). Second, the investment phase. In this phase, the VC commits capital to the project and acts as a monitoring agent to reduce moral hazard (Bergemann and Hege, 1998; Neher, 1999; Cornelli and Yosha, 2003). Third, the exit phase. In this phase, the VC exits from the funding relation and acts as a certifying agent to reduce information asymmetry between the portfolio firm and the new investors (Barry et al., 1990; Cumming and Johan, 2008a; Gompers and Lerner, 1999b; Megginson and Weiss, 1991). In this thesis, I focus on the VC's role as a monitoring agent in the investment phase and as a certifying agent in the exit phase.3 The scope of my thesis is to improve understanding of agency cost related to ven-

2

3

asymmetries between corporate V C s and their corporate mothers, and its impact on exit outcomes. However, the focus of my thesis is the principal-agent framework between the V C and the entrepreneur. Since V C s add value t o their portfolio firms, the relationship between the V C and the entrepreneur is not a one-directional principal-agent problem. Moreover, there exists a double-sided principal-agent problem (Casamatta, 2003; Houben, 2002; Inderst and M ü l l e r , 2004; Repullo and Suarez, 2004; Schmidt, 2003). T h i s is not in the focus of my thesis. There are only few papers related t o the V C ' s role as a screening agent ( A m i t et al., 1990; Berglund and Johansson, 1999; C h a n et al., 1990; Hellmann, 1998; K a p l a n and Stromberg, 2001; Ueda, 2004; Van Osnabrugge, 2000). T h i s strand of the literature is mostly theoretical or qualitative in nature. A m a j o r reason for t h a t is d a t a availability. There is only private information owned by the V C s themselves about the firms they rejected t o fund.

Chapter

1.

Introduction

17

ture investments. I add new implications to this field, showing that (1) agency cost are time-variant in the investment phase, (2) the information rent earned by the VC in a funding relationship is biased under asymmetric information and (3) certification by the VC in the exit phase is most important if new investors face high search and screening costs. The results are in line with general theories regarding the compensation of risky employment (Amernic, 1984; Eisenhardt, 1989), credit assessment in loan portfolios (Duffie and Landò, 2001) and certification in markets with imperfect information (Baron, 1982; Chan, 1983). My thesis provides researchers and practitioners an improved toolkit to better understand the VC-entrepreneur relationship. The following three chapters coincide with the content of three my papers. For papers I wrote with co-authors, the personal pronoun 'we' is used in the thesis. For my single-author paper, the personal pronoun Τ is used. Further, the papers are formally revised for the thesis. In Chapter 2 Moral Hazard in VC Finance 4, we examine time-dependence of agency risk and its interaction with market risk. So far, papers have considered agency risk in venture projects to be independent time and environmental conditions (Bergemann and Hege, 1998; Neher, 1999; Cornelli and Yosha, 2003). However, we demonstrate that the environmental conditions (market risk in our set-up) significantly impact the incentives of the entrepreneur over time. The economic rational for this relation is as follows: the entrepreneur has motivation to expend high effort to grow the firm if future market conditions are seemingly favorable and the prospects of the firm are high. But, his motivation decreases if expectations about future market conditions change. In periods of high market risk, the entrepreneur's expected payoff from the project is downgraded. As a consequence, private benefits from managing the firm become more attractive to the entrepreneur than the realization of the project. We introduce a formal model, in which the VC has to cope with two tasks: he T h i s chapter includes my paper Moral Hazard in VC Finance: More Expensive than You Thougt. I ' m the corresponding author of this paper. Co-authors are Marie Lambert and Hans-Peter Burghof. I contributed t o this paper by the i n i t i a l idea, the formal model, the empirical research design and c o - w r i t i n g every chapter. Moreover, I was responsible for d a t a collection and preparation, the empirical analysis and interpretation of the results.

18

has to manage his investment with respect to market risk and he has to manage the project with respect to agency cost. We show that the VC optimally reacts to market shocks differently with respect to the relevance of entrepreneur's human capital for the success of the firm. In the case of high relevance of human capital, optimal investments to incentivize the entrepreneur to stay with the VC raise substantial opportunity loss from a real options perspective of investing. We test the implication from the formal model empirically and find support for our theory. In Chapter 3 Learning in Funding Relationships 0, I examine how learning of the VC about the prospects of an entrepreneurial firm in the funding relationship is impacted by asymmetric information between the VC and the entrepreneur. Asymétrie information provide the entrepreneur discretion in managing the firm and exposes learning to agency cost (Bergemann and Hege, 1998; Casamatta, 2003; Neher, 1999). To date, learning on the firm-level is only analyzed theoretically (Bergemann and Hege, 1998, 2005; Hsu, 2010). The reason for this is data availability (Sorensen, 2008). I provide contribution to this topic, analyzing learning on the firm-level empirically. For this purpose, I create an unique data set and analyze the impact of an informational update about the firm's planned capital appropriation on its valuation. I find that the VC needs comprehensive information about the firm to efficiently learning about its prospects and that efficient learning is distorted if comprehensive information exists, but is distributed asymmetrically between the VC and the entrepreneur. In Chapter 4 VCs as Intermediaries in Exits 6, we examine the role of the VC as an intermediary between the portfolio firm and the new investors in the exit phase. To improve the understanding of VC's role as T h i s chapter includes my paper Learning in Funding Relationships: Are Venture Capitalists really Different?. I ' m single author of this paper T h i s chapter includes my paper VCs as Intermediaries in IPO and Trade Sale Exits : Evidence from Industry-Level Analysis in Europe and the U.S.. I ' m the corresponding author of this paper. Co-author is is Alexander Steeb. T h e paper is based on a preliminary work of Steeb (2017). I n this paper, I extend the i n i t i a l idea, the d a t a basis and the analytical framework of the preliminary work. I c o n t r i b u t e d t o this paper by suggesting the i n i t i a l idea, d a t a collection, b u i l d i n g an analytical framework t o extend the i n i t i a l research idea, as well as c o - w r i t i n g every chapter. Moreover, I was responsible for the development of the empirical research design, the empirical analysis, and interpretation of the results.

Chapter

1.

Introduction

19

a certifying agent, we model exit outcomes on an industry-level. We do so, to analyze the VC's role for the different exit conditions in the different industries. So far, industry classification was only used as an explanation for the likelihood to exit to an IPO (Cumming and Johan, 2008a; Gompers and Lerner, 1999b). We find that the need for certification by the VC is conditional on the exit conditions within an industry and that certification by the VC is more important to exit to an IPO in industries with high IPO rates. The result indicates that VCs need to signal their ability as strong intermediaries to successfully exit their portfolio firms to an IPO if IPO investors face severe search and screening costs. In Chapter 5, I summarize the main findings and contribute to the discussion on the VC's role as an active investor.

21

Chapter 2

M o r a l H a z a r d in V C Finance 2.1

Introduction

Entrepreneurial firms operate in highly dynamic markets where future market conditions are particularly uncertain and can change very fast. In this environment, the VC has to cope with two tasks: he has to manage his investment with respect to market risk, and he has to manage the project with respect to agency cost. We analyze the resulting optimal investment decision of the VC in a real options framework. So far, the real options theory is rarely approached to the field of venture capital. Real options are rights but not obligations to take some actions in the future. This provides the decision maker flexibility to act upon informational updates (Dixit and Pindyck, 1994; McDonald and Siegel, 1986; Trigeorgis, 1996). In the real world, decisions are not static. For example, the VC can defer the initiation of a new project to wait for additional information about the future market conditions and initiate the project in the future only if market conditions turn out to be positive. Also, once started a project, the VC can abandon the project

22

.

Introduction

at any given stage of its development if environmental conditions turn worse, or he can expand it if environmental conditions turn more favorable (Trigeorgis, 1996). Flexibility is valuable in the real options theory if investments are exposed to two types of risk: exogenous risk that is irreducible through organizational activity and endogenous risk that can be influenced by organizational activity (Folta, 1998; McGrath et al., 2004). Hence, it is reasonable to analyze the funding decision of the VC in a real options framework if the following two assumptions are fulfilled (Hsu, 2010; Li, 2008). The first assumption is that market risk is exogenous to organizational activities, which means that it can not be impacted by the VC or the entrepreneur. We assume that market risk is related to unexpected market developments, e.g. technological shocks, trending consumer behavior or competitor's response that impact the value of the entrepreneurial project, but is out of the control of the VC and the entrepreneur and resolves primarily with the passage of time (Dixit and Pindyck, 1994; Li, 2008; McGrath, 1997; Pindyck, 1993). The second assumption is that agency risk is endogenous to organizational activity. Agency cost in entrepreneurial projects are related to the relationship between the VC and the entrepreneur. The entrepreneur inhibits the role of a contracting agent and owns unique human capital, such as specific skills or knowledge, essential to realize the project (Hart and Moore, 1994). We assume that this provides the entrepreneur power to threaten the VC with hold-up, because the VC can not continue the project without the entrepreneur (Neher, 1999). However, the VC can incentivize the entrepreneur to stay with him and realize the project through organizational activity, for example through contingent control allocation (Chan et al., 1990; Kirilenko, 2001) and the use of convertible securities (Casamatta, 2003; Cornelli and Yosha, 2003; Repullo and Suarez, 2004; Schmidt, 2003). In this chapter, we focus on the optimal timing of follow-on investments in an existing funding relationship. In the sense of real options, staging the total investment to a number of financing rounds provides the VC the opportunity to manage his investment based on two types of options (Li, 2008): the delay option to defer follow-on investments and the learning option to immediately commit investments. The delay option provides the

Chapter

2.

Moral

Hazard

in VC

Finance

23

VC the opportunity to wait for informational updates about the market conditions and to commit additional funds in the future if market conditions turn out to be favorable, or to abandons the project in the future to confine high downside losses (Li, 2008). If uncertainty about future market conditions is high, waiting for informational updates is economically more valuable than immediate investing (Dixit and Pindyck, 1994; McDonald and Siegel, 1986; Trigeorgis, 1996). The learning option provides the VC the opportunity to expand the project by immediately committing additional funds to learn about the prospects of the firm by observing its performance. Further, by observing the firm's performance the VC aggregates beliefs about the effort of the entrepreneur. This is valuable if human capital of the entrepreneur is relevant and if there is high uncertainty about the entrepreneur's effort, because the VC can learn about optimal incentivation of the entrepreneur. In this chapter, we consider the optimal timing of investments in this set-up if market risk and agency risk are interdependent. This interdependence has not yet been considered. So far, literature focuses on the optimal timing of investments considering independence of market risk and agency risk (Li, 2008; Gompers, 1995; Neher, 1999; Bergemann and Hege, 1998). However, interdependence is reasonable for the following reason: the entrepreneur has motivation to expend high effort to the project if the future market conditions are seemingly favorable and the prospects of the firm are high, but his motivation decreases if expectations about the future market conditions change and depress the prospects of the firm. The interdependence of market risk and agency risk hence refers to the idea of a relative value of private benefits. In periods of high market risk, the entrepreneur's expected payoff from the project is downgraded. Private benefits from managing the firm become more attractive to the entrepreneur relative to the realization of the project. In this way, the entrepreneur's incentive to expend effort decreases and his incentive to behave opportunistically increases. As a consequence, the treatment of market risk must affect optimal investments differently with respect to the relevance of entrepreneur's human capital for the success of the firm. In our set-up, the timing of follow-on investments can be interpreted as the decision of the VC whether to capitalize the delay option, which allows

24

2.2.

Formal

Model

him to reduce losses from unexpected (negative) market developments, or to capitalize the learning option, which allows the VC to discover the prospects of the firm and to optimally incentive the entrepreneur. Capitalizing the delay option charges agency costs to the VC, because he stays uninformed about the true potential of the entrepreneurial firm and hence can not optimally incentivize the entrepreneur. Capitalizing the learning option charges opportunity cost to the VC. This is because the VC must incentivize the entrepreneur for uncertain effort under market risk. As market risk tightens agency risk in our set-up, incentivation is very costly in this case. We introduce a formal model to show how incentives of the entrepreneur change in a world of uncertainty and analyze the optimal investment decision of the VC. We demonstrate that the entrepreneur has high incentive to consume private benefits if his human capital is essential for the realization of the project and market conditions worsen. In this situation, the VC mus accelerate investments to projects with a high relevance of the entrepreneur's human capital, relative to projects with a low relevance of the entrepreneur's human capital, to provide the entrepreneur incentive to stay with the VC. As a consequence, the VC suffers opportunity loss, because he must abandon the delay option when it is most valuable. We find empirical support for our theory analyzing the event history of individual financing rounds from European VC-backed entrepreneurial firms.

2.2 2.2.1

Formal M o d e l The Entrepreneurial Project

A financially constrained entrepreneur E initiated a project with uncertain returns. The project can be expanded by a follow-on investment. The funds are provided by a non-constrained venture capitalist VC. E must provide effort ε to realize the expansion of the project, with ε = {0,1}. The effort of the entrepreneur is a critical resource for the success of the firm, the VC cannot realize the project without the human capital of the entrepreneur (Hart and Moore, 1994; Neher, 1999). We describe the expan-

Chapter

2.

Moral

Hazard

in VC

Finance

25

sion of the entrepreneurial project as a one-shot problem. This assumption implies that the project can be expanded one time in its life-time and that the expanded project cannot be shrunk any more. The VC and the E are considered to be risk-neutral. The VC provides the total investment in equity financing only. The VC and E agree on a sharing contract in t = 0. The sharing contract defines the entrepreneur's share Se of the project's payoff and the follow-on investment. The project requires an investment Κ to be expanded. Investments are sunk once committed to the project. Time is standardized to the interval [0,1]. The discount rate is d = 0. The VC decides about the expansion at time T, with 0 < Τ < 1. Since the VC decides about the expansion of the project, he can realize all profits from the timing decision. In t = 1, a stochastic market shock π is realized with probability (1 — _p), with ρ [0,1]. (1 —ρ) is ex-ante known by the E and the VC. If the market shock is being realized in t = 1, the project fails and is valueless. Otherwise, the project generates the payoff V(e). The project's value follows a Bernoulli distribution.

G

(2.1) with V(l) > 0 and V(0) = 0. In other words, E has to expend effort to realize the project. This shows the relevance of the human capital of the entrepreneur for the success of the project. If the entrepreneur does not expend effort, the project is also valueless. The probability (1 — p) is out of the control of the E and the VC and resolves with the passage of time. There is a confidence level θ π(Τ) at time Τ that π is realized in t = 1. It describes the aggregation of beliefs about the future market conditions over time. θ π(Τ) is based on the square-rootof-time rule. The rule allows to scale down a higher frequency risk estimate to a lower frequency risk estimate. It is commonly used when financial risk is time aggregated. The confidence level θ π(Τ) is θ π(Τ) = τ 0·5

(2.2)

26

2.2.

2.2.2

Formal

Model

M a x i m i z a t i o n Function of the V C

The market risk (1 — ρ) constrains the project's expected payoff. Ko describes the case where the VC immediately invests Κ in t = 0. In this case, the VC immediately expands the project and capitalizes the learning option. The VC' s expected payoff P(0) v c is P(0) v c

= -Ko + (1 - SE)pV{e)

(2.3)

Over time, the VC becomes more confident about the shock π being realized in t = 1. Κ ι describes the case where the VC invests Κ in t = 1. In this case, the VC capitalizes the delay option. The follow-on investment is delayed until uncertainty about the realization of the shock is disclosed. The expected payoff of the VC still depends on p, because the VC only expands the project if the shock is not being realized. We assume that the VC can not expand the project later than t = 1. For t > 1 the investment opportunity vanishes and all options expire. The expected payoff P{l) vc is P{l) v c = {-K, + (1 - SE)V{e))p (2.4) The economic value of the delay option D(T) is positively related to the probability (1 — p) and delay Τ > 0. D(T) = P(T)

VC

- P(0) v c

(2.5)

= θ τ(Τ)[-Κ(ρ- 1)] The VC maximizes his total expected payoff P{T) P(T)

2.2.3

VC

= argmax{(l - SE)pV(e) -K τ

(2.6) VC

+ θ π(Τ)[-Κ(ρ

- 1)]}

(2.7)

Participation Constraint of the Entrepreneur

E chooses between the two effort levels ε = 1 and ε = 0. If E expends effort (ε = 1) the value of the project is F ( l ) with probability p, and zero otherwise. Since E is financially constrained and does not provide funds, the delay of investments has no economic value to him. His expected payoff

Chapter

2.

Moral

Hazard

in VC

27

Finance

P(t,e) E is Ρ(1,ε) Ε = P( 0,ε) Ε = Ρ(ε) Ε , where Ρ(1,ε) β specifies the case when the F C invests Κ in £ = 1 and Ρ(0,ε) β specifies the case when the VC invests Κ in t = 0. Since effort of the entrepreneur is essential to realize the project, there is a return α being realized on the human capital of the entrepreneur. It is α = 0 if the human capital of the entrepreneur is irrelevant for the realization of the project and α > 0 if the human capital of the entrepreneur is relevant for the realization of the project. For Τ > 0 the VC is uninformed about the prospects of the firm and cannot optimally incentivize E. Because E i s human capital is relevant, he can threaten the VC with a premature expiry of the investment opportunity, by leaving the project in T. In this case, E expends effort ε = 0 to the project and allocates his human capital to activities that go only to himself. In detail, E can realize an alternative profit C if he does not wait for the decision of the VC to expand or abandon the the project in T. Instead, E employs his human capital for alternative use until T. For example, E can start working as an adviser where he is compensated accordingly to the accumulated return of his human capital. We can think of this as experience. C is C = olT

(2.8)

α is assumed to be a logarithmic return function, so that the above equation is valid. E maximizes Ρ(ε) Ε = &Ygmdix{sEpV (ε) - C}

(2.9)

ε

The entrepreneur will choose ε = 0 and leave the project if C > sEpV( £) C

sEpV (ε)

> 1

(2.10) (2.11)

The ratio describes the relative value of private benefits. E i s incentive to behave opportunistically (E chooses ε = 0) is positively related to probability (1 — p) and to his return of human capital α. Thus, given a relative value of private benefits, highly qualified entrepreneurs are more likely to

28

2.2.

Formal

Model

leave a project than less qualified entrepreneurs if the project is treated by market risk. The entrepreneur is indifferent between ε = 1 and ε = 0 if his expected payoff from the project equals his private benefit C. The participation constraint P.C. E is sEpV( 1) > C sEpV(l)

2.2.4

-OLT > 0

(2.12) (2.13)

Solution w i t h o u t Relevance of H u m a n C a p i t a l

In the first case, there is no relevance of human capital. By definition this is α = 0. The entrepreneur's private benefit C from choosing ε = 0 is C = OLT = 0

(2.14)

Inserting C — 0 into the participation constraint of the entrepreneur P.C. E , the minimum payoff P ( l ) E to incentivize E to choose ε = 1 is P(1) E"

= 0

(2.15)

E earns nothing from the realization of the project and will always stay in the project. The VC maximizes P(T) V C. T** and S%* is

(2.16) Si? = 0

(2.17)

The payoff to the VC is P ( T * * ) V C = [-K + V(1 )}p

(2.18)

The VC earns the entire payoff from the project and realizes the total value of the delay option. This shows that entrepreneurs without valuable human capital can not realize a substantial profit in a VC-entrepreneur relation.

Chapter

2.2.5

2.

Moral

Hazard

in VC

29

Finance

Solution w i t h Relevance of H u m a n C a p i t a l

In the second case, E owns human capital and there is return from it. By definition this is α > 0. For Τ > 0, E can realize a private benefit C from choosing ε = 0. The VC has to vest share Se of the project's payoff to E to incentivize him to stay in the project and to choose ε = 1. With respect to the participation constraint of the entrepreneur P.C. E , Se is P.C. E : sEpV{ 1) = OLT

(2.19)

For V(l) = 1, we maximize the payoff P(T) T* =

(p-

VC

of the VC. Τ* and S% are

l)2

(2.21) (ρ- l ) 2 5* = ^ L (2.22) * 4 ap The return of the entrepreneur's human capital a and the market risk (1 — p) increase the share Se vested to the entrepreneur and reduce the residuum share of the VC. Since entrepreneur's human capital is positively related the profit the entrepreneur can realize in the VC-entrepreneur relation, higher qualified entrepreneurs are more likely to retain a larger share in their start-up. The payoffs to the VC and the E are P ( T * ) V C = (1 - S%)pV{ 1) - Κ + θ π(Τ*)[-Κ(ρ P(l)

E

= ceT*

- 1)]

(2.23) (2.24)

In this case, the VC and E share the surplus from the project. E earns the return from his human capital and the VC receives the residuum payoff and the value of the delay option. 2.2.6

O p p o r t u n i t y Loss o f t h e V C

In a world of uncertainty (1 — p) > 0, the option to delay investments is economically more valuable than immediate investing. This is shown in

30

2.2.

Formal

Model

(2.6): the value of delay option D(T) is positively related to the probability (1 — p). Since the optimal delay T* is negatively related to the relevance of human capital a, agency costs force the VC to advance investments. Opportunity loss is present if the VC can not capitalize the entire delay option, D(T) < D( 1). In the first best case, the VC invests Κ in t = 1 and realizes D{ 1). There is no opportunity loss. In the second best case, the VC maximizes his payoff P{T) V C with respect to the participation constraint of the entrepreneur P.C. E . For Τ > 0 agency costs C arise. For C > 0, the VC has to vest share Se > 0 to E to provide him incentive to choose ε = 1. The VC will benefit from further delaying the investment as long as the marginal profits from the delay option exceed the marginal compensations of E. dP(0) ^ ^ -IFAWO)-K

VC

+

, dD(T) dP{l) ^ ^ - ^ ^

E

(2 25)

'

+ T°' 6[—K(P -1)]) ^ ^ f c p y ( i )

at

-

at

1

j

For V(l) = 1, -α - ^

0 and T* < 1 opportunity loss arises. We present the resulting opportunity loss D( 1) — D(T*) from the optimal timing of the investment in figure 2.2. For any level of α > 0, opportunity loss decreases in market risk (1 — p). This means that the VC delays the investment to capitalize the delay option if market risk is high. This leads to our first hypothesis. Hypothesis H.2.1: VCs delay follow-on investments if investments are treated by market risk. Conversely, for any level of market risk (1 — p) < 1, opportunity loss increases in a. This means that the VC advances the investment to provide E incentive to choose ε = 1. This leads to our second hypothesis. Hypothesis H.2.2: VCs advance follow-on investments if the human capital of the entrepreneur is relevant. At least, opportunity loss increases by the factor , for a > 0. This is because the relative value of entrepreneur's private benefit increases in market risk. The VC optimally accelerates the investment to projects with a high relevance of the entrepreneur's human capital for the success of the project, relative to firms with a low relevance of the entrepreneur's human capital, to provide entrepreneurs with a high return

32

2.2.

Formal

Model

on their human capital incentive to expend effort to the project. This depresses the value of the delay option when the return on the entrepreneur's human capital is high. This leads to our third hypothesis.

Hypothesis H.2.3: Investments to firm with a high relevance of the entrepreneur's human capital are accelerated when market risk is high, relative to investments to firms with a low relevance of the entrepreneur's human capital.

Figure 2.1: Optimal Timing of Investments a is the return on human capital of the entrepreneur, a = 0 if the human capital of the entrepreneur is irrelevant for the realization of the project, a > 0 if the human capital of the entrepreneur is relevant for the realization of the project. (1 — p) is the level of the market risk.

Chapter

2.

Figure ments

2.2:

Moral

Hazard

Opportunity

in VC

33

Finance

Loss of the VC from

Optimal

Timing

of

Invest-

a is the return on human capital of the entrepreneur, a = 0 if the human capital of the entrepreneur is irrelevant for the realization of the project, a > 0 if the human capital of the entrepreneur is relevant for the realization of the project. (1 — p) is the level of the market risk.

34

2.3.

2.3 2.3.1

Empirical

Approach

Empirical Approach Sample

Our sample covers individual financing rounds from VC-backed entrepreneurial firms based in 15 European countries 1 for the period from 2003/01/01 to 2015/12/31 from Dow Jones VentureSource. The sample period covers the post-Dotcom bubble period and the financial crisis of 2007. Furthermore, the sample period covers the European sovereign debt-crisis in 2009 that came along w i t h an expansive monetary policy of the European Central Bank. We exclude firms from the energy and utilities sector, since energyrelated infrastructure projects and renewable energy production were strongly supported and highly regulated by the European Union within the sample period. Information about the entrepreneurial firm characteristics and the deal characteristics is from VentureSource. Ratios from listed peers are from Compustat global. To identify listed peers, all firms are included that are listed at the major exchange in on the 15 European countries considered. A l l non-Euro amounts are converted to Euro-equivalences by end of the day exchange rates from Thomson Reuters Datastream. A l l amounts are deflated to the reference year of 2003 by Eurostat inflation rate. Quotations of public market indices are also from Thomson Reuters Datastream. The GICS structure for industry classifications is from MSCI.

2.3.2

Econometric

Method

We analyze the time elapsed between two subsequent financing rounds a firm receives. We call this the duration between the financing rounds. A n increasing duration indicates the delay of follow-on investments and thus that the VC capitalizes the delay option. Conversely, a decreasing duration Belgium, A u s t r i a , Prance, Germany, Greece, Ireland, Italy, Luxembourg, Netherlands, Portugal, Spain, U n i t e d K i n g d o m , Denmark, Finland, Sweden

Chapter

2.

Moral

Hazard

in VC

35

Finance

indicates an advancement of follow-on investments and thus that the VC capitalizes the learning option (Gompers, 1995; Li, 2008). We use a difference-in-difference approach to analyze how the timing of investments is impacted differently for projects with a high relevance of the entrepreneur's human capital when treated by market risk. This effect is measured by the impact of the treatment on the duration between two financing rounds for entrepreneurial firms with a high relevance of the entrepreneur's human capital and for a control group with a low relevance of the entrepreneur's human capital. We use an accelerated-failure time model to fit information about the duration into a parametric regression model. Failure is the completion of the follow-on round. The censoring date is the date of the initial public offering, the acquisition or bankruptcy of the firm. Ongoing investments are censored after five years (1825 days). To follow each event for five years, we restrict our analysis to financing events that took place between 2003/01/01 and 2010/12/31. This leaves us with 7,336 observations. The survival function S(t) is the probability that a financing round is completed later than t S(t)=p(T>t )

The failure time is modeled by a linear effect composed of the co-variates and a random disturbance term e. e is a vector of errors assumed to come from a known distribution and σ is an unknown scale parameter. We fit the model to the natural logarithm of the duration and Weilbull distribution of the error term. The model fits the chapter's focus on the timing of investments since the estimation parameters in the accelerated failure time model can be directly interpreted as the influence of the explanatory variables on the failure time. Therefore, the model is commonly used in duration studies (Gompers, 1995; Li, 2008). The parameters are estimated by the maximum likelihood method w i t h a Newton-Raphson algorithm. The standard errors are estimated from the inverse of the observed information matrix. The model is LOG(T

ii k)

= ß0

+ βλ Riskt + β A Xi,k

+ ß2Asseti + crci yk

+ ß3[Risk t*

Asseti]

36

2.3.

Empirical

Approach

where T ^ is the time elapsed between the financing round k and the followon round k + 1 for firm i. Risk t is a time-series variable that captures the level of the market risk at funding date β ι estimates the time-serial effect of market risk on the timing of the follow-on investment. Asseti is a vector of cross-sectional variables that captures firm Vs asset structure to approximate the relevance of entrepreneur's human capital. β 2 estimates the cross-sectional effect of agency risk on the timing of future investments. ßs is the difference-in-difference estimator. X iì k is a vector of covariates specifying the deal characteristics of financing round k and the firm characteristics of firm i, β4 is the vector of its estimation parameters.

2.3.3

Measures

M a r k e t Risk The market risk is related to unexpected market developments that are out of the control of the entrepreneur and the VC (Li, 2008; McGrath, 1997; Pindyck, 1993; Dixit and Pindyck, 1994). We relate the market risk to market price volatility. It captures the level of accumulated market uncertainty (Cochrane, 2005). We use the level of the Euro Stoxx 50 Volatility Index (VSTOXX) at the funding date to measure the market price volatility. The index is based on Euro Stoxx 50 options prices and reflects the one year ahead market expectation of volatility.

Relevance of Entrepreneur's H u m a n C a p i t a l Entrepreneurial firms have almost no tangible assets and face high expenditures to set-up their business models, but earn only little revenues. Hence, the major asset of the firms is their growth opportunity. This is shown in the VC valuation method, which differs substantially from traditional valuation methods, such as the DCF approach or a multiples approach. The VC method to value a young entrepreneurial firm is based on a scenario approach, in which the development of the firm in favorable and unfavorable scenarios is forecasted and a potential final exit-value is estimated (Smith et al., 2011).

Chapter

2.

Moral

Hazard

in VC

Finance

37

The growth opportunity of an entrepreneurial firm primary emerges from the respective industry's growth potential and innovativeness (Brush and Chaganti, 1999; Chandler and Hanks, 1994; Covin et al., 1990; McDougall et al., 1992). The growth opportunity is an intangible asset of the firm. For firms with a high fraction of intangible assets, the human capital of the entrepreneur is found to be a critical determinant for success (Colombo and Grilli, 2010; Davila et al., 2003; Shrader and Siegel, 2007). So, we can easily the firm's asset structure to approximate the relevance of entrepreneur's human capital. We approximate an entrepreneurial firm's asset structure by the asset structure of listed peers. We approximate the asset structure by the market-to-book ratio and R&D-expenditures-to-sales ratio. To find an appropriate peer group of listed companies, we match sector classifications provided by Dow Jones VentureSource to GICS classification scheme used by Compustat global. Industry structure slightly differs for the GICS classification and the classification by Dow Jones. We match the classification schemes manually based on industry descriptions. We measure the industry's growth potential by the peer group's median market-to-book ratio at previous year's end and innovativeness by peer group's median R&Dexpenditures-to-sales ratio at previous year's end from Compustat global.

Contract Design The financing instrument defines whether the VC is more actively involved in the operations of the entrepreneurial firm or inherits a passive role. Providing equity, the VC receives power in the board of directors. This provides him the opportunity to monitor the entrepreneur's performance closely and approve significant decisions. In this way, equity possesses strong governance abilities compared to debt, since equity emphasizes behavioral control of the entrepreneur (Fama and Jensen, 1983). We control for the impact of debt-like and an equity-like instruments on the discretion of the entrepreneur. We assume pure debt and convertible debt to posses the characteristics of a debt instrument, whereas equity swaps and pure equity posses the characteristics of an equity instrument.

38

2.3.

Empirical

Approach

Further, smaller funding amounts limit the opportunistic behavior of the entrepreneur, because his bargaining power is reduced (Neher, 1999). A small funding amount will prevent the investor's claim from a bid down, and reduce entrepreneur's control in managing the firm. VCs regularly syndicate their investments to obtain a second opinion about the quality of the entrepreneurial project in the selection phase and to provide improved value-adding in the investment phase (Brander et al., 2002; Lerner, 1994). Moreover, those tasks are not independent from each other as an efficient selection of the project increases the effectiveness of the VC's involvement (Casamatta and Haritchabalet, 2007). VCs that syndicate their investment will be able to monitor the entrepreneur more efficiently. We control for the impact of syndication on the discretion of the entrepreneur. We define deals with two or more investors involved to be syndicated investments.

Firm-Specific Information According to Neher (1999), relevance of entrepreneur's human capital is reduced with a further development of the firm, since the entrepreneur's human capital mitigates to the firm in the form of patents and prototypes over time. We link the availability of such types of tangible assets to the firm's age. W i t h ongoing operations, the existence of tangible assets becomes more likely and the bargaining power of the entrepreneur reduces. We control for this impact on the asset structure of the firm. We include the natural logarithm of firm's age in our estimations. Moreover, entrepreneurial projects differ in quality. The VC gathers this information in the screening process (Casamatta and Haritchabalet, 2007; Sahlman, 1990). However, the screening process is exposed to adverse selection concerns. Higher quality projects have a more urgent need for capital to finance their future growth (Li, 2008). As a consequence, entrepreneurs initiating low quality projects have incentive to imitate higher quality projects to increase their funding. Hence, staging is more important for projects that pretend to be of high quality, since stage financing rejects low promising firms (Sahlman, 1990). We control for the quality

Chapter

2.

Moral

Hazard

in VC

Finance

39

of the project to separate adverse selection from moral hazard. This is important because the effort of the entrepreneur is a necessary condition, but not a sufficient condition for success of the project. Even a promising project will not succeed without entrepreneur's effort, however, even if the entrepreneur expends high effort to a low quality project, it will not succeed. We measure quality by an ex-post indicator of success which is an exit of the firm (Brander et al., 2002; Gompers, 1995; Gompers and Lerner, 2000b; Li, 2008; Sorenson and Stuart, 2001). We use information about an exit of the firm based on the information available in VentureSource at the date of data generation to control for project's quality. We define projects that exited to an IPO or a trade sale as high quality projects.

Further Environmental Conditions VCs adjust their investments according to market signals such as the general boom and bust cycles of the public equity markets and economic growth. We control for total VC funding in the previous year, which is positively related to good funding conditions (Cherif and Gazdar, 2011; Félix et al., 2013; Gompers et al., 2008). A favorable institutional environment, such as a strong corporate governance and investor protection, increases the of the VC to mitigate agency risk (Cumming et al., 2010; Jeng and Wells, 2000). We control for the entrepreneurial firm's country of incorporation to control for the impact of different institutional environments on the timing of investments. At last, we control for the entrepreneurial firm's industry sector to account for unobserved sector-specific fixed effects.

2.3.4

Empirical Results

In our sample, the average duration between two financing rounds is 593 days, approximately 19.5 months. Table 2.1 presents descriptive statistics and correlations of the variables. None of the correlations are sources of concern for multi-collinearity.

40

2.3.

Empirical

Approach

To apply the difference-in-difference approach, we create dummy variables to identify firms fraught w i t h agency risk based on industries' R&Dto-sales ratios and market-to-book ratios. Further, we create a dummy variable to identify financing rounds treated by market risk. We use logrank test statistic to estimate break-point values for the creation of the dummy variables. We apply the method of Contai and O'Quigley (1999) to estimate the break-point values. The method is an outcome-oriented approach to compute break-point values corresponding to its most significant relation with the outcome and is designed for survival analysis with censored data, which matches our purpose. We use this approach to account for non-linearity in the relation between the identification variables and the dependent variable. We compare break-point values and median values of the identification variables in table 2.2. The break-point value of the R&D-expenditures-to-sales ratio is higher than its median value. The higher break-point value indicates that only high levels of R & D intensity increase agency risk. The break-point value of the market-to-book ratio equals its median value. The break-point value of the V S T O X X is lower than its median value. The lower break-point value indicates that even low levels of market risk provide the VC an opportunity to take advantage of the delay option. We show the results from the duration analysis in table 2.3. Estimation (1) is the baseline model that includes the control variables. We find that VCs adjust the timing of investments to the contract design, the firm-specific information and further environmental conditions. Estimations (2) to (6) add information about market risk and agency risk. The new information does not change the impact of the control variables. H.2.1 states that VCs delay subsequent investments to wait for informational updates about the market conditions if market risk is high. We include the dummy variable for market risk in estimations (2), (4), (5) to analyze its impact on the duration. We find that VCs defer follow-on investments if the current investment is treated by market risk. In estimation (2), we model the duration based on the control variables and market risk. The treatment delays follow-on investments on average by approximately two months. This corresponds to 9% of the average duration between two financing rounds. In estimation (4), we add information about the firm's agency risk. The impact of the treatment remains unchanged. In esti-

Statistics and Correlation

Matrix

(a)

0.291

0.000

1.004

Max

(a)

VC

-0.014

(f)

-0.057 ***

-0-001

-0-013

-0.025**

(e)

0.006

(h)

0.461 ***

(i)

-0.031 ***

0.045 ***

0.064 ***

-0.008 -0.083*** 0.125*** 0.066*** 0.068*** 0.002 0.108*** -0.053 *** 0.051 *** 0.047***

(g)

-0.003

0.893

(d)

0.020* -0.006 -0.004 0.102***

(c)

-0.019 0.036 *** 0.066 *** 0.067*** 0.013 -0.152***

5,529 '

I°tai· Funding

2,671 '

Ü) VJ/

927

Firm Age IPO Tradesale

3,707 '

(g) (h) (i)

VC

Î 0taf. Funding

(f) (g) (h) (i)

(j) VJ/

-0.185*** 0.424

(b)

FundinS 3.185 8.705 0.002 502.7 0.095*** -0.019 Amount Syndication 0.576 0.494 0.000 1.000 0.054*** Firm Age 5.126 5.444 0.00 104.5 0.008 -0.020* IPO 0.060 0.238 0.000 1.000 0.141 *** 0.022* Tradesale 0.266 0.442 0.000 1.000 -0.021 *

arket-to- 6.188 3.561 1.070 28.93 -0.002 book VStoxx 0.235 0.101 0.116 0.875 -0.022* Debt 0.099 0.299 0.000 1.000 0.583

M

Min

(e)

(c) (d)

(b)

R&Dexpendituresto-sales 0.200

Std.

2. Moral

Mean

Pearson correlation coefficients reported. R&D-expenditures-to-sales=firm ,s industry R&D-expenditures-to-sales ratio. Market-to-book=firm's industry market-to-book ratio ratio. VStoxx= VStoxx 50 Index at the funding date. Debt=l, if funding is provided as debt. Funding Amount= Funding Amount in 2003 Euros. Syndication= 1 if deal is syndicated, 0 otherwise. Firm age=(Funding date - founding date of the company), in years. IPO=l if firm was exited to an IPO, 0 otherwise. Tradesale=l, if firm was exited to a Tradesale, 0 otherwise. Total VC Funding =( Total VC Funding per year, in 2003 Euros). *** indicates significance at the 1% significance level, * ^indicates significance at the 5% significance level, * indicates significance at the 10% significance level.

Table 2.1: Descriptive

Chapter Hazard in VC Finance

41

42

2.3.

Table 2.2: Variable

Empirical

Approach

Break-Points

Contai and O'Quigley (1999) break-point values reported. R&D=1, if R&Dexpenditures-to-sales> CO-break-point, and zero otherwise. R&D=1, if R&Dexpenditures-to-sales> CO-break-point, and zero otherwise. Growth=l, if Market-tobook> CO-break-point, and zero otherwise. Risk=l, if VSTOXX> CO-break-point, and zero otherwise. R&Dexpendituresto-sales Median CO-break-point

0.071 0.113

Market-tobook 5.573 5.538

VSTOXX 21.160 19.900

mation (5), we include the interaction term. In this model, the impact of market risk on the duration increases in magnitude. We find that the treatment delays follow-on investments by approximately three months if the human capital of the entrepreneur is not relevant. This shows that there is strong interdependence between market risk and agency risk. H.2.2 states that VCs advance follow-on investments to firms with a high relevance of the entrepreneur's human capital. We include the dummy variables for firm's asset structure in estimations (3), (4), (5). We find that VCs advance investments w i t h respect to relevance of entrepreneur's human capital. In estimation (3), we model the duration based on the control variables and firm's R & D intensity and growth opportunity. R & D intensity decreases time to follow-on investments by approximately two and a half months. This corresponds to 13% of the average duration between two financing rounds. Growth does not impact the duration. In estimation (4), we add information about the market risk. The impact of the other variables remains unchanged. When interdependence of market risk and agency risk is considered in estimation (5), the impact of the variable R & D slightly increases compared to estimations (3) and (4). In estimation (5), R & D intensity decreases time to follow-on investments by approximately three months. This corresponds to 14% of the average duration. H.2.3 states that the treatment of market risk impacts the timing of follow-on investments differently for firms with a high relevance of the entrepreneur's human capital. The difference-in-difference estimate shows

Chapter

2.

Moral

Hazard

in VC

Finance

43

that the treatment does not affect the duration between financing rounds differently w i t h respect to R & D intensity in estimation (5). However, the difference-in-difference estimate is significant for growth opportunities. The treatment affects delay of follow-on investments differently for the treatment group in this case. Here, investments to projects w i t h a high relevance of the entrepreneur's human capital are accelerated by approximately two and a half months relative to the control group. This result indicates that VCs increase their allocation of funds towards projects with a high relevance of the entrepreneur's human capital if investments are treated by market risk.

2.3.5

Robustness Checks

In estimation (6), we control for the aggregate level of equity valuations by the Euro Stoxx 50 Equity Price Index. This is done for two reasons. First, fund managers window dress their investments towards extant valuations to impress sponsors (Lakonishok et al., 1991). Risk might be related to window dressing of the VC if market risk and aggregate equity valuations are correlated. Since this is reasonable, the impact of market risk might account for window dressing activities instead of a strategic investment decision. Second, agency risk estimated by the market-to-book ratio might be biased towards the aggregate level of equity valuations. By its nature, the level of the market-to-book ratio is impacted by equity valuations in the time series. We may assign relatively more projects to the treatment group whose funding take place in periods of high aggregate valuations. In estimation (6), the market risk increases the delay of investments by approximately four and a half months. This corresponds to 24% of the average duration between two financing rounds. The impact of market risk does not vanish when controlling for window dressing activities, indicating that market risk accounts for strategic investing of the VC. In estimation (6), growth opportunities also do not impact the duration for a low level of market risk, indicating that the positive impact estimated in estimation (5) is driven by the selection bias. However, the difference-in-difference estimate is still negative, its magnitude does not change compared to estimation (5). This confirms robustness of hypotheses H.2.3.

Time Model : Funding Duration

Risk*Growth

(1)

(2)

(3)

(4)

(5)

(6)

~

(Table continued on next page)

-0.121 **

-0.119 ** (0.057)

11.354*** 13.100*** 11.286*** 12.894*** 12.678*** (1.257) (1.410) (1.252) (1.411) (0.131) -0.120*** -0.113 *** -0.134*** -0.096** (0.032) (0.032) (0.044) 0.017 0.030 (0.099 ** 0.069 (0.029) (0.029) (0.043) 0.085*** 0.078** 0.131 *** 0.212*** (0.030) (0.030) (0.043) (0.045) 0.022 0.012 (0.057)

(0.057)

(0.057)

(0.044)

(0.045)

8.699*** (1.579)

2.3.

Risk*R&D

Risk

Growth

R&D

Intercept

Dependent: Log(Funding Duration)

Regression coefficients reported. Standard errors in parentheses. Weibull distribution of the error term. R&D=1 if venture firm's industry R&D-expenditures-to-sales ratio is above the Contai and O'Quigley (1999) break-point value, and 0 otherwise. Growth=l if venture firm's industry market-to-book ratio ratio is above the Contai and O'Quigley (1999) break-point value, and 0 otherwise. Risk=l if the VStoxx 50 Index is above the Contai and O'Quigley (1999) break-point value at the funding date, and 0 otherwise. Debt=l, if funding is provided as debt. Funding Amount=log(Funding Amount in 2003 Euros). Syndication= 1 if deal is syndicated, 0 otherwise. Firm age=log(Funding date - founding date of the company). IPO=l if firm was exited to an IPO, 0 otherwise. Tradesale=l, if firm was exited to a Tradesale, 0 otherwise. Total VC Funding =Log (Total VC Funding in previous year, in 2003 Euros). Stoch Market=Log(Euro Stoxx 50 Price Index at the funding date). *** indicates significance at the 1% significance level, * ^indicates significance at the 5% significance level, * indicates significance at the 10% significance level. a indicates that the classification variables are significant at the 1% significance level based on type-3-effect analysis.

Table 2.3: Accelerated Failure

44 Empirical Approach

(1)

(2)

(3)

(4)

(5)

(6)

7,336 1.079 -10,120

7,336 1.079 -10,116

7,336 1.080 -10,113

7,336 1.080 -10,110

7,336 1.083 -10,108

-10,092

Hazard

7,336 1.079

2. Moral

Observations Weibull Shape Log Likelihood

-0.489 ***

Dependent: Log(Funding Duration)

Time Model: Funding Duration (continued)

-0.485 *** -0.481 *** -0.478 *** -0.478 *** -0.466 *** (0.042) (0.042) (0.042) (0.042) (0.042) (0.042) Funding Amount -0.018 * -0.019 * -0.017 * -0.017 * -0.017 * -0.024 ** (0.010) (0.010) (0.010) (0.010) (0.010) (0.010) Syndication -0.077 *** -0.076 *** -0.078 *** -0.077 *** -0.077 *** -0.072 ** (0.028) (0.028) (0.028) (0.028) (0.028) (0.028) Firm Age 0.062 *** 0.060 *** 0.061 *** 0.060 *** 0.059 *** 0.058 *** (0.009) (0.009) (0.009) (0.009) (0.009) (0.009) IPO -1.019*** -1.013 *** -1.014*** -1.009*** -0.827*** -0.994*** (0.052) (0.052) (0.052) (0.052) (0.030) (0.051) Tradesale -0.845 *** -0.835 *** -0.834 *** -0.827 *** -0.478 *** -0.813 *** (0.030) (0.030) (0.030) (0.030) (0.042) (0.030) Total VC Funding -0.183 *** -0.264*** -0.179*** -0.254*** -0.245*** -0.219*** (0.057) (0.064) (0.056) (0.064) (0.065) (0.064) Stock Market 0.431 *** (0.077) Location-fix yes a yes a yes a yes a yes a yes a Industry-fix yes a yes a yes a yes a yes a yes a

Debt

Accelerated Failure

Chapter in VC Finance 45

Time Model : Funding Duration , Early and Later Stages

0.041

Risk

Risk* Growth

(7)

(9)

(10)

(0.100)

(0.068)

14.082 *** 13.898 *** (2.632) (1.694) (1.713) -0.066 * -0.093 * (0.080) (0.037) (0.053) 0.021 0.093 * (0.076) (0.034) (0.052) 0.132*** 0.185*** (0.074) (0.037) (0.053) 0.032 (0.105) (0.067) -0.125*

(8)

Empirical

(Table continued on next page)

-0.151

10.059 *** (2.605) -0.256 *** (0.060) 0.113 (0.052) 0.128* (0.054) -0.045

2.3.

Risk*R&D

0.035

-0.260 ***

Growth

R&D

Intercept 10.571 ***

Dependent: Log(Funding Duration)

Regression coefficients reported. Standard errors in parentheses. Weibull distribution of the error term. R&D=1 if venture firm's industry R&D-expenditures-to-sales ratio is above the Contai and O'Quigley (1999) break-point value, and 0 otherwise. Growth=l if venture firm's industry market-to-book ratio ratio is above the Contai and O'Quigley (1999) break-point value, and 0 otherwise. Risk=l if the VStoxx 50 Index is above the Contai and O'Quigley (1999) break-point value at the funding date, and 0 otherwise. Debt=l, if funding is provided as debt. Convertible Debt=l, if funding is provided as convertible debt. Equity Swap=l, if funding was equity swap. Funding Amount=log(Funding Amount in 2003 Euros). Syndication= 1 if deal is syndicated, 0 otherwise. Firm age=log(Funding date - founding date of the company). IPO=l if firm was exited to an IPO, 0 otherwise. Tradesale=l, if firm was exited to a Tradesale, 0 otherwise. Total VC Funding =Log (Total VC Funding in previous year, in 2003 Euros). Stoch Market=Log(Euro Stoxx 50 Price Index at the funding date). *** indicates significance at the 1% significance level, **indicates significance at the 5% significance level, * indicates significance at the 10% significance level. a indicates that the classification variables are significant at the 1% significance level based on type-3-effect analysis.

Table 2.4'· Accelerated Failure

46 Approach

2,421 1.137 -3,095

Observations Weibull Shape Log Likelihood

(8)

(9)

(10)

4,915 1.077 -6,913

Hazard

4,915 1.076 -6,915

2. Moral

2,421 1.138 -3,094

-1.619*** -0.280*** -0.279*** (0.399) (0.060) (0.072) Convertible Debt -1.351 *** -0.463 *** -0.466 *** (0.185) (0.037) (0.053) Equity Swap 0.097 *** 0.092 *** (0.059) (0.170) Funding Amount 0.034 * 0.034 * -0.030 ** -0.030 ** (0.018) (0.018) (0.013) (0.013) Syndication -0.047 -0.047 -0.049 -0.049 (0.048) (0.048) (0.035) (0.035) Firm Age 0.077*** 0.076*** 0.120*** 0.118*** (0.012) (0.012) (0.020) (0.021) IPO -0.809 *** -0.803 *** -0.970 *** -0.971 *** (0.108) (0.107) (0.060) (0.060) Tradesale -0.645*** -0.641 *** -0.787*** -0.788*** (0.058) (0.058) (0.037) (0.037) Bankrupt 0.509*** 0.514*** 0.319*** 0.324*** (0.088) (0.088) (0.059) (0.059) Total VC Funding -0.183 -0.161 -0.328 *** -0.320 *** (0.119) (0.120) (0.076) (0.077) Location-fix yes a yes a yes a yes a Industry-fix yes c yes c yes a yes a

(7) -1.642*** (0.399) -1.351 *** (0.185)

Dependent: Log(Funding Duration)

Time Model : Funding Duration , Early and Later Stages (continued)

Debt

Accelerated Failure

Chapter in VC Finance

48

2.3.

Empirical

Approach

Krohmer et al. (2009) analyze the investment decision of the VC related to agency risk and market risk by the development stage of the project. They argue that agency risk is more present in the early stage of a project when uncertainty about entrepreneur's effort and ability is high, whereas market risk is more present in the later stage of a project when the VC decides about an exit or writing off the project. To adress this issue, we do sub-sample regressions based on the firm's investment status. We model the failure time of a firm's initial financing round to address timing of investments in the early stage of a project, and we model the failure time of a firm's subsequent financing rounds to address the timing of investments in the later stage of a project. The results are presented in table 2.4. In estimation (7) and (8), we model the duration from the first financing round a firm receives to the second one. The average duration to the second financing round is 664 days, approximately 22 months. In estimation (7), we estimate the effects of market risk and agency risk separately. In estimation (8), we consider interdependence of the effects. The treatment has no impact on the timing of follow-on investments in estimation (7), but has a slightly significant impact in estimation (8). Because market risk only slightly impacts the duration between the financing rounds, this indicates, that the delay option is not valuable in the early stage of a project. However, R & D intensity strongly decreases the duration to the second financing round by almost seven months, corresponding to 30% of the average duration. This indicates that the learning option is highly valuable in the early stage of a project. In estimation (9) and (10), we model the duration based on the same firms' subsequent financing rounds. The average duration between two financing rounds is 564 days, approximately 18.5 months. In estimation (9), we estimate the effects of market risk and agency risk separately and in estimation (10) we consider interdependence of the effects. The treatment increases the duration by approximately two and a half months in estimation (9), corresponding to 14% of the average duration between two financing rounds, and by almost four months in model (10), corresponding to 20% of the average duration between two financing rounds. This indicates that the delay option is valuable in the later stage of a project. R & D intensity decreases the duration by approximately one month in model (9),

Chapter

2.

Moral

Hazard

in VC

Finance

49

corresponding to 7% of the average duration between two financing rounds, and by almost two months in estimation (10), corresponding to 10% of the average duration between two financing rounds. Growth opportunities do not impact the duration in estimation (9) and increase the duration by almost two months in model (10), corresponding to 10% of the average duration between two financing rounds. The difference-in-difference estimate is negative. The magnitude of the coefficient is almost the same compared to estimations (5) and (6). The results indicate that in the later stage of a project, when market risk becomes a more relevant risk factor, there is opportunity loss from the abandonment of the delay option if the human capital of the entrepreneur is relevant for the success of the firm. We test the fit of estimation (5) compared to alternative parametric estimations. We present likelihood-ratio test statistics and correlations of predicted and observed failure times in table 2.5. The log-likelihoods for the exponential and log-normal distributions of the error term are slightly higher compared to the Weibull model we use. However, our results do not change if we apply an exponential or log-normal model. Further, we calculate the correlation between the predicted and the observed failure times to report a pseudo R? measure. We proceed as follows: we predict the expected failure time for each financing round by estimation (5) and transform the predicted failure times into censored data. For this purpose, we censor predicted failure times that exceed the prefixed time of five years. Then, we correlate the simulated sample and the original sample to estimate the correlation coefficient. The correlation coefficients are approximately 0.38 for all models and statistically significant at at the 1% significance level.

2.4

Conclusion

In this chapter, we analyze the optimal timing of follow-on investments to expand existing entrepreneurial projects in a set-up with market risk and agency risk. We consider interdependence of market risk and agency risk in that the strength of agency risk is related to the level of market risk. This offers some interesting new implications for optimal investing.

50

2 Λ.

Conclusion

In particular, we find that highly qualified entrepreneurs are more likely to leave a project than less qualified entrepreneurs if the project is treated by market risk. This is due to a higher relative value of private benefits if the entrepreneur can realize a high return on his human capital. As a consequence, we find that VCs optimally accelerate investments to projects with a high relevance of the entrepreneur's human capital for the success of the project, relative to firms with a low relevance of the entrepreneur's human capital. As a result, VCs allocate relatively more funds towards projects with a high relevance of the entrepreneur's human capital if investments are treated by market risk. This chapter has several implications for theory and practice. On the one hand, our results might help to explain observed return patterns of VC portfolios. Since agency conflicts are the main reason for the existence of the VC industry (Amit et al., 1998), it is reasonable that the agency cost also have a considerable impact on the realized returns of VCs. Cochrane (2005) shows that return patterns of VC portfolios are characterized through a high alpha and a high market beta. First, the alpha accounts for abnormal return that can not be explained by common risk factors. The real options perspective might explain the phenomenon: stage financing results in an option like payoff structure where the option premium accounts for the alpha. Second, the high market beta accounts for a high sensitivity of the returns to market risk. The relative value of private benefits might explain the phenomenon: when investments are treated by market risk, VCs allocate relatively more funds to projects with a high relevance of the entrepreneur's human capital. As a result, VCs tend to allocate funds towards relatively riskier projects when market risk is high. On the other hand, the results suggest the implementation of contractual claims that focus in particular on the mitigation of agency conflicts in periods of high market risk. From the perspective of the entrepreneur, market risk can be viewed as a high-risk employment situation. The entrepreneur will lose his employment and future income if the VC decides to abandon the project. Amernic (1984) and Eisenhardt (1989) show that the principal has to compensate the cost of risky employment to retain managers in high-risk situations. For projects with a high relevance of the

PLNORM

PLLOG

PGAMMA

PEXP

PWEIB

Sample

-10,108 -10,128 -10,041 -10,024 -10,136

hood

Likeli-

Log

0.380*** 0.379*** 0.380*** 0.377*** 0.376***

Sample

0.999*** 0.995*** 0.978*** 0.974***

PWEIB

0.995*** 0.978*** 0.974***

PEXP

0.993 *** 0.991 ***

PGAMMA

0.997***

PLLOG

Log Likelihood and correlation coefficients reported. Sample is the observed failure time. PWEIB is the predicted failure time by the accelerated failure time model using Weibull distribution of the error term. PEXP, PGAMMA, PLLOG, PLNORM CIRE the predicted failure from exponential, general-Gamma, log-logistic modeling, and log-normal modeling. The log likelihood is reported for the respective models.

2. Moral

Table 2.5: Model Fit

Chapter Hazard in VC Finance

51

52

2 Λ.

Conclusion

entrepreneur's human capital, a pre-defined compensation payment in the case of abandonment could act like a put-option held by the entrepreneur: the entrepreneur can sell his shares to the VC at a pre-fixed price if the project's value falls below a critical level. The compensation payment reduces the relative value of private benefits in periods of high market risk and thus relaxes the entrepreneur's incentive to leave the project. This will allow the VC to capitalize on a larger proportion of the delay option when it is most valuable.

53

Chapter 3

Learning in Funding Relationships 3.1

Introduction

Business models of entrepreneurial start-ups are genuinely new and innovative. W i t h their new business models, entrepreneurial firms tap into existing markets to initiate fundamental changes in established processes and to capitalize on the change from a first mover advantage (Choi and Shepherd, 2004; Lieberman and Montgomery, 1988; Robinson et al., 1994; Suarez and Lanzolla, 2007; Wernerfelt and Karnani, 1987). Also, creating new products and services, entrepreneurial firms can establish novel customer needs and create genuinely new markets that did not exist before. As entrepreneurial activity destructs established processes by its nature, the final effect of entrepreneurial activity on the organizational structure of a market is uncertain. As a consequence, entrepreneurial projects come along with highly uncertain payoffs and the entrepreneur as well as the VC can only guess about the future value of the entrepreneurial firm. VCs show highly explorative behavior funding entrepreneurial firms as they allocate funds to projects with initially unknown return distributions

54

3.1.

Introduction

and must actually learn about the prospects of the investment opportunity. Hence, identifying promising and unpromising projects early is essential to a VC's success. When prospects of a project are initially unknown, identification can be possessed through learning (Bergemann and Hege, 1998, 2005; Hochberg et al., 2010; Hsu, 2010). This is, the VC accumulates information about the projects over time to continuously improve his assessment of the firm's prospects. The more efficient this learning process proceeds, the earlier a VC identifies promising and unpromising projects, and the more successful he can be. VCs face two alternatives how to learn about a firm's prospects. First, they can rely on public information. Industry life-cycles regularly squeeze out exploited investment opportunities and raise new ones. On a general level, VCs can learn about the rise of new investment opportunities from other investors' behavior, signaled by the public market activity (Goldfarb et al., 2007; Linnainmaa, 2006; Lundvall and Johnson, 1994). Second, VCs can rely on private information from their own investment into a project. VCs usually stage their investment into an entrepreneurial firm to a number of financing rounds. This provides them the opportunity to build a relation with the entrepreneur and to gather information about the project through monitoring (Bergemann and Hege, 1998; Berger et al., 1999; Boot, 2000). As such, VCs learn about the prospects of individual a firm over time. Both, learning from public and private information, improves VCs' understanding of a new investment opportunity and thereby improves their future investment decisions. However, learning on the private level is much more important for VCs than learning on the public level. Once a growth opportunity emerges, it is also in the focus of other market participants and competing entrepreneurial firms. To succeed, an entrepreneur must win the race for a new technology, product or service. Providing too much information about his new business models publicly at an early stage decreases the success probability of the project, because competitors can free ride on the entrepreneur's first mover investment (Hoppe, 2000; Reinganum, 1985; Shankar et al., 1998). Consequently, only few information about a new investment opportunity will be available to the public and therefore the learning by VCs takes place to a large extent on a private level. Conse-

Chapter

3.

Learning

in Funding

Relationships

55

quently, VCs spend most of their time monitoring their portfolio firms. Gorman and Sahlman (1989) find that VCs spend around 60% of their time managing their investments. But, information production in funding relationships is exposed to agency conflicts if information asymmetry between the VC and the entrepreneur exists (Aghion and Bolton, 1992; Gompers, 1995; Lerner, 1994). Information asymmetry is reasonable, as it is the entrepreneur operating inside the firm and thus is assumed to have control over the allocation of information to the VC (Bergemann and Hege, 1998). This provides the entrepreneur discretion in managing the firm and hence disturbs efficient learning when the entrepreneur behaves opportunistically (Bergemann and Hege, 1998; Casamatta, 2003; Neher, 1999). For example, whereas the VC has interest in an early stopping of unpromising projects, the entrepreneur most often also has personal interest in realizing his business idea and hence has interest in continuing the project as long as possible. In Cornelli and Yosha (2003), it is shown that this motivates the entrepreneur to manipulate information allocated to the VC to ensure continuation of the firm. As a consequence, under asymmetric information, VCs do not identify promising and unpromising firms as early as they would under symmetric information. In this way, asymmetric information disturb efficient learning and results in the continuation of unpromising firms for too long. This is costly for the VC. To date, learning on project level is only analyzed theoretically (Bergemann and Hege, 1998, 2005; Hsu, 2010). The reason for this is data availability (Sorensen, 2008). I provide contribution to this topic, analyzing learning on a project level empirically. To the best of my knowledge, I ' m the first to do so. For this purpose, I create a unique data set. I analyze planned capital appropriation of entrepreneurial firms to analyze the value of informational updates, and to show conditions for varying value of updates. There are two main results of this chapter. First, I find that comprehensive information has a stronger impact on the valuation of an entrepreneurial firm than uncomprehensive information. This indicates that VCs learn faster given comprehensive information. Second, I find that

56

3.2.

Learning

and

Comprehensiveness

information asymmetry between the VC and the entrepreneur result in a valuation of the entrepreneurial firm based on a pool value of risk. This indicates that asymmetric information disturbs efficient learning.

3.2

Learning and Comprehensiveness

According to Berger et al. (1999), learning on a private level is defined by the following three criteria. First, the intermediary gathers information beyond public available information. Second, information gathering takes place over time through multiple interactions w i t h the agent. Third, information gathered remains confidential to the principal and the agent. In venture capital transactions, learning is performed through a staging of the investment to multiple financing rounds (Gompers, 1995; Lerner, 1994). This provides the VC the chance to monitor the development of the entrepreneurial firm closely and to learn about its prospects from informational updates disclosed at each financing round. The criteria defining learning are met for the relationship between the entrepreneur and the VC, since (1) information gathered at each financing round is project specific and hence goes beyond public information, (2) informational updates are disclosed through multiple interactions between the entrepreneur and the VC in the staging process, and (3) the informational updates remain confidential to the entrepreneur and the VC. Learning results in an informational advantage of the intermediary over outside investors that he can capitalize. VCs internalize the information rent by adjusting their future investment decisions based on the ninformational updates disclosed. According to March (1991) and Sorensen (2008), VCs engage in exploitative and explorative behavior, where exploitation describes the allocation of funds towards projects w i t h already known returns to capitalize information available, and exploration describes the allocation of funds towards projects with unknown returns to improve understanding of this investment opportunity. However, Sorensen (2008) relates exploitation to the internalization of public information and exploration to the internalization of private information. As a result, Sorensen (2008) finds that the rent earned by VCs is larger for explorative behavior.

Chapter

3.

Learning

in Funding

Relationships

57

Contrary to the proposition of Sorensen (2008), I propose that both, exploiting and exploring, is related to learning on the private level. This is, the business model of an entrepreneurial project is initially unproven and returns are unknown. Directing capital to the project, the VC explores the investment opportunity and learns about the prospects of the firm. Then, the VC exploits this information adapting his investment behavior in response to the new information. This is, funding is continued if an informational update encourages the VC to expect high prospects of the firm, and stopped otherwise. In this way, the VC internalizes the information rent from learning by rejecting low quality projects and expanding high quality projects (Fried and Hisrich, 1994). To identify promising and unpromising projects early, informational updates about the firm must be comprehensive. Comprehensive updates comprise complete information about the project and the market conditions. In this way, comprehensive information is a precise and unbiased estimate of the project's true success probability. Given the true success probability, the VC can easily estimate the true value of the entrepreneurial firm. In contrast, uncomprehensive information is imprecise and biased as it contains only incomplete information about the project and the market conditions. As a consequence, the project's true success probability remains uncertain and so the true value of the firm. In this case, the VC must continue information gathering, before he finally decides about the completion or abandonment of the project. This is costly, because the likelihood to continue an unpromising project increases and hence the information rent earned by the VC decreases. Thus, as comprehensive information is more valuable to the VC than uncomprehensive information, the value of an informational update is related to the comprehensiveness of the information comprised. Hypothesis H.3.1: Comprehensive information is positively related to the value of an informational update. Because information production is costly, the VC might restrict information production to a small set of information that he assumes is most relevant to approximate the project's true success probability. Furthermore, in the real world, information is always uncomprehensive, because it

58

3.3.

Frictions

in the Learning

Process

is impossible to think very far ahead in a complex and unpredictable world and to plan for all various contingencies. Also, for genuinely new business models and industries, determinants of projects' success may simply be unknown. In this case, the VC unintentionally produce uncomprehensive information until he explored the relevant variables. As such, informational updates can become more comprehensive over time as business models and industries mature. However, more comprehensive information helps the VC to identify promising and unpromising projects earlier. In this way, efficiency of learning is related to the comprehensives of information comprised in an informational update. If more valuable information about project's success probability are allocated to the VC, he can identify promising and unpromising projects earlier. Hypothesis H.3.2: Learning proceeds more efficiently if information is comprehensive.

3.3

Frictions i n t h e L e a r n i n g Process

The entrepreneur is operating inside the firm and thus is assumed to have control over the allocation of information to the VC (Bergemann and Hege, 1998). Hence, the learning process is exposed to asymmetric information between the VC and the entrepreneur. This provides discretion to the activities of the entrepreneur and disturbs efficient learning when the entrepreneur behaves opportunistically (Bergemann and Hege, 1998; Cornelli and Yosha, 2003; Gompers, 1995; Neher, 1999). For example, whereas the VC has interest in an early stopping of unpromising projects, the entrepreneur most often also has personal interest in realizing his business idea and hence has interest in continuing the project as long as possible. Cornelli and Yosha (2003) show that continuation preference motivates the entrepreneur to manipulate information allocated to the VC. This is, the entrepreneur tends to focus on short term goals to improve intermediate results and ensure funding in the next financing round. In this case, the information allocated to the VC is too positive and the VC continues

Chapter

3.

Learning

in Funding

Relationships

59

unpromising firms for too long. Also, the entrepreneur can signal high prospects of the firm shifting project's risk to increase funding in the next financing round. In this case, the VC will allocate higher portfolio weights to the high risk project and return volatility of his portfolio increases. VCs separate cash-flow from voting rights to allocate the rights in a way that opportunistic behavior of the entrepreneur is reduced. They do so, through the allocating of non-voting stocks and stock options, and through explicit contracts (Denis, 2004; Kaplan and Stromberg, 2001, 2003). First, VCs allocate cash-flow rights to the entrepreneur in a way that the entrepreneur's total compensation is very sensitive to the firm's success. So, the interest of the VC and the entrepreneur converge, and the entrepreneur has incentive to increase effort that also goes to the interest of the VC (Harris and Raviv, 1979; Hölmstrom, 1979; Innés, 1990). Second, VCs allocate control rights in a way that they have voting rights and board seats. So, they can enforce their interests and prevent actions that go only to the entrepreneur (Aghion and Bolton, 1992; Dewatripont and Tirole, 1994). If the entrepreneur and the VC have the same interest in the project, the incentive of the entrepreneur to disclose information truthfully increases, and the expectations of the investor and the entrepreneur converge early. This is valuable for the VC, because the probability that he continues an unpromising project too long decreases. Hence, informational updates comprise an information premium for the mitigation of asymmetric information. The VC internalizes the premium over time as investment decisions become more reliable when information asymmetry is reduced. Hypothesis H.3.3: The information premium increases in the level of information asymmetry between the VC and the entrepreneur. However, severity of agency conflicts is related to the characteristics of the project. For example, if work of the entrepreneur is highly interdependent. In this case, it is difficult for the VC to incentivize the entrepreneur expending effort, because external effects have a significant impact on the outcome of the entrepreneur's activity and hence the entrepreneur has discretion to manage resources inefficiently (Gompers, 1995). In this case, the informational update is not reliable and the VC has difficulties identifying

60

3.4.

Formal

Hypotheses

promising and unpromising projects. Hence, under asymmetric information, VCs identify promising and unpromising firms later as they would under symmetric information. In this way, efficiency of learning is related to the level of information asymmetry. Hypothesis H.3.4: Learning proceeds less efficient under asymmetric information.

3.4

Formal Hypotheses

Figure 1 visualizes the effect of positive and negative informational updates on the firm value for both symmetric and asymmetric information. Vo is the value of the entrepreneurial firm in to, before an informational update is disclosed. In to, the uncertainty about the true value of the project is high. In t i , an informational update about the prospects of the firm is disclosed, and uncertainty about the firm value is reduced. Positive information ( / + ) increases the VC's expectation about the project's success probability and the VC increases the valuation of the firm ( A F ( / + ) > 0). For negative information ( / " ) , the VC's expectation about the firm's success probability decreases, and the VC decreases the valuation of the firm ( A V ( I ~ ) < 0). For symmetric information, the effect of an informational update on the valuation of the firm is only related to its information value, which is conditional on the comprehensiveness of the information. The change in the valuation of the firm corresponds to the value of the new information. The information disclosed in t\ can either be comprehensive ( I c ) or uncomprehensive (lue)Given uncomprehensive information, high uncertainty about the firm's true success probability persists and the true value of the firm also remains uncertain. The impact of new information on the firm value will be small. Given comprehensive information, the true success probability of the project is uncovered, and so the true value of the firm. The impact of new information on the firm value will be large. This is true for positive and negative updates. Hypothesis H.3.1:

\AV(I C)\

>

\AV(I

UC)\

Chapter

3.

Learning

in Funding

61

Relationships

Given a peer group of firms, initial firm values in to, before information is disclosed, will not spread widely from one another, because individuals' success probabilities are unknown and the valuations must be based on a pool value of risk. If uncomprehensive information is disclosed for all firms in t i , uncertainty about individuals' true success probabilities persist. As a consequence, true firm values also remain uncertain and the valuations are furthermore based on the pool value of risk. Hence, firms' valuations will not spread widely from one another even after new information is disclosed. On the other hand, comprehensive information allows the VC to identify promising and unpromising projects early. After comprehensive information disclosed in t = 1, firms can be valued based on their individual success probabilities. In this case, firms' valuations will spread widely from one another after new information is disclosed. As a result, the efficiency of learning is observed in the variance of individual firms' valuations within a peer group of firms, after new information is disclosed. For increasing comprehensiveness of information, I expect increasing variance in the valuation of peer firms after information is disclosed. Hypothesis H.3.2:

a(AV(I c))

> a(AV(I

uc))

AV(I C)

A n informational update reduces information asymmetry between the entrepreneur and the VC. Hence, the informational updates must comprise an information premium (P(A)) for the reduction of asymmetric information. Consequently, the premium must be higher for a higher level of asymmetric information. In figure 3.2, describes a high level of asymmetric information, and A- describes a low level of asymmetric information. Since information asymmetry is reduced for positive and negative information, the premium is positive in value for both. Therefore, an informational update impacts the valuation of the firm conditional on the cumulative value of the information itself and the information premium. This is AV(I,A) = AV(I) + Ρ (A). I assume that information between the VC and the entrepreneur converges by Bayesian updating.Therefore, the marginal value of the information premium is decreasing over time as multiple interactions between the VC and the entrepreneur take place. Given an informational update comprising positive information ( / + ) , the cumulative effect of the information content and the information premium

62

3.4.

Formal

Hypotheses

Figure 3.1: Impact of Informational Updates on the Firm Value , Comprehensive and Uncomprehensive Information Hypotheses H.3.1. and H.3.2. illustrated. Ic is comprehensive information , lue is uncomprehensive information , I + is positive information , I~ is negative information , V(I) firm value given information I, AV(I) change in firm value after information I disclosed , σ ( Δ V ( I ) ) variance of individual firms' valuations in a peer group, after I disclosed. is m u l t i p l i c a t i v e . Consequently, t h e i m p a c t of a positive i n f o r m a t i o n a l update o n t h e f i r m value m u s t be larger for a higher level of a s y m m e t r i c information. G i v e n an i n f o r m a t i o n a l u p d a t e c o m p r i s i n g negative i n f o r m a t i o n (J7~), t h e c u m u l a t i v e effect of t h e i n f o r m a t i o n content and t h e i n f o r m a t i o n prem i u m is ambiguous. T h e r e is a negative i m p a c t of t h e i n f o r m a t i o n content o n t h e v a l u a t i o n of t h e firm, b u t a positive i m p a c t of t h e i n f o r m a t i o n premium.

Hence, t h e i n f o r m a t i o n p r e m i u m p a r t l y diminishes t h e i m p a c t of

negative i n f o r m a t i o n .

Since t h e value of t h e i n f o r m a t i o n p r e m i u m is in-

creasing i n t h e level of i n f o r m a t i o n a s y m m e t r y , t h e i m p a c t of a negative i n f o r m a t i o n a l u p d a t e o n t h e f i r m value m u s t be smaller for a higher level of a s y m m e t r i c i n f o r m a t i o n .

H y p o t h e s i s H.3.3a:

A+) >

A.)

H y p o t h e s i s H . 3 . 3 b : AV(I~,

A+) < AV(I~

,A-)

Chapter

3.

Learning

in Funding

63

Relationships

A s y m m e t r i c i n f o r m a t i o n provides t h e entrepreneur discretion i n m a n aging t h e firm. I f t h e entrepreneur aims t o ensure t h e c o n t i n u a t i o n of an u n p r o m i s i n g p r o j e c t , he w i l l t r y t o m i m i c a p r o m i s i n g p r o j e c t t o ensure funding. M i m i c k i n g a p r o m i s i n g p r o j e c t makes i t h a r d for t h e V C t o discover t h e t r u e q u a l i t y of t h e p r o j e c t .

T h i s is how learning is d i s t u r b e d .

G i v e n a peer group of firms, t h e t r u e f i r m values r e m a i n uncovered even after new i n f o r m a t i o n is disclosed i f h i g h i n f o r m a t i o n a s y m m e t r y exits. Consequently, firms w i l l be valued based o n a p o o l value of risk, and t h e valuations w i l l n o t spread w i d e l y f r o m one another. I f i n f o r m a t i o n asymm e t r y is low, t r u e f i r m values are discovered after new i n f o r m a t i o n is disclosed. I n t h i s case, f i r m values w i l l spread w i d e l y f r o m one another after new i n f o r m a t i o n is disclosed, because t h e V C can observe t h e t r u e q u a l i t y of t h e f i r m and value firms i n d i v i d u a l l y .

A g a i n , t h e efficiency of learn-

i n g is observed i n t h e variance of i n d i v i d u a l firms' valuations, after new i n f o r m a t i o n is disclosed.

Hypothesis H.3.4: a(AV(A +)) < a(AV(A_))

3.5 3.5.1

Data and Methodology Sample

M y sample covers i n d i v i d u a l financing rounds f r o m V C - b a c k e d entrepreneurial projects based i n 15 E u r o p e a n c o u n t r i e s 1 t h a t received venture c a p i t a l f u n d i n g i n t h e p e r i o d f r o m 2 0 0 3 / 0 1 / 0 1 t o 2 0 1 5 / 1 2 / 3 1 . T h e d a t a is f r o m D o w Jones VentureSource.

I exclude s t a r t - u p s f r o m t h e energy and u t i l -

ities sector. Energy-related i n f r a s t r u c t u r e projects and renewable energy p r o d u c t i o n were s t r o n g l y s u p p o r t e d and h i g h l y regulated b y t h e E u r o p e a n U n i o n w i t h i n t h e sample period. I n f o r m a t i o n a b o u t t h e entrepreneurial f i r m characteristics and t h e deal characteristics is f r o m VentureSource. A l l n o n - E u r o values are converted t o Euro-equivalences b y end of t h e day exchange rates f r o m T h o m s o n Reuters 1

Belgium, A u s t r i a , Prance, Germany, Greece, Ireland, Italy, Luxembourg, Netherlands, Portugal, Spain, U n i t e d K i n g d o m , Denmark, Finland, Sweden

64

3.5.

Figure 3.2: Impact of Informational ric and Asymmetric Information

Data and

Methodology

Updates on the Firm Value, Symmet-

Hypotheses U.S. and H.4- illustrated. / + is positive information, I~ is negative information , A+ indicates a high level of information asymmetry between the VC and the entrepreneur , A- indicates a low level of information asymmetry, V(J, A) is the firm value given information I and level of information asymmetry A, AV(I, A)is the change in the firm value after information I is disclosed given the level of information asymmetry A, a(AV(I,A)) variance of individual firms' valuations in a peer group, after lis disclosed given the level of information asymmetry A.

Datastream. All amounts are deflated to a reference year of 2003 by Eurostat inflation rate.

3.5.2

Measures

Valuation The value of an entrepreneurial firm is the pre-money valuation at the funding date. The pre-money valuation is calculated as the post-money valuation reported less the financing amount received. This is the value of the firm after new information is disclosed, but before additional funds are committed.

Chapter

3.

Learning

in Funding

Relationships

65

Information Content of Updates I use the information about an entrepreneurial firm's planned capital appropriation to approximate the informational update disclosed to the VC. Planned capital appropriate provides information about future activities, and discloses information about the development stage of the firm. A further development of the firm increases its success probability (Agarwal and Gort, 2002; Das et al., 2003; Davidsson, 1991). For example, entrepreneurs planning to use funds to accelerate the development of the firm successfully completed the incubation of an entrepreneurial idea. This increases the expected success probability of the firm. In contrast, if the entrepreneur plans to allocate funds to the incubation of the entrepreneurial idea, this means that the entrepreneur did not yet create a lasting business, and the likelihood that the project fails is great. Usually, capital appropriation is confidential to the entrepreneur and the VC, therefore this is an appropriate information for my purpose. Fund details in VentureSource include a description of the planned capital appropriation for a limited number of firms. The descriptions are nonstandardized and comprise up to 926 characters. By text analysis, I identify the key information about firms' planned capital appropriation. Based on that, I define three types of information that can potentially be disclosed to the VC through an informational update. The information signals the development stage of the firm to approximate the VC's expectation about the success probability of the firm. The information is: The entrepreneur plans to allocate fund to (1) the incubation of an entrepreneurial idea, (2) the acceleration of the project, or (3) to grow the firm. I define whether an informational update is positive or negative in the cross-section of observations by a reference signal approach. To define a positive informational update, the reference signal is given by the information that the entrepreneur allocates funds to the incubation of an entrepreneurial idea. Hazard is highest in this development stage, and thus the firm value is small. In this case, hazard is lower, and thus the firm value is higher, if the entrepreneur would allocate funds to the acceleration of the project, or the growth of the firm. Thus, accelerating the project or growing the firm pose a positive signal relative to the incubation stage

66

3.5.

Data and

Methodology

of an idea. To define a negative informational update, the reference signal is given by the information that the entrepreneur allocates funds to the growth of the firm. Hazard is lowest in this development stage, and thus the firm value is highest. In this case, hazard is higher, and thus the firm value is lower, if the entrepreneur would allocate funds to the incubation of an idea or the acceleration of the project. Thus, incubating an idea or accelerating the project pose a negative signal relative to the growth stage of a firm. I typify operational activities described in the planned capital appropriation to identify the relevant information content. I define the incubation of an entrepreneurial idea by funds being used to develop a product or to start initial sales activities. I define the acceleration of a project by funds allocated to the commercialization of a product and to extend the sales activities. Furthermore, acceleration is defined by the professionalization of business operations, which is investments into production capacities and the recruitment of employees. I define growth of the firm by funds being allocated to differentiate products, to expansion sales internationally, and to form alliances and conduct acquisitions. I create a dummy variable for each category that is equal to one if the information given matches with the specific development stage, and zero otherwise. Some of the descriptions cannot unambiguously be categorized. In these cases, I respect a pecking order in primary capital appropriation to assign each financing round uniquely to one development stage. The pecking order acknowledges the fact that, by nature, a product is commercialized before it is sold on a large scale. Therefore, I assign information that comprises attributes of the incubation and the acceleration stage based on the information about the product development. And I assign information that comprises attributes of the acceleration and the growth stage based on the information about the sales activities. In both cases, all the information available is considered, and therefore I define the information to be comprehensive (Ic)> Additionally, I create uncomprehensive information (lue)For this purpose, I intentionally ignore information about either product development or sales activity. This corresponds to situations when the VC has

of Financing B,ounds I

1

The comapny will use the funding for product development and rapid expansion into additional markets in Europe and North America.

^

0

0

0

0

0

0

1

1

1

0

incu

1

IR&D

(Table continued on next page)

The funding will be used to focus on new applications for the technology and for an q q engineering center in East Croydon, UK.

0

q

grow

0

0

0

0

0

0

acc

0

0

1

0

ISALE grow

0

0

0

incu

0

0

0

acc

1

grow

in Funding

The funds will be used to develop the company's forthcoming "social-music" router. 1

ι

Ic acc

3. Learning

The company will use the financing for research and development efforts and to commercialize new products.

incu

Ic is comprehensive information, IR&D information about the product development (uncomprehensive information), I s ALE information about the sales activity (uncomprehensive information), /+ is positive information, I~ is negative information

Table 3.1: Classification

Chapter Relationships

67

q

0

IC

1

0

1

acc

1

0

grow

0

IRSZD

0

0

0

incu

0

1

1

acc

0

0

0

ISALE grow

0

0

0

incu

1

0

1

1

0

acc

grow

3.5.

(Table continued on next page)

Funding will be used to expand operations into South East Asia, Australia and the Americas. The funding will also be used accelerate the expansion of Aconite's Affina Enterprise into the new growth markets.

0

0

incu

of Financing B,ounds I (continued)

Thefinancing will be used to further develop the company's next-generation telecom revenue-assurance portfolio, accelerate its geographic expansion across the Americas and Asia Pacific, and increase support for existing and new customers.

The company plans on using proceeds from this round to accelerate the commercial development of its proprietary non contact sleep monitoring technology and to continue the development of clinical products for sleep apnoea screening and disease management.

Classification

68 Data and Methodology

0

1

0

0

1

The company plans on using proceeds from this round to start distribution activity and hire a sales team. 0

1

0

0

0

0

0

0

0

1

0

0

1

1

0

grow

0

IR&D

0

0

0

0

0

0

incu

0

0

0

0

0

0

acc

0

0

1

0

0

ISALE

grow

0

1

0

0

0

incu

0

0

1

0

1

1

grow

0

0

0

0

acc

in Funding

The company will use the funding to develop the new service and to strengthen its position in Miami, Seattle, and Denver, US.

The funding will be used to accelerate growth into new markets such as Asia and the Americas and to recruit new staff. The new capital investment will allow the company to accelerate its growth through acquisitions and partnership agreements.

hiring.

0

Ic

acc

3. Learning

The company will use thefinancing for staff

company's strategy and positioning.

incu

of Financing B,ounds I (continued)

The funding will be used to reinforce the

Classification

Chapter Relationships

of Financing B,ounds II

1

^

0

0

q

1

T-

R&CD

T+

(Table continued on next page)

The funding will be used to focus on new applications for the technology and for an ^ engineering center in East Croydon, UK. ^

1

The comapny will use the funding for product development and rapid expansion into addi- 0 tional markets in Europe and North America.

The funds will be used to develop the company's forthcoming "social-music" router. 0

0

C

T1

The company will use the financing for research and development efforts and q 1 commercialize new products.

C

Γ+ 1

1

q

0

q

1

R&CD

T+

q

0

0

SALE

T-

SALE

LE is comprehensive information, IR&D information about the product development (uncomprehensive information), Is ALE information about the sales activity (uncomprehensive information), /+ is positive information, I~ is negative information

Table 3.2: Classification

3.5. Data and Methodology

q

1

q

(Table continued on next page)

^

0

1

1

1

Γ+

^

Â

Γ-

1

1

R&D

q

0

1

SALE

SALE

in Funding

Funding will be used to expand operations into South East Asia, Australia and the Americas. The funding will also be used to ^ accelerate the expansion of Aconite's Affina Enterprise into the new growth markets.

1

0

Γ-

3. Learning

Thefinancing will be used to further develop the company's next-generation telecom revenue-assurance portfolio, accelerate its geographic expansion across the Americas and Asia Pacific, and increase support for existing and new customers.

1

Γ+ Γ- Γ+

of Financing B,ounds II (continued)

The company plans on using proceeds from this round to accelerate the commercial development of its proprietary non contact sleep monitoring technology and to continue the development of clinical products for sleep apnoea screening and disease management.

Classification

Chapter Relationships

7

1 1

1

C

0

1

C

J

i l l

0 1

The company will use the funding to develop the new service and to strengthen its position in Miami, Seattle, and Denver, US.

The company plans on using proceeds from this round to start distribution activity and hire a sales team.

0

0

Γ-

0

0

Γ+

0

0

R&CD

0

1

0

into new markets such as Asia and the Amerl 0 0 icas and to recruit new staff. The new capital investment will allow the company to accelerate its growth through 1 0 0 acquisitions and partnership agreements.

The funding will be used to accelerate growth

hiring.

The company will use thefinancing for staff

company's strategy and positioning.

J

Γ+ Γ- Γ+

0

0

1

0

SALE

1

0

0

1

Γ-

of Financing B,ounds II (continued)

The funding will be used to reinforce the

Classification

0

1

SALE

1

1

3.5. Data and Methodology

Chapter

3. Learning

in Funding

Relationships

73

incomplete information. I match the entrepreneurial activity to the development stage of the firm based on information about the product development or sales activity only. W i t h this approach, I create two types of uncomprehensive information, IR&D and I s ALE- Both information inhibit the characteristics of uncomprehensive information, because the information is biased towards one specific aspect of firm development. I present examples from the classification procedure in table 3.1.

Information A s y m m e t r y Information asymmetries mitigate with multiple interaction between the VC and the entrepreneur. In early financing rounds, the VC accumulated only few information about the project and must estimate the prospect of the firm to a great extent based on the information provided by the entrepreneur. In later financing rounds, the VC accumulated information from multiple interactions and expectations og the VC and the entrepreneur converge. I define three funding stages of the start-up company to approximate the level of information asymmetry: Seed- and first-financing rounds (A+++), second- and third-financing rounds and forth- and later-financing rounds where + indicates the the level of information asymmetry. It is A + + + > A++ > A+. I build a dummy variable for each category that is equal to one, if the financing round matches with the respective funding stage, and zero otherwise.

Learning Paths I simulate different learning paths (L) in the cross-section of observations interacting informational updates and financing rounds. For example, the information about the entrepreneur planning to use funds to accelerate the project can be disclosed in seed- and first-financing rounds, or in second- and third-financing rounds. Although the update comprises the same information content, the VC undergoes a different learning path in the two cases. If the information is disclosed in a seed- or first-financing

74

3.5.

Data and

Methodology

round, information is allocated asymmetrically between the VC and the entrepreneur, and the update is less reliable. In contrast, if the information is disclosed in a second- and third-financing round, the information is more reliable, because expectations of the VC and the entrepreneur converges through multiple interaction. However, the value of the information premium is negatively related to the reliability of updates. Since information is asymmetrically distributed between the VC and the entrepreneur in an early financing round, any information disclosed in seed- and first-financing rounds is of high value to the VC. In later financing rounds, expectations converged through multiple interactions and the information premium on convergence of information is small. The learning paths are represented by any combination of information content and information asymmetry. In this way, a set of synthetic learning paths is simulated. I can compare (1) the value of informational updates comprising different information at the same value of the information premium, and (2) the value of informational updates comprising similar information at different values of the information premium. In a regression, the estimated coefficients for L measures the valuation jump, when switching from one learning path to another. This is AV(I ÌA) Ì and estimates the difference in the values of the informational updates.

3.5.3

Econometric Specification

Pre-money Value Regression The empirical analysis focuses on the impact of an informational update on the entrepreneurial firm's valuation. I use a generalized least square regression model (GLS) to do so. The model is Log{V

ii k)

= ß0 + ßiL ii k

+ ß2I+l~

+

+

ι3·1)

Vitk is the pre-money valuation of the entrepreneurial firm i at the funding round k. L is the matrix * ßi is the vector of parameters of interest. is a vector that defines the information to be positive or negative, and is a vector that defines the level of information asymmetry, e^k is the estimation error. Since the sample includes multiple

Chapter

3.

Learning

in Funding

Relationships

75

observations, e.g. a firm receives multiple rounds of financing, I cluster estimation errors at the firm level.

Survivorship-Bias Correction The VC decides about the continuation of the firm based on the informational update. Since I can only observe information about planned capital appropriation for continued projects, the estimation is exposed to a survivor-ship bias. M y sample only includes observations where the informational update lead to a continuation of the funding, but the sample does not include observations where informational update resulted in such a dramatic downgrading of firm's prospects that the VC abandoned the project. As a consequence, the impact of positive updates will be overestimated and the impact of negative updates will be underestimated. I solve this issue by an approach similar to Cochrane (2005). In this paper, Cochrane (2005) proposes an approach to tackle the survivorshipbias in the estimation of a VC's portfolio return. This is, the final values of portfolio firms can only be observed if they are exited. Hence, estimating the average internal rates of return (IRR) for VC portfolios is biased, because unsuccessful firms are underrepresented in the estimation. Cochrane (2005) therefore adjusts the IRR calculation for the firm's hazard and finds that average IRRs are overestimated when not correcting for this bias. I use a proportional hazard model (Cox et al., 1979) to estimate the firm's hazard based on a set of co-variates. The hazard is the probability that the firm does not receive a subsequent financing round. I relate survival to the age of the entrepreneurial firm. This is estimated based on all observations in my sample that provide a complete funding history. The hazard model is hi(Age) = ho(Age)Z*

(3.2)

Age is the natural logarithm of firm V s age at the funding date. ho(Age) is an arbitrary function of time. It is not specified and one does not need to assume a specific distribution function for the hazard ratio. Xi(Age) is a vector of co-variates for firm i at the given age. ß\ is the vector of

76

3.5.

Data and

Methodology

parameters to be estimated. Firm's hazards is e^ 1 , which does not depend on age an more.

Endogeneity correction The information I use to estimate the impact of learning is not truly private as this is information from VentureSource and hence accessible to a restricted group of people. However, only for a few financing rounds information about the capital appropriation is available, indicating that this information is rather close to private information than to public information. However, the availability of the information might be biased across observations to situation where the VC has an advantage from providing naturally private information to a restricted group of people. For example, VCs might provide access to this information to attract attention in their community and to impress sponsors. In this case, VCs report positive updates to emphasize their ability and effort, but hide negative updates to prevent reputation loss. I solve the bias problem through a two stage estimation procedure considering an endogenous treatment effect. This is a common approach to control for a possible selection bias (Hand, 2007). First, I estimate a selection equation on the probability that information about the capital appropriation is available through a probit model. The model is p(h) = ßo + ßiXk + ek

(3.3)

Ik is a dummy variable equal to one if a description about planned capital appropriation is available for funding round k. Xk is a vector of deal specific characteristics for funding round and ß\ is the vector of parameters to be estimated, e^ is the estimation error. Then, an inverse mill ratio type of control variable is computed (Heckman, 1979). Given estimates from the probit model, the inverse Mills ratio is 1MB

=

Φ(βι * Vk) Φ(βι * Vk)

(3.4)

Chapter

3.

Learning

in Funding

Relationships

77

φ is the normal probability density function and Φ is the normal cumulative density function. ß i * V k is the matrix from vectors of estimated coefficients and independent variables.

3.6

Results

3.6.1

S u m m a r y Statistics a n d Selection Regressions

Figure 3.3 shows the results from the classification of financing rounds. The numbers reported outside the bars show the cumulative funds committed, the numbers inside the bars show the average financing amount per deal. In seed- and first-financing rounds, most funds are allocated to incubating activities. In second- and later-financing rounds, funds are mostly allocated to acceleration and growth activities. Overall, 21.6% of the funds are allocated to incubation activities and 41.2% to acceleration activities. The amounts almost equal the allocation of funds to seed- and first-, and second- and third-financing rounds (21.9% and 40.1%). This shows that my classification scheme appropriately matches development stages of the firm w i t h its funding process. Average financing amount per deal increases for similar activities across financing rounds. The average financing amount for incubating activities is 2.12 mio. Euros in seed- and first-financing rounds, and 3.53 mio. Euros, respectively 7.07 mio. Euros in second- and third-financing rounds, respectively forth- and later-financing rounds. Similar patterns are observed for acceleration and growth activities. This shows that VCs are willing to increase their engagement in similar activities of the entrepreneur, if funding conditions improve. Also, the variance of funds committed to different activities increases among financing rounds. Whereas average funding amount allocated to incubation activities differs only marginally from the average funding amount allocated to growth activities in seed- and first-financing rounds, funding amounts spread widely from one another in forth- an later-financing rounds. In seed- and first-financing rounds entrepreneurs receive on average 2.12 mio. Euros for incubation activities, and 1.97 mio. Euros for

78

3.6.

Results

growth activities. In contrast, in forth- and later-financing rounds, entrepreneurs receive on average 7.07 mio. Euros for incubation activities, and 12.77 mio. Euros for growth activities. This shows that VCs consider entrepreneurs' activities more individually, if funding conditions improve. In figure 3.4, average pre-money valuations are reported. The figure provides a first insight into the valuation patterns of the different learning paths. This is, positive information results in a valuation premium. This is shown as firms that plan to allocate fund to accelerartion or growth activities are value higher than firms allocating the funds to inbubation activities, in all funding stages. Also, we observe that the valuation of firms within the same development stage increases if funding conditions improve. For example, in seed- and first-financing rounds, firms in the incubating stage are valued at an average of 7.03 mio. Euros, whereas firms in the incubation stage are valued at an average of 23.79 mio. Euros, in forth- and later-financing rounds. The variance in the firms' valuations increases among financing rounds. In seed- and first-financing rounds, the average valuation of firms in the incubation stage is 7.03 mio. Euros, and average valuation of firms in the growth stage is 7.54 mio. Euros. In contrast, in forth- and later-financing rounds, the average valuation of firms in the incubation stage is 23.79 mio. Euros, and average valuation of firms in the growth stage is 61.51 mio. Euros. This indicates that VCs value firms on a pool value of risk in early financing rounds, and consider firms more individually in later financing rounds. Before reporting the results of the pre-money valuation that are the focus of the next section, table 3.2 reports estimates from the selection equations. Column 1 presents estimates on the likelihood that information about the planned capital appropriation is available in VentureSource. Higher financing amounts, increasing age of the firm and syndication of the deal increase the probability that information is provided. First, according to the summary statistics, higher financing amounts are related to a further development of the firm. If the firm is further developed, providing original private information publicly is less costly for the VC, because firms

Chapter

3.

Learning

in Funding

Relationships

79

1,100.00

1,068.98 1,031.43

wwh 27

4.03

1,000.00

948.72 4.52

900.00 800.00 700.00 600.00 £ ο β

500.00

536.29 2.12

547.49 j.rü 458.27 2.62

400.00 1.97 300.00

261.63

200.00

100.00 0.00 mcu

acc

grow

Seed & 1st

Figure

3.3: Average

incu

acc

grow

incu

2nd & 3rd

and Aggregate

Investments

acc

grow

4th & Later

Inf. Update Fin.Round

to Projects

Amounts reported in mio. Euros. Numbers reported outside the bars show the cumulative funds committed, numbers inside show the average financing amount per deal, incu is the incubation stage of the firm, acc is the acceleration stage, grow is the growth stage.

80

3.6.

Results

70.0061.51 60.00-

50.00-

40.40

40.00-

30.0023.79 20.00-

14.99

16.9 I

12.20 10.00-

7.03

JÎ21

7.54

acc

grow

0.00 J incu

incu

Seed & 1st

Figure

3.4'· Average

Pre-Money

acc

grow

2nd & 3rd

Valuations

incu

acc

grow

4th & Later

of Entrepreneurial

Inf. Update Fin.Round

Firms

Amounts reported in mio. Euros. Numbers show the average pre-money valuations of the entrepreneurial firms, incu is the incubation stage of the firm, acc is the acceleration stage, grow is the growth stage.

Chapter

3.

Learning

in Funding

Relationships

81

competitive advantage may rely to a smaller extent on confidential information. Second, w i t h increasing age, information about the firm is disclosed naturally as they attract attention in their community. Hence, the intrinsic value of truly private information decreases in firm's age, and learning about matured firms relies to a smaller extent on private information. Third, internal information flow is exposed to agency risk among the syndicating partners (Casamatta and Haritchabalet, 2007; Grossman and Hart, 1980; Hölmstrom, 1982; Pichler and Wilhelm, 2001; Wilson, 1968). Hence, making relevant information semi-public, this means making it public to a restricted group of people, might be an efficient way to reduce agency cost in a syndicate. Further, the effect of exiting the firm to an IPO is negativly related to the likelihood that information is available in VentureSource. Publishing information is a standardized process for capital market transactions. To offer stocks and bonds to the broader public, issuers must fulfill various information duties. For private placements, there is no such standardized process and VCs must reduce transaction cost by voluntarily providing this information. Our sample does not cover any information provided to the public from information duties but rather constitutes a voluntary information allocation. Service providers, such as Dow Jones, might substitute the standardized information provision when investors focus on private placements. Column 2 presents estimates of the hazard of the firm. First, hazard decreases in the funding amount the firm receives. VCs aim to identify promising and unpromising projects to reduce investments to unpromising projects and to expand investments to promising projects. Hence, VCs commit higher amounts to firms that they expect to survive (Agarwal and Gort, 2002; Das et al., 2003; Davidsson, 1991). Second, firm's hazard is higher in seed- and first-financing rounds, and for syndicated deals. Information asymmetry between the VC and the entrepreneur is higher in the early stage of a project. As a consequence, entering a funding relationship is exposed to adverse selection, therefore investments into unpromising firms are more likely in early financing rounds (Berglund and Johansson, 1999). In syndicates, learning performs less efficiently due to free riding among syndicating partners (Casamatta and Haritchabalet, 2007; Gross-

82

.6.

Results

man and Hart, 1980; Hölmstrom, 1982; Pichler and Wilhelm, 2001; Wilson, 1968). Hence, syndicating partners individually provide reduced effort identifying promising and unpromising projects, therefore investments into unpromising firms are more likely. Table 3.3 presents estimates from regressing pre-money valuation on a matrix of learning paths. In estimations (1), (3), (5), the learning paths are simulated for positive informational updates. Alternative informational content posing the reference signal is the incubation stage of the firm. Hence, I is equal to one if the funds are allocated to acceleration or growth activities, and zero otherwise. The reference signal is disclosed in forth- and later-financing rounds, under a low level of information asymmetry. Hence, the information premium comprised in the reference signal is small. I chose to do so, because the impact of positive updates is expected to increase in the information premium, and so the cumulative impact of the reference update on the valuation is small. In contrast, in estimations (2), (4), (5), learning paths are simulated for negative informational updates. Thus, alternative informational content posing the reference signal is the growth stage of the firm. I is equal to one if funds are allocated to incubating or acceleration activities, and zero otherwise. The reference signal is disclosed in seed- and early-financing rounds, approximating for funding conditions w i t h a high level of information asymmetry. Hence, the information premium comprised in the reference signal is high. I chose to do so, because the impact of positive updates is expected to increase in the information premium, and so the cumulative impact of the reference update on the valuation is small as well. In models (1) and (2), I simulate learning based on comprehensive informational updates. In contrast, in models (3) to (6), learning paths are simulated based on uncomprehensive information. Hypothesis H.3.1 states that comprehensive updates have a higher information value than uncomprehensive updates, and hence have a greater effect on firm valuations. First, I look at positive information content of the update. For learning path 1 ( L I ) , firm values increase on average by 168.0% after comprehensive information is disclosed (Ic)> For uncompre-

Chapter

3.

Learning

in Funding

83

Relationships

Table 3.3: Probit Regression (1) and Cox Proportional eling (2): Selection Equations

Hazard Bate Mod-

Regression coefficients reported. Standard Errors in parentheses. P-values *** < 0.01, ** < 0.05, * < 0.10. Funding-Amount =Log (Funding-Amount). Fundingamounts in non-Euro currencies are converted to Euro equivalences by end-of the day exchange rates and deflated to a reference year of 2003 by Eurostat inflation rate. Seed & 1st =1, if the financing round is the first financing round the start-up received, and 0 otherwise. IP0=1 if the start-up exited to an IPO, and 0 otherwise. Firm-Ag e=Log (funding date - founding date of the company). Syndication= 1, if the deal is syndicated, and 0 otherwise. Industry-fix effects are included, industry classifications are Business & Financial Services, Consumer Services, Healthcare, Information Technology, Others. Dependent:

(2)

(1) Funding-Amount Seed & 1 s t IPO Firm-Age Syndication Intercept Indusry-fix Observations Log-Likelihood AIC Likelihood-Ratio Wald

0.169 (0.015) 0.004 (0.041) -1.547 (0.135) 0.085 (0.019) 0.327 (0.045) -3.516 (0.23) yes 5295 -2,943.48

Dependent:

***

-0.161 * * * (0.01) 0.445 * * * (0.031)

*** *** ***

0.27 * * * (0.035)

*** yes 5295 72,319.23 722.03 723.2

hi(Age)

84

.6.

Results

hensive information, the firm values increase on average by 72.5% if the update comprises information about the progress of research and development activities {IR&D), and by 101.5% if the update comprises information about the progress of the sales activities (I S ALE)- For learning path L2, firm values increase on average by 120.1% for / c , and by 69.0%, respectively 64.3%, for I R&D and I s ALE- A l l estimates are significant at the 1% level. To compare estimation coefficients across models, I control for unequal variances of the estimates. I test for the difference in the coefficients applying Welch's unpaired t-test. The confidence interval is 95%. The test shows that the estimate for comprehensive information is larger than the respective estimates for uncomprehensiv information on the 1% significance level. This indicates that the information value of comprehensive information is strictly larger than the information value of uncomprehensive information, for positive informational updates. For negative information content of the update, average firm values decrease by 27.4% after comprehensive information is disclosed (1/2). For uncomprehensive information, firm values decrease on average by 31.3%, respectively 9.5%, for I R & D and Is ALEThe estimates are statistically significant on the 1% level for Lq and for I r & d - Differences in L2 for comprehensive and uncomprehensive information are also statistically significant on the 1% significance level. However, I observe that for I r & D uncomprehensive information has a higher information value than comprehensive information. For L3, firm values decrease on average by 53.5% for I c , and by 51.9%, respectively 45.7% for Ir&D and I s ALE- L3 is statistically significant on the 1% significance level for I c and for I r & d , and on the 10% significance level for I s ALE- Differences in L3 for comprehensive and uncomprehensive information are also statistically significant on the 1% significance level. This indicates that information value of comprehensive information is, in general, larger than the information value of uncomprehensive information, for negative informational updates. Hypothesis H.3.2 states that learning proceeds more efficiently for comprehensive information. I compare the standard deviations of the estimates across models to test this hypothesis. I relate standard deviations to a reference value of 1, which corresponds to a pool-value of risk before an informational update is disclosed. For the learning paths LI and L2,

Chapter

3.

Learning

in Funding

Relationships

85

the standard deviations of the estimates are 0.240, respectively 0.258 for comprehensive information. For uncomprehensive information, standard deviation are strictly smaller for all learning paths. This means that firms' valuations spread wider from one another after comprehensive information is disclosed. In detail, after positive and comprehensive information is disclosed, 95% of the firms are valued at a premium between 120.0% and 216.0%, for learning path LI. For uncomprehensive information, the premium is between 39.7% and 105.3% for updates that comprise information about the progress of the research and development activities (IR&D), and between 67.3% and 135.7% for updates that comprise information about the progress of the sales activities (I S ALE)- In 1/2, 95% of the firms are valued at a premium between 171.7% and 68.5% after comprehensive information is disclosed, and at a premium between 101.0% and 37.0%, respectively 100.1% and 28.5% , after uncomprehensive information is disclosed. Hypotheses H.3.3a and H.3.3b state that the value of informational updates is conditional on the information premium comprised. H.3.3a states that positive informational updates are more valuable when the information premium is high. For learning path L I , information is disclosed in seed- and first-financing rounds, and for learning path L2, information is revealed in second- or third-financing rounds. For comprehensive information and uncomprehensive information, LI is larger than L2 for all estimations. Controlling for unequal variances, differences between LI and L2 are statistically significant on the 1% significance level. The effect of positive updates on the firm value is hence strictly larger if the information is disclosed under high information asymmetry. This indicates that the information premium increases in the level of information asymmetry. H.3.3b states that negative informational updates are less valuable if the information premium is high. For learning path L3, information is disclosed in forth- and later-financing rounds. For comprehensive information and uncomprehensive information, L3 is larger than L2 for all estimations. Controlling for unequal variances, differences between L3 and L2 are statistically significant on the 1% significance level. The effect of negative updates on the firm value is strictly larger if information is disclosed under

86

3.7.

Conclusion

a low level of information asymmetry. This indicates that the information premium increases in the level of information asymmetry. Hypothesis H.3.4 states that learning proceeds less efficiently under asymmetric information. I compare the standard deviations of the coefficients within the models to test this hypothesis. I relate standard deviations to a reference value of 1, which corresponds to a pool-value of risk before an informational update is revealed. For positive informational updates, standard deviations of LI and L2 are very close to each other in all models. Given Ic, standard deviation is 0.240 for LI and 0.258 for L2. Given I R&D, standard deviations are 0.164 and 0.160, and given I s ALE, 0.171 and 0.179. Hence, firm valuations do not spread significantly wider from each other, if positive information is revealed under low information asymmetry. The results indicate that information asymmetry does not impact the efficiency of learning if positive information is disclosed. H.4. is rejected. However, for negative informational updates, the standard deviation of L3 is strictly larger than the standard deviation of L2, in all models. Given I c , standard deviation is 0.157 for L3 and 0.098 for L2. Given I R&D, standard deviations are 0.139 and 0.108, and given Is ALE, 0.250 and 0.120. Hence, firms' valuations spread significantly wider from each other, if negative information is revealed under low information asymmetry. The results indicate that information asymmetry disturbs efficient learning if negative information is disclosed.

3.7

Conclusion

In this chapter, I analyze entrepreneurial firm's planned capital appropriation in a unique data set to estimate the value of an informational update. Thereby, I focus on the relevance of comprehensive information and information asymmetry for efficient learning. I contribute to the literature in that it is the first to analyze learning on the project level empirically. There are two main findings that are important to improve the understanding of learning in funding relationships.

ic

(3)

T+

(4)

T-

(5)

Dependent: Log(Vi,k) SALE

T+

(6)

SALE

T-

(Table continued on next page)

1.680*** 0.725*** 1.015*** (0.240) (0.164) (0.171) L2 1.201 *** -0.274 *** 0.690 *** -0.313 *** 0.643 *** -0.095 (0.258) (0.098) (0.160) (0.108) (0.179) (0.120) L3 -0.535 *** -0.519 *** -0.457 * (0.157) (0.139) (0.25) I -1.490*** -0.024 -0.676*** 0.090 -0.801 *** -0.211 *** (0.232) (0.068) (0.143) (0.073) (0.161) (0.075) A+++ -2.378 *** -1.478 *** -1.768 *** 0.229 0.135 0.164 A++ -1.572 *** 0.58 *** -0.966 *** 0.579 *** -1.061 *** 0.521 *** (0.249) (0.072) (0.131) (0.067) (0.165) (0.064) A+ 1.332 *** 1.328 *** 1.276*** (0.123) (0.119) (0.115)

(2)

T-

in Funding

LI

(1)

Γ+ Ic

Regression coefficients reported. Standard Errors in parentheses. P-values *** < 0.01, ** < 0.05, * < 0.10. L1=I*A+++. L2=I *A++. L3=I*A-\-. 1=1, if the information matches the type of information specified, and 0 otherwise. A+4.+ , A+ indicates the level of information asymmetry, hi is firms individual hazard. IMR is an inverse mill ratio type of control variable. Year- and Industry-fix effects are included, industry classifications are Business & Financial Services, Consumer Services, Healthcare, Information Technology, Others.

Table 3.4·' GLS Regression: Pre-Money Valuation

Chapter 3. Learning Relationships

87

(1)

(2)

le

(3)

yes 1597 0.434

yes

(6)

-0.145 (0.103) 7.790*** (0.694) 12.874*** (0.335) yes

(5)

I SALE

-0.198 * (0.102) 7.319*** (0.674) 14.311 *** (0.394) yes

(4)

I R&D ^fi&D

Observations 1597 1597 1597 1597 1597 B2 0.475 0.435 0.438 0.431 0.447

-0.180 * -0.225 ** (0.099) (0.104) IMR 8.7*** 7.380*** 7.515*** (0.623) (0.688) Intercept 14.600*** 12.862*** (0.372) (0.333) Year-fix yes yes yes yes IndustryJ ^ yes yes yes yes

hi -0.160

le

1

SALE

-0.197 * (0.101) 7.245*** (0.651) 14.344*** (0.387)

Valuation (continued)

Dependent: Log(V^fc)

GLS Regression: Pre-Money

(0.696) 12.956*** (0.337)

(0.103)

88 3.7. Conclusion

Chapter

3.

Learning

in Funding

Relationships

89

First, I prove relevance of comprehensive information for efficient learning. This is, only w i t h precise and unbiased information early identification of promising and unpromising projects is feasible. This is shown as firm values are more affected by comprehensive information than by uncomprehensive information. This is also shown in a higher variance of firm values in a portfolio after comprehensive information about the firms is disclosed. This indicates that firms' prospects are assessed more individually given comprehensive information. In contrast, given imprecise and biased information, firms are valued based on a pool value of risk, indicating that promising and unpromising projects are not identified. To improve their performance over time, VCs must not only base their investment decision on the expected return from the current investment itself, but also on the potential to learn from the investment. This is also shown by Sorensen (2008). This is why the experience of the VC has a large impact his investment behavior. This is shown by Casamatta and Haritchabalet (2007). Also, this is an explanation for prevailing literature finding that VCs highly specialize in their investment scope (Gompers et al., 2009; Norton and Tenenbaum, 1993). Specialization bundles the VC's experience and resources, and increases the speed of learning. Second, I prove disruption of efficient learning through information asymmetry between the entrepreneur and the VC. The results point up that entrepreneurs try to ensure the continuation of unpromising projects for personal reasons. Entrepreneur intentionally try to make it hard for the VC to identify promising and unpromising projects. This is, given discretion to their activities, entrepreneurs manipulate negative information to reduce information value of new information. This is shown as firm values are less affected by negative information, given a high level of information asymmetry. Promising and unpromising projects are thus harder to identify, and learning proceeds less efficiently. This is also shown as the variance of firm values in a portfolio is small after negative information is disclosed, given a high level of information asymmetry. In contrast, entrepreneurs do not have an incentive to behave opportunistically if the firm performs well. Given positive informational updates, entrepreneurs must not fear abandonment of the project and disclose in-

90

3.7.

Conclusion

formation truthfully. This is shown as firm values are more affected by positive information, given a high level of information asymmetry. This indicates that promising and unpromising firms are identified early, and that learning proceeds more efficiently. Consequently, the variance of firm values after new information is disclosed is constantly high for all levels of information asymmetry. This indicates that learning always proceeds efficiently if the prospects of the firm are favorable. To conclude, the results indicate that, for unpromising projects, the information premium is transferred to the entrepreneur in the form of private benefits. The results are consistent w i t h prevailing literature on credit assessments in loan portfolios (Duffie and Landò, 2001). Hence, although a defining characteristic of VCs is their active involvement in their portfolio firms, learning patterns of VCs do not substantially differ from those of traditional intermediaries. This is, VCs receive negative information about their portfolio firms too late.

91

Chapter 4

V C s as I n t e r m e d i a r i e s i n Exits 4.1

Introduction

Funding entrepreneurial start-ups differs substantially from funding established firms. Entrepreneurial start-ups have almost no tangible assets and face high expenditures to set-up their business models, but earn only little revenues. This restricts their access to debt funding, because they cannot provide collateral and have no positive cash flows from their operations to pay interest. Thus, VCs regularly fund entrepreneurial start-ups by equity. Although equity is, in principle, provided for an indefinite period of time, the VC and the entrepreneur lead a temporary relationship (Cumming and Johan, 2010; Cumming and Macintosh, 2001; Sahlman, 1990). As start-ups lack cash flows to pay dividends to their equity investors, divesting their equity stake at a premium is the only source of profit for VCs. Hence, the divesting process is of high relevance for the whole VC industry. If VCs do not earn an adequate return on their divesting strategy, they will have difficulties to raise new capital in the future and will drop out of the market at some point (Gompers and Lerner, 2004).

92

4.1.

Introduction

Like other private investments, shares of entrepreneurial firms are highly illiquid assets. Hence, it is a major issue for the VC to find an appropriate exit vehicle to divest the firm. In general, VCs have two options to divest their portfolio firms. They can either transform their illiquid equity stake in the company into liquid shares through an IPO to sell it to the public, or they can sell the equity stake in total to another private investor in a private transaction. Whether a public or private transaction is preferred relates to a variety of factors, including the ability of new owners to solve information asymmetries and to discipline the entrepreneur (Bienz and Leite, 2008; Cumming and Macintosh, 2003a,b; Schwienbacher, 2005, 2008), the transaction cost of effecting an IPO and ongoing cost of operating as a public firm (Bayar and Chemmanur, 2011; Cumming and Macintosh, 2003b; Lerner, 1994), and the allocation of control rights between the VC and the entrepreneur (Berglöf, 1994; Bascha and Walz, 2001; Cumming, 2008; Lerner, 1994) VCs have interest in exiting their portfolio firms to an IPO because of higher reputation and return prospects. According to the grandstanding hypothesis (Gompers, 1996), VCs benefit from exiting their portfolio firms to an IPO through a reputation effect. Exiting to an IPO, VCs attract high attention in their community which increases investor's participation in follow-on IPOs (Barry et al., 1990; Gompers, 1996; Megginson and Weiss, 1991) and further fund raising from sponsors (Lee and Wahal, 2004). Also, performing an IPO, VCs earn on average highest returns. For example, in a sample of U.K. private equity exits, Nikoskelainen and Wright (2007) show an average IRR of 136.9% for IPOs and 23.0% for trade sales. Also, in a sample of U.S. venture capital investments, Cochrane (2005) finds average log returns are 81% for IPOs and 50% for trade sales. VCs consider potential exit strategies when they initially decide about the funding of a firm (Cumming and Johan, 2008b). However, Cumming and Johan (2008b) find that VCs are quite bad in preplanning their exit strategy. They find that VCs have preplanned an exit strategy via an IPO or a trade sale for 31% of their portfolio firms at the time of the initial funding. But stressing the difference between preplanned and actual exits, they find that only 19% of the actual IPOs were preplanned, and only 36% of the actual acquisitions were preplanned. Nevertheless, Cumming and

Chapter

4.

VCs as Intermediaries

in Exits

93

Johan (2008b) do not provide a discussion about why VCs fail to preplan their exits. However, VCs inherit a central role in exits as they act as intermediaries between their portfolio firms and the new investors. We assume that one reason for failing preplanned exit outcomes is that there is still an inadequate understanding of VCs' intermediary role in the exit transaction. In this role, the VC has to reduce information asymmetries efficiently pre-exit, so that information asymmetry between the entrepreneur and new investors is lower at the exit than it would have been without VC intervention (Baker and Gompers, 2003; Barry et al., 1990; Gompers and Lerner, 2004; Hochberg et al., 2007; Hochberg, 2011; Lerner, 1994; Megginson and Weiss, 1991). Thus, it is our scope to further analyze the role of the VC's intermediation for exit outcomes. In general, the need for strong intermediation by the VC differs in the exit type and matters to a much greater degree for an IPO than for a trade sale (Barry et al., 1990; Cumming and Johan, 2008a; Gompers and Lerner, 2004; Megginson and Weiss, 1991). In a trade sale, the firm is sold in total to a single acquirer, most often a strategic investor. A n acquirer has the ability to carry out a due diligence to reduce information asymmetry and has industry specific knowledge to assess the prospects of the entrepreneurial firm. In contrast, in an IPO, a significant portion of the firm is publicly offered and sold to a disperse group of shareholders. Those have neither the ability to carry out a due diligence, nor the industry specific knowledge to assess the prospects of the firm properly. Hence, VCs who are able to better signal their role as a strong intermediary will be more likely to exit to an IPO. This chapter provides a major contribution to better understand the role of VCs as intermediaries in exit transactions by modeling exit outcomes on an industry-level. So far, industry classification was only used as an explanation for the likelihood to exit to an IPO. For example, Gompers and Lerner (2004) and Cumming and Johan (2008a) argue that life science and technology firms have a higher likelihood to exit to an IPO, because they pose higher growth opportunities and thus IPO investors have higher appetite for those firms. We perform an industry-level analysis and find that the impact of the VC's intermediation on exit outcomes is not only related to the type of exit, but also to further environmental conditions

94

4.2.

Agency

Cost in IPO and Trade Sale Exits

surrounding the exit. We find that certification of the deal by the VC is more important for IPO exits in industries w i t h high IPO rates, namely Business & Financial Services and Healthcare industries. In contrast, in industries w i t h lower IPO rates, we do not find that strong intermediaries can reduce information asymmetry more efficiently.

4.2 4.2.1

A g e n c y C o s t i n I P O a n d T r a d e Sale E x i t s Adverse Selection P r e - E x i t

There are two main sources of asymmetric information that impact exit outcomes: First, ex-ante information asymmetries between the selling VC and the new investors that lead to adverse selection, second, ex-post information asymmetries between the new investors and the entrepreneur that lead to moral hazard. Adverse selection results from the fact that the selling VC had a relationship with the entrepreneur. During the funding process, the VC accumulated information about the entrepreneurial firm that improved his assessment of the firm's business model and technology (Bergemann and Hege, 1998). This information is valuable as it improves the VC's expectation about the prospects of the firm. Since this information is only available to the selling VC inside the funding relationship, the VC has an informational advantage over the new investors that he can capitalize (Akerlof, 1970). Transaction costs related to adverse selection depend on the new investors' access to these information. In an IPO, a significant portion of the firm is publicly offered and sold to a disperse group of shareholders. For those, free riding hampers individual's incentive to gather private information about the start-up company (Grossman and Hart, 1980; Hölmstrom, 1982; Pichler and Wilhelm, 2001; Wilson, 1968). As such, IPO investors are exposed to adverse selection and will demand for a discount in the offering price (Akerlof, 1970; Baron, 1982; Cumming and Macintosh, 2003b; Jensen and Meckling, 1976; Michaely and Shaw, 1994). This dilutes the share of the VC and depresses his return prospects. In contrast, in a trade sale, the entrepreneurial firm is sold

Chapter

4.

VCs as Intermediaries

in Exits

95

in total to a single acquirer. In absence of the free-riding problem, the acquirer has an incentive to carry out a due diligence to gather private information about the firm. A trade sales is thus preferred to minimize transaction cost if the value of private information is larger than the cost of carrying out a due diligence. Further, an IPO and a trade sale differ in the deal structure, which has an effect on the information disclosure. IPOs conduct in the form of an auction and hence the renegotiation of new investors is eliminated. In contrast, trade sales conduct in the form of individual contracts thus information disclosure is maximized through explicit incentives. For example, instead of a one-time payment, the acquirer can propose a performance based payment scheme comprising a fixed and a variable component, where the total payment allocated to the VC increases in the ratio of fixed to variable compensation and in the entrepreneurial firm's quality. This contract design anticipates renegotiation of new investors. In this way, agency cost is minimized from an ex-post perspective because the VC has incentive to disclose the quality of the firm truthfully. This further maximizes the advantage of a trade sales over an IPO when the transaction is exposed to adverse selection risk.

4.2.2

Moral Hazard Post-Exit

Moral hazard results from opportunistic behavior of the entrepreneur. The entrepreneur is operating inside the firm and hence controls the allocation of information to investors (Bergemann and Hege, 1998; Cornelli and Yosha, 2003). This gives the entrepreneur the chance to manipulate information distributed to the investors and to capitalize on an informational advantage. During the funding process, VCs implement several mechanisms like contingent control allocation (Chan et al., 1990; Hellmann, 1998; Kirilenko, 2001), convertible securities (Casamatta, 2003; Cornelli and Yosha, 2003; Repullo and Suarez, 2004; Schmidt, 2003), and stage financing (Bergemann and Hege, 1998; Neher, 1999) to reduce opportunistic behavior of the entrepreneur. Transaction costs related to moral hazard depend on the new owners' ability to efficiently monitor and control the entrepreneur.

96

4.2.

Agency

Cost in IPO and Trade Sale Exits

In an IPO, new shares are issued to a disperse group of shareholders and thus the ownership structure of the company itself becomes more dispersed. Again, free riding among investors hampers individual's incentive to exert effort to monitor the entrepreneur and, as a result, the entrepreneur regains more discretion in managing the firm (Grossman and Hart, 1980; Hölmstrom, 1982; Pichler and Wilhelm, 2001; Wilson, 1968). However, in a trade sale, the acquirer usually purchases the company as a whole and takes full control of the firm. In this case, the acquirer is not exposed to free riding cost and thus has an incentive to engage in close monitoring and control activities to minimize opportunistic behavior of the entrepreneur. The advantage of concentrated ownership over dispersed ownership depends on the effectiveness of the firm's internal corporate governance (Bollitori et al., 2010; Bruton et al., 2009; Hochberg, 2011; Hoque, 2014; Krishnan et al., 2011). A strong corporate governance disciplines managers since it reduces their discretion in managing the company (Hart, 1995; Hoskisson and Turk, 1990; Uhlaner et al., 2007; Van Ees et al., 2009). VCs professionalize start-ups corporate governance modifying the firm's organizational structure and hiring outside directors to re-allocate control (Hellmann and Puri, 2002; Jain and Tabak, 2008; Khanin et al., 2009; Suchard, 2009). In this way, decision committees are build inside the firm and the discretion of the entrepreneur in managing the firm is reduced. The advantage of a trade sale over an IPO to minimize transaction cost hence depends on the ability of the new owners to control the entrepreneur over the effectiveness of the corporate governance system implemented by the VC. Furthermore, an IPO and a trade sale differ in the incentive structure, which impacts the effort of the entrepreneur. Post-IPO, the entrepreneur owns a substantial share in the company and therefore is in control of the firm and participates in its future success, whereas, in a trade sale, control is transferred to the acquirer and the entrepreneur is excluded from further profit sharing (Black and Gilson, 1998). Post-IPO, the entrepreneur and the new investors have the same interest in the company and the entrepreneur has an incentive to expend effort to maximize the shareholder value (Bienz and Leite, 2008). In contrast, post-trade sale, the entrepreneur and the new owner have divergent interests. The entrepreneur has no incentive to maximize the shareholder value, but has an

Chapter

4.

VCs as Intermediaries

in Exits

97

incentive to extract an informational rent from his informational advantage as a manager (Aghion and Bolton, 1992; Chan et al., 1990; Gompers, 1995; Hansen, 1992; Hellmann, 1998; Kirilenko, 2001). The effort of the entrepreneur is especially important if he has specific human capital essential for the future success of the firm (Amit et al., 1998; Burchardt et al., 2014; Hart and Moore, 1994; Neher, 1999). Thus, to maximize the cost advantage of concentrated ownership, the acquirer must have the ability to compensate entrepreneur's human capital post-trade sale.

4.3

Investor Characteristics

In their role as an intermediary between the portfolio firm and the new investors, the VC reduces information asymmetries, so that information asymmetry between the entrepreneur and the new owners is lower than it would have been without the VC's intervention (Baker and Gompers, 2003; Barry et al., 1990; Gompers and Lerner, 2004; Hochberg et al., 2007; Hochberg, 2011; Lerner, 1994; Megginson and Weiss, 1991). To do so, VCs play an active role in monitoring the entrepreneur (Amit et al., 1998; Gompers, 1995; Kaplan and Stromberg, 2001). Monitoring is hence particularly important if information asymmetry is large, for example in early financing rounds when information is based to a large extent on private information of the entrepreneur, and in projects in which entrepreneurs have a high personal interest (Gompers, 1995). To strengthen their role as a monitor, VCs can syndicate their investments to improve their screening and value adding activities (Brander et al., 2002; Casamatta and Haritchabalet, 2007; Hopp and Rieder, 2011; Sah and Stiglitz, 1986). In syndicates, VCs have superior information processing capacity and benefit from the experience of their partners to improve the assessment of information provided by the entrepreneur. This improves the quality of their decisions and hence signals their screening and value adding ability. Furthermore, syndicating also signals the quality of investment opportunities according to the reciprocate theory (Ferrary, 2010; Lockett and Wright, 2001; Seppä and Jääskeläinen, 2002; Sorenson and Stuart, 2001; Wright and Lockett, 2003). This is, lead VCs grant non-lead VCs access to their high quality investment opportuni-

98

4.3.

Investor

Characteristics

ties through syndication. By syndicating out promising investments, VCs expect reciprocate behavior of their syndicating partners. This improves VCs access to high quality investment opportunities in a network. In this way, syndication size signals the lead-VC's status within his community and thus his access to promising investment opportunities. H y p o t h e s i s H . 4 . 1 : Syndication is positively related to the likelihood of exiting to an IPO. VCs have interest in providing the entrepreneurs incentive to expand effort to improve firm's success probability. But as the entrepreneur controls the allocation of information to the investors, he has a certain discretion in managing the firm. The entrepreneur reduces his effort to maximize the shareholder value if he has personal interest in the firm's projects and can maximize his utility from the consumption of private benefits (Bergemann and Hege, 1998; Cornelli and Yosha, 2003; Gompers, 1995). For example, this is the case for research oriented projects that provide the entrepreneur also high reputation in his community. The entrepreneur has an incentive to reduce effort in maximizing shareholders' interest if he expects to earn a higher personal return from trimming it towards the interest of his community than from trimming the research project towards the commercial interest of the shareholders (Gompers, 1995). As a consequence, VCs provide funds continuously over time through a staged capital infusion to provide the entrepreneur an incentive to expand effort to the firm (Bergemann and Hege, 1998; Gompers, 1995). As staging provides the VC the opportunity to abandon the project at each subsequent financing round, it provides the VC power to threaten the entrepreneur with the abandonment of the project. Providing the entrepreneur just as much money as required to meet the next intermediate development stage of the project forces the entrepreneur to expand effort to the interest of the VC to ensure the continuation of the project (Schwienbacher, 2005). In this way, staging signals the role of the VC as an active monitor.

H y p o t h e s i s H . 4 . 2 : Staging is positively related to the likelihood of exiting to an IPO.

Chapter

4.

VCs as Intermediaries

in Exits

99

Effectiveness of a trade sale over an IPO is related to the cost of the new investors to access the private information from the selling VC. Carrying out a due diligence is costly and resource intensive as it takes quite a long time to gather the information, and regularly M & A advisors are hired to do this. In contrast, cost of information production is reduced if the VC and the new owner lead a special relationship and share information on a regular basis. This is the case for corporate VCs (CVCs) (Dushnitsky and Lenox, 2006; Dushnitsky and Shaver, 2009; Gompers and Lerner, 2000a; Maula et al., 2005; Weber and Weber, 2007). CVCs are closely linked to their corporate mothers as the mother inherits a double role as a limited partner and a potential acquirer. There is extensive information sharing between CVCs and their corporate mothers. Thus, effectiveness of exiting a portfolio firm to the corporate mother over exiting a portfolio firm to another acquirer is high. Further, in a trade sale, the concentration of ownership reduces the entrepreneur's incentive to exert effort. In the case of exiting the firm to the corporate mother, the acquirer is assumed to have strategic interest in the entrepreneurial firm and hence to have the ability and industry specific knowledge to compensate the human capital of the entrepreneur. H y p o t h e s i s H . 4 . 3 : CVC affiliation is negatively related to the likelihood of exiting to an IPO. If the new investors cannot compensate the human capital of the entrepreneur, incentive comparability of dispersed ownership over concentrated ownership is related to the effectiveness of the firm's corporate governance. For example, modifying the firm's organizational structure, the VC redistributes control among a group of externally hired managers and, by doing this, the discretion of the entrepreneur is reduces through the establishment of a decision-making committee. Then, outside control is less important to prevent opportunistic behavior of the entrepreneur and the advantage of concentrated ownership to control the entrepreneur mitigates. VCs usually retain a share in the company for a certain period of time post-IPO and hence remain in the firm's supervisory board. It is found that experienced VCs exhibit more active post-IPO involvement in the corporate governance through a higher probability of retaining their

100

4.4.

Firm

Characteristics

shares and board seats than their less experienced peers (Hochberg, 2011; Krishnan et al., 2011). Thus, the VC's experience signals his role as an active monitor post-IPO. H y p o t h e s i s H . 4 . 4 : Lead-investor's experience is positively related to the likelihood of exiting to an IPO.

4.4

F i r m Characteristics

In absence of a strong intermediary, new investors must directly rely on the firm characteristics to mitigate adverse selection. In this case, the availability of public information about the company is an important determinant of exit outcomes. If there is only few information publicly available, a trade sale is beneficial, because then private information can be gathered through a due diligence. In contrast, if information asymmetry between the entrepreneur and the new investors is low, because information about the firm is publicly available, a trade sale only has a small benefit over an IPO, with respect to information gathering. In this way, an IPO becomes more likely as soon as profound information about the prospects of the firm become public. Young firms have a short operating history and therefore no track record of operations is publicly available. In absence of public information, the only way to gather information about an entrepreneurial firm is to enter into a private relationship and to learn about the firm's prospects. In this case, private information is valuable. In contrast, more matured firms have tracked operations. W i t h increasing firm age, promising firms attracted attention in their communities, approving the potential of their new technologies to outside investors. On the other hand, for matured firms that did not attract attention in their communities it is reasonable that their technology or business model is less promising. Hence, with increasing age, information about a firm becomes public and the value of private information is reduced. In this, the informational advantage of private information mitigates and the likelihood to exit to an IPO increases.

Chapter

4.

VCs as Intermediaries

in Exits

101

H y p o t h e s i s H . 4 . 5 : Firm's age is positively related to the likelihood of exiting to an IPO. Traditional valuation approaches, such as DCF methods or multiples approaches, build a common ground for transaction prices based on publicly available information. But, they are no proper method to assess the value of an entrepreneurial start-up, because start-ups entail excessive cost for research and development and do not earn substantial operating profits. Therefore, the VC valuation method differs substantially from the traditional valuation approaches. Usually, the valuation of start-ups is based on a scenario approach, in which the development of the firm in favorable and unfavorable scenarios is forecasted and a potential final exit-value is estimated (Smith et al., 2011). The VC valuation method thus cause high agency cost as estimates are strongly related to private information of the VC and the entrepreneur. In contrast, if a young firm already operates profitable at the time of the exit, valuation by traditional approaches can be applied to reduce asymmetric information. This reduces the need for private information to assess the value of the entrepreneurial firm and the likelihood to exit to an IPO increases. H y p o t h e s i s H . 4 . 6 : Operating profitability is positively related to the likelihood of exiting to an IPO.

4.5 4.5.1

Sample and Methodology Sample

Our sample consists of IPO and trade sale exits from VC-backed entrepreneurial firms provided by Dow Jones VentureSource. We consider exits from firms that received their initial financing round between 2003/01/01 and 2015/12/31 and exited in the same period. We analyze 871 exits from firms based in 15 European countries 1 and 2,493 exits from firms based in the U.S.. Information about the entrepreneurial firm characteristics 1

Belgium, A u s t r i a , Prance, Germany, Greece, Ireland, Italy, Luxembourg, Netherlands, Portugal, Spain, U n i t e d K i n g d o m , Denmark, Finland, Sweden

102

4.5.

Sample and

Methodology

and the deal characteristics is from VentureSource. To analyze differences between industries we use sector classifications provided by Dow Jones VentureSource and aggregate firms into four broader industries: Healthcare, Information Technology, Consumer Services and Business & Financial Services.

4.5.2

Methodology

We use a logistic regression model to predict the probability that the exit is an IPO relative to the probability that the exit is a trade sale. We use an IPO dummy that can take two values for the exits observed. IPO equals 1 if firm % was exited to an IPO, and 0 if firm % was exited to a trade sale. The model is

p(IPOi)

= ßo + ß\X% + €i

(4.1)

X i is a vector of co-variates for firm % and ß\ is the vector of estimated coefficients, e^ is the error term that is assumed to come from an exponential distribution. Because of the non-linear relation between the dependent and independent variables, we report odds-ratios. The odds describe the probability that the exit is an IPO relative to the probability that the exit is a trade sale. The odds-ratio shows in which relation the odds change if the independent variable changes by one unit. We log-transform continuous predictors to the base of 2. So, we can interpret the odds-ratio associated with a doubling of the predictor. We bias-correct our estimation to avoid quasi-complete separation of data points by Firth (1993) penalized maximum likelihood estimation. Data separation is especially an issue in the sub-sample regressions. Further, we use classification trees to identify the value of additional information for the prediction of an IPO exit. Classification trees are commonly used in machine learning approaches for predictive modeling. The advantage of such a type of model is that it uncovers the structure of data and hence provides an insight into data patterns. W i t h a classification tree, input data is split based on a given splitting strategy to cluster the data.

Chapter

4.

VCs as Intermediaries

in Exits

103

We apply splitting based on the chi-square test. This is used for categorical response variables and splits input data by testing for the difference in the response variable. The splitting is done by recursive partitioning, starting w i t h all the observations, which are represented by the node at the top of the tree. At each step, a split is determined by finding the best predictor variable and the best cut point that assigns observations in the parent node to two child nodes. To avoid overfitting of the model, which is a common issue when using a classification tree model, we define the minimum number of firms in a child node to be at least 30 and prune the tree to a total of ten final groups.

4.5.3

C o n t r o l Variables

We control our model for time- and location-fix effects, and additional deal characteristics. Among others, Ritter and Welch (2002) show that IPO activity is time variant. Pastor and Veronesi (2005) demonstrate that IPO waves are caused by high stock prices and decline in expected returns, indicating market timing of issuers. To control for a time variant use of IPOs as exit vehicle, we control our estimations for year-fix effects. Location-fix effects account for the fact that the liquidity of the IPO market varies among regions. For example, capital markets in Europe provide less liquidity to young growth firms compared to the U.S. (Robbie and Mike, 1998; Schwienbacher, 2005). This restricts the use of European capital markets as a potential divesting channel for VCs. For this reason, we keep the European sample and U.S. sample separated in our analysis. We also consider further location-fix effects within our samples. In our European sample, we control for firms being located in the U.K.. In a sample of U.K. and France VC-backed IPOs, Chahine et al. (2007) show that VC-backing reduces IPO underpricing in the U.K., but increase underpricing in France. This indicates that IPOs are a more efficient exit vehicle in the U.K., biasing the choice of exit vehicle for U.K. based portfolio firms. In our U.S. sample we control for firms being located in California. We use the location to account for the Silicon Valley effect. Firms located in the Silicon Valley have access to the largest and most liquid VC market in

104

4.5.

Sample and

Methodology

the U.S. and thereby have better access to the capital market than firms in other regions, which leads to a higher likelihood of completing IPOs (Zhang, 2007). We control for the regional distance between the lead-VC and the portfolio firm. VCs show a strong local bias in their investments (BabcockLumish, 2009; Chen et al., 2010; Cumming and Dai, 2010). This is reasoned by the close involvement of VCs in the business strategy of their portfolio firms (Bottazzi et al., 2008; Bruton et al., 1997; Hellmann and Puri, 2002; Sapienza et al., 1996). We approximate regional distance in the U.S. by the location of the firm's and the lead-VC's headquarters. We accumulate states to four geographic regions as defined by the U.S. Census Bureau (Northeast, Midwest, South, West) and we control whether the VC and the portfolio firm have their headquarters in the same region. In Europe, besides distance, there is an additional factor influencing the VC's involvement, namely, language barriers. Involvement is closely related to barrier-free communication between the VC and the entrepreneur. Thus, for Europe, we control whether the headquarters are located in the same country. At least, we control for the VC's investment duration. Barry et al. (1990) and Megginson and Weiss (1991) demonstrate an impact of the investment duration on the exit outcome. The investment duration of the VC describes the time from his initial investment into the firm to his exit. Cumming and Macintosh (2001) consider that VCs aim to minimize information asymmetries at the time of the exit and hence investment duration varies depending on firm's initial characteristics at the first funding. Further, Giot and Schwienbacher (2007) consider a pecking order in VC exits, which is, VCs exit to a trade sale if they cannot exit to an IPO. To control for investment duration, we control our estimations for the time between the initial funding received by the firm and the exit date.

Chapter

4.

VCs as Intermediaries

in Exits

4.6

Results

4.6.1

Descriptive Statistics

105

Table 4.1 presents means and standard deviations for the European sample, and table 4.2 for the U.S. sample. Overall, 85.0% of the exits in Europe and 89.0% of the exits in the U.S. account for trade sales. Hence, trade sales are the most important exit vehicle in both regions, whereas IPOs are only performed for a few firms. The ratio of IPO to trade sale exits is slightly higher in Europe compared to the U.S.. This is especially surprising since U.S. capital markets are found to be more liquid for young growth firms (Robbie and Mike, 1998; Schwienbacher, 2005). For both samples, the ratio of IPO exited firms to trade sale exited firms is the highest in the Healthcare and Business & Financial Services industries. Splitting the samples based on the exit vehicles, we find that IPO exited firms are in general older, operate more often profitable and received more financing rounds than firms exited to a trade sales. Also, IPO exited firms were backed by larger syndicates and more experienced leadinvestors. However, we find differences in the investor's activity between IPO exited companies in Europe and in the U.S.. IPO exited companies based in the U.S. received more financing rounds and are backed by more experienced investors as well as larger syndicates compared to their European peers. In contrast, firm characteristics of IPO exited companies do not vary significantly in the U.S. and in Europe. Clustering observations on the industry-level, we find differences in investor and firm characteristics of IPO exited firms across industries. In Europe, IPO exited firms in the Healthcare industry are backed by the largest syndicates among all IPO exited firms. IPO exited firms operating in the Healthcare and Business & Financial Services industries were on average backed by more experienced lead-investors than IPO exited firms from other industries. At least, IPO exited firms in the Consumer Services and Information Technology industries were more often profitable than IPO exited firms in other industries.

106

4.6.

Results

In the U.S. sample, IPO exited firms in the Healthcare industry are also backed by the largest syndicates among all IPO exited firms. IPO exited firms in the Business & Financial Services and Consumer Services industries are backed by the most experienced lead-investors among industries. Also, IPO exited firms in the Consumer Services and Information Technology industries are more often operating profitably than IPO exited firms in other industries.

IPO

Trade IPO

Trade IPO

Trade

Business Consumer Services Services

in parentheses.

871

131

13

121

(Table continued on next page)

740

14

162

VCs as Intermediaries

Observations

3.50 2.32 2.85 1.93 2.50 2.10 (1.89) (2.72) (1.65) (1.52) (1.38) (1.7) (1.36) Staging 3.35 4.02 3.23 4.23 2.81 3.86 3.04 (1.61) (2.12) (1.47) (2.09) (1.09) (2.96) (1.24) Lead-CVC 0.16 0.07 0.18 0.00 0.17 0.07 0.19 (0.37) (0.25) (0.38) (0.00) (0.37) (0.27) (0.39) Lead-Experience 14.82 17.65 14.31 25.31 12.95 18.93 16.54 (14.72) (13.72) (14.84) (18.29) (9.92) (19.56) (21.97) Firm-Age 7.77 7.71 7.78 6.66 7.83 7.14 6.68 (6.64) (5.36) (6.84) (4.16) (5.28) (2.70) (9.13) Profitable 0.13 0.18 0.12 0.23 0.15 0.43 0.07 (0.34) (0.39) (0.33) (0.44) (0.36) (0.51) (0.25) Location 0.32 0.21 0.34 0.23 0.40 0.29 0.30 (0.47) (0.41) (0.47) (0.44) (0.49) (0.47) (0.46) Same Region 0.81 0.82 0.80 0.85 0.78 0.79 0.76 (0.40) (0.39) (0.40) (0.38) (0.42) (0.43) (0.43) Duration 4.47 4.60 4.45 4.14 3.99 4.84 3.64 (2.51) (2.66) (2.48) (2.22) (2.30) (2.55) (2.07)

Syndication 2.49

Financial

Standard errors

4.

all

Means reported.

Table J^.l: Summary Statistics , European Sample

Chapter in Exits

107

Observations

Duration

Same Region

Location

Profitable

Firm-Age

Lead-Experience

Lead-CVC

Staging

Syndication

Trade IPO

Information . . Technology Trade IPO Trade

Others

61

107

21

292

22

58

3.21 2.81 2.36 2.55 1.83 (3.09) (2.24) (2.54) (1.56) (2.06) (1.42) 4.28 3.89 4.14 3.38 3.14 2.71 (2.08) (1.82) (2.37) (1.54) (1.04) (1.12) 0.03 0.21 0.19 0.17 0.09 0.16 (0.18) (0.41) (0.40) (0.37) (0.29) (0.37) 17.75 14.35 14.48 13.85 15.05 13.24 (12.52) (15.43) (11.54) (11.66) (10.36) (11.25) 7.34 7.64 10.76 8.18 6.78 8.93 (4.08) (3.97) (9.82) (4.89) (3.54) (12.49) 0.05 0.06 0.48 0.18 0.09 0.05 (0.22) (0.23) (0.51) (0.39) (0.29) (0.22) 0.16 0.41 0.24 0.29 0.27 0.47 (0.37) (0.49) (0.44) (0.46) (0.46) (0.50) 0.82 0.80 0.86 0.83 0.77 0.84 (0.39) (0.40) (0.36) (0.37) (0.43) (0.37) 4.82 5.4 4.77 4.77 3.94 4.25 (2.80) (2.63) (2.99) (2.58) (2.31) (2.31)

4.46

IPO

TT ι,-ι Healthcare

Summary Statistics , European Sample (continued)

108 4.6. Results

0.37

Lead-CVC

0.05

0.48

Profitable

Location

4.46

Observations 2493

Duration

Same Region

6.59

Firm-Age

Trade IPO

Trade IPO

Trade

275

34

406

27

(Table continued on next page)

2218

450

6.08 3.22 4.91 3.05 5.56 3.06 (2.36) (3.38) (1.99) (2.67) (1.86) (3.42) (1.89) 6.44 3.93 6.15 3.84 7.07 3.47 (2.17) (2.70) (1.93) (2.72) (1.67) (4.06) (1.52) 0.20 0.39 0.15 0.38 0.22 0.35 (0.48) (0.40) (0.49) (0.36) (0.48) (0.42) (0.48) 20.67 27.32 19.85 30.06 20.75 32.89 18.51 (18.78) (20.39) (18.41) (13.75) (16.99) (18.66) (16.70) 8.14 6.39 8.50 6.72 8.44 4.95 (4.00) (3.71) (4.00) (3.39) (4.36) (3.42) (3.15) 0.12 0.05 0.18 0.07 0.37 0.02 (0.23) (0.32) (0.21) (0.39) (0.26) (0.49) (0.15) 0.45 0.49 0.50 0.44 0.37 0.57 (0.50) (0.50) (0.50) (0.51) (0.5) (0.49) (0.50) 0.75 0.73 0.76 0.82 0.77 0.70 0.76 (0.43) (0.44) (0.43) (0.39) (0.42) (0.47) (0.43) 6.10 4.26 6.05 4.25 5.74 3.45 (2.53) (2.46) (2.46) (1.96) (2.41) (2.21) (2.07)

IPO

Business Consumer Services Services

in parentheses.

VCs as Intermediaries

Lead-Experience

4.21

Staging

Syndication 3.54

Financial

Standard errors

4.

all

Means reported.

Table 4-2: Summary Statistics , U.S. Sample

Chapter in Exits

10

0.21

Lead-CVC

0.39 0.70 6.08

Location

Same Region

Duration 160

0.04

Profitable

Observations

7.98

Firm-Age

25.93

6.30

Staging

Lead-Experience

6.55

Syndication

Trade IPO

Trade IPO

Information . . Technology Trade

Al1

Others

342

39

935

15

89

3.59 5.51 3.22 6.13 3.51 (3.62) (2.25) (2.62) (1.88) (3.18) (2.67) 4.45 6.38 3.96 7.60 4.42 (2.44) (2.21) (2.03) (1.94) (3.62) (2.81) 0.36 0.15 0.43 0.33 0.26 (0.41) (0.48) (0.37) (0.50) (0.49) (0.44) 21.94 27.54 19.10 25.53 22.34 (19.57) (22.63) (19.30) (16.56) (39.75) (29.32) 7.44 8.36 6.43 7.88 7.80 (4.10) (4.21) (2.47) (3.86) (3.33) (4.62) 0.05 0.23 0.05 0.07 0.06 (0.19) (0.22) (0.43) (0.21) (0.26) (0.23) 0.34 0.69 0.53 0.60 0.36 (0.49) (0.48) (0.47) (0.50) (0.51) (0.48) 0.71 0.79 0.78 0.73 0.63 (0.46) (0.46) (0.41) (0.42) (0.46) (0.49) 5.05 6.84 4.33 5.22 4.66 (2.68) (2.57) (2.28) (2.48) (1.42) (2.69)

IPO

Healthcare

TT U1

Summary Statistics , U.S. Sample (continued)

10 4.6. Results

Chapter

4.6.2

4.

VCs as Intermediaries

in Exits

111

Multivariate Analysis

In this section, we test H.4.1 to H.4.6 on the industry-level to analyze how the impact of investor and firm characteristics varies across industries. W i t h this approach we aim to identify differences in the intermediary role of the VC across industries. Table 4.3 presents results from the European samples. Model (1) is the baseline model, exit outcomes are predicted using all observations. In the baseline model, investor characteristics impact the exit outcome. Hypotheses H.4.1 to H.4.4 are confirmed. The estimates are statistically significant at least on the 10% significance level. In detail, a doubling of the syndication size increases odds to exit to an IPO by 37.0%. For staging, odds to exit to an IPO increase by 30.2% if the number of financing rounds received doubles. A doubling in the experience of the lead-investor increases odds to exit to an IPO by 18.9%. Further, if a CVC was a leadinvestor at least in one financing round, the likelihood to exit to a trade sale instead of an IPO is 4.0 times larger compared to entrepreneurial firms not affiliated with a lead-CVC. Considering firm characteristics, hypothesis H.4.5 is rejected. The firm's age is not related to exit outcomes. However, H.4.6 is confirmed. Operating profitable at the time of the exit increases odds to exit to an IPO by 3.3 times. The estimate is statistically significant on the 1% significance level. In models (2) to (6), exit outcomes are predicted on an industry-level. In the Business & Financial Services industry, investor characteristics impact the exit outcome. H.4.1 to H.4.4 are confirmed. The estimates are statistically significant at least on the 10% significance level. A doubling in the syndication size increases odds to exit to an IPO by 2.2 times. A doubling in the number of financing rounds received increases odds to exit to an IPO by 57.9%, and a doubling in the lead-investor's experience increases odds by 94.5%. CVC affiliation increases the likelihood to exit to a trade sale instead of an IPO by 12.2 times. Firm characteristics do not impact exit outcomes in the Business & Financial Services industry. H.4.5 and H.4.6 are rejected.

112

4.6.

Results

In the Consumer Services industry, investor characteristics do not impact the exit outcome. H.4.1 to H.4.4 are rejected. However, firm characteristics impact exit outcomes. H.4.6 is confirmed. The effect is statistically significant on the 1% significance level. Odds to exit to an IPO increase by 6.9 time if the entrepreneurial firm is operating profitable at the time of the exit. The estimate is statistically significant on the 1% significance level. In the Healthcare industry, investor characteristics impact exit outcomes. Hypothesis H.4.1, H.4.3 and H.4.4 are confirmed. The estimates are statistically significant at least on the 5% significance level. A doubling in the syndication size increases odds to exit to an IPO by 72.6%, and a doubling in the lead-investor's experience by 41.1%. CVC affiliation of the company increases the likelihood to exit to a trade sale instead of an IPO by 9.2 times. Firm characteristics do not impact exit outcomes in the Healthcare industry. H.4.5 and H.4.6 are rejected. In the Information Technology industry, both, investor characteristics and firm characteristics, impact exit outcomes. However, the impact of investor characteristics is limited to CVC affiliation. The estimate is statistically significant on the 1% significance level, but the effect is economically small as CVC affiliation increases the likelihood to exit to a trade sale only by 11.6%. However, hypotheses H.4.5 and H.4.6 are confirmed. The estimates are statistically significant at least on the 5% significance level. A doubling in the firm's age increase odds to exit to an IPO by 77.5%, and odds increase by 5.1 times if the firm is operating profitable at the time of the exit. Summarizing the results from table 4.3, we find that either investor characteristics or firm characteristics are helpful in explaining exit outcomes across industries in Europe. Further, the results indicate that exit outcomes are related to investor characteristics in industries w i t h higher IPO rates, namely Business & Financial Services and Healthcare industries. In industries with low IPO rates, namely Consumer Services and Information Technology, investor characteristics are found to have weak impact on the exit outcome. Hence, we conclude that the need for intermediation is related to the number of potential investment opportunities that IPO

Chapter

4.

VCs as Intermediaries

in Exits

113

investors face. Search and screening cost of IPO investors increase in the number of potential investment opportunities and hence there is a severe need for signaling to minimize the transaction cost. VCs thus signal their high ability as intermediaries by syndicating and staging their investments. W i t h less IPO conducts, intermediation is less important, because search and screening cost of IPO investors are low. But, w i t h less IPO conducts, investors face difficulties in the pricing of new issues, because only few peer transactions exist. Thus, valuation risk is a major source of agency cost in this case. The application of traditional valuation methodsis the only way to reduce asymmetric information based on publicly available information. This is only possible if the firm operates profitable at the time of the exit. Table 4.4 presents results from the U.S. samples. Model (1) is the baseline model, exit outcomes are predicted using all observations. In the baseline model, investor characteristics impact exit outcomes. Hypotheses H.4.1 to H.4.4 are confirmed. The estimates are statistically significant at least on the 5% significance level. In detail, doubling the syndication size increases odds to exit to an IPO by 2.1 times, and a doubling in the number of financing rounds received increases the odds by 4.3 time. For staging and syndication activities the effects estimated in the U.S. sample are noticeably larger compared to the effects estimated in the European sample. Further, a doubling of lead-investor's experience in the VC market increases odds to exit to an IPO by 16.1%. CVC affiliation increases the likelihood to exit to a trade sale instead of an IPO by 4.3 times. The effects of lead-investor's experience and CVC affiliation are roughly the same in magnitude compared to the effects estimated in the European sample. Considering firm characteristics, hypothesis H.4.5 is rejected. Firm's age is not related to exit outcomes. However, hypothesis H.4.6 is confirmed. Firm's profitability is related to IPO exit outcomes. The estimate is statistically significant on the 1% significance level. Odds to exit to an IPO increase by 5.5 times if the firm operates profitable at the time of the exit. This effect is larger in magnitude compared to the effect estimated in the European sample. In models (2) to (6), exit outcomes are predicted on an industry-level. Investor characteristics impact exit outcomes in all industries. H.4.1 to H.4.4 are confirmed for all industries, except for Healthcare. The estimates

114

4.6.

Results

are statistically significant at least on the 5% significance level. The effect on syndication is nearly stable across industries. The odds to exit to an IPO increases in a range between 83.5% and 117.5 % if the syndication size doubles. However, the magnitude of the other effects varies across industries. For firm characteristics, H.4.5 is rejected for all industries. However, H.4.6 is confirmed for all industries, except for the Healthcare industry. The estimates are statistically significant at least on the 5% significance level. Since there is an impact of investor and firm characteristics across most industries, we compare the industry-level effects to the effects estimated in the baseline model to analyze whether the sensitivity of exit outcomes to the characteristics changes relative to the baseline model. To control for unequal variances of the estimates, we test for the difference in the regression coefficients applying Welch's unpaired t-test. In the Business & Financial Services industry, changing sensitivity to investor characteristics is ambiguous. The effect of staging is smaller in magnitude compared to the baseline model. The regression coefficients are different from each other on the 1% significance level. A doubling of the number of financing rounds received only increases odds to exit to an IPO by 2.9 times. In contrast, the effect of lead-investor experience is larger in magnitude compared to the baseline model. The regression coefficients are different from each other on the 1% significance level. A doubling in lead-investor's experience increases odds to exit to an IPO by 87.5%. For firm characteristics, sensitivity of exit outcomes to firm characteristics decreases. The effect of profitability on the exit outcome is smaller in magnitude compared to the baseline model. Operating profitable at the time of exit increases odds to exit to an IPO only by 3.4 times. Also, regression coefficients are different from each other on the 1% significance level. In the Consumer Services industry, sensitivity of exit outcomes to investor characteristics increases. The effect related to staging increases in magnitude relative to the effects estimated in the baseline model. A doubling in the number of financing rounds increases odds to exit to an IPO by 5.1 times. The estimates are different from each other on the 1% sig-

Chapter

4.

VCs as Intermediaries

in Exits

115

nificance level. Although odds strongly increase for investor experience, the difference is not statistically significant. Sensitivity of exit outcomes to firm characteristics also increases. The effect of the firm's profitability is larger in magnitude compared to the effect estimated in the baseline model. The odds to exit to an IPO increase by 10.6 times for firms operating profitable at the time of the exit. The difference is significant on the 1% level. In the Healtchcare industry, sensitivity of exit outcomes to investor characteristics decreases. The effect related to staging decreases in magnitude relative to the effect estimated in the baseline model. A doubling in the number of financing rounds received increases odds to exit to an IPO by only 3.2 times. The difference is significant on the 1% level. Furthermore, hypothesis H.4.3 is rejected. Lead-investor's experience is not related to exit outcomes. Also, exit outcomes are not related to investor characteristics. H.4.5 and H.4.6 are rejected. In the Information Technology industry, sensitivity of exit outcomes to investor characteristics does not differ from the baseline model. However, sensitivity of exit outcomes to firm characteristics increases. The effect of firm's profitability is larger in magnitude compared to the effect estimated in the baseline model. The odds to exit to an IPO increases by 9.9 times for firms operating profitable at the time of the exit. The difference is significant on the 1% level. Summarizing the results from table 4.4, we find that both, investor and firm characteristics, are helpful in explaining the exit outcomes across industries in the U.S.. Further, the results indicate that exit outcomes are more sensitive to changes in investor characteristics in industries with higher IPO rates, namely Business & Financial Services and Healthcare industries. In industries with low IPO rates, namely Consumer Services and Information Technology, exit outcomes are less sensitive to changes in investor characteristics. However, in industries w i t h low IPO rates, exit outcomes show increased sensitivity to changes in the firm's profitability. The results are similar to the results in Europe and confirm the assumption that the need for strong intermediaries to successfully perform an IPO is related to the search and screening costs IPO investors face.

116

4.6.3

4.6.

Results

Classification Tree Analysis

Figures 4.1 to 4.6 show the decision trees from applying classification tree analysis. We report the results from the first three splits applied. The trees can be understood as follows: The node on the top of the tree is the parent node, including all observations. The splitting criteria is quoted below the node. The splitting criteria splits observations of the parent node into two child nodes, where the child node labeled 1 contains observations that fulfill the splitting criteria, and the child node labeled 0 contains observations that reject the splitting criteria. Splits are applied descending to their impact on the response variable, based on the chi-square criterion. Information given in the node is the fraction of IPO exits (labeled 1) in the sub-sample, and the residual fraction of trade sale exits (labeled 0). The aim of this analysis is to analyze which characteristics efficiently reduce agency concerns of IPO investors. For this purpose, we aggregate observations from the Business & Financial Services and Healthcare industries to observations from industries w i t h high IPO rates, and Consumer Services and Information Technology industries to observations from industries with low IPO rates. In industries with high IPO rates, exit outcomes are found to be more sensitive to investor characteristics in the multivariate analysis. In contrast, in industries with low IPO rates, exit outcomes are found to be more sensitive to firm characteristics. We thus expect that VC intermediation is a potent mechanism to mitigate agency concerns of IPO investors in industries with high IPO rates, whereas firm characteristics are potent to mitigate agency concerns in industries with low IPO rates. This expectation results in two suppositions (S.4.1 and S.4.2). First (S.4.1), in industries with high IPO rates, investor characteristics are the most important splitting criteria to separate IPO exited firms from trade sale exited firms. In contrast, in industries with low IPO rates, firm characteristics are the most important splitting criteria. Hence, we expect the first split applied in industries w i t h high IPO rates to be related to investor characteristics, and the first split applied in industries with low IPO rates to be related to firm characteristics.

1.189*

(3)

0.461 3.105 1.082 1.039 0.771 no no

1.945**

no

1.411 **

0.109***

1.726***

(5)

(6)

0.874

7.6

46 7

128.0

80 26 4

72.3 1Q

2

1.775** 3.486 0.296*** 0.524 0.350***

0.620

0.843

1.629

^ , Others

1.112

1.075 1.715 0.553

Information , . Technology

1.068 1.210 0.896***

„ Healthcare

1.418 1.198 6.972*** 0.760 5.111 *** 1.028 0.293 *** 0.793 1.485 1.141 0.786 0.911 0.358*** no no no no no

0.969

1.421

(4)

Consumer . Services

0

168 313 188.8

no

16 4

2.205* 0.872 1.579* 0.082* 0.347

(2)

Observations 871 134 176 AIC 657.1 64.3 75.8 Likelihood Ra- igg ß 26 5 tio Wald 128.6 17.2 15.3 30.3 23.9

Firm-Age 1.161 Profitable 3.290*** Location 0.408*** Same Region 0.937 Duration 0.548*** Year-fix yes a no Industry-fix yes a

Ji/XDerience

Ìf ad".

(1)

1.370** 1.302*** 0.251 ***

Business Financial c . Services

4. VCs as Intermediaries

Syndication Staging Lead-CVC

n

all

Dependent: IPO

Odds-ratios reported. Chi-squared ρ-values *** < 0.01, ** < 0.05, *< 0.1 ,a < 0.01 based on Type-3 effect analysis. Syndication=log 2 (maximum number of investor in a syndicate). Staging=log 2 (number of financing rounds). LeadVC=1, if a CVC was lead-investor in afinancing round, 0 otherwise. Lead-Experience=log 2 (most matured leadinvestor affiliated with the start-up). Firm-ag e=log 2 (exit date - founding date of the company), Profitable= 1, if company operates profitable at the exit date, 0 otherwise.

Table J^.S: Logistic Regression: European Sample

Chapter in Exits

117

U.S. Sample

(1)

2.060*** 4.278*** 0.234***

(3)

1.835 ** 2.175 ** 2.938** 0.145***

(2)

Business Financial Services

Observations AIC 1636.6 Likelihood Ratio Wald 358.0

49.9

92.4

67.3

2

64.8

13.7

137

971

98 8

103 22

8

8.109

5.970*** 0.671

4.6.

41.1

ß41

2493 440 477 502 211.0 181.4 588.4 297.2 ? 59^ 88 7

0.769

0.728 9.907***

1.723 4.884*** 0.128***

(6)

Information ^ , . Others Technology

2.072 *** 3.199*** 0.296***

(5)

„ Healthcare

2.136 *** 5.076*** 0.225**

(4)

Consumer ~ . Services

ïf ad". 1.161 ** 1.875 *** 1.542 * 1.057 1.383 * 0.543 ** Ji/Xperience Firm-Age 1.114 1.368 2.350* 0.803 1.773 Profitable 5.517*** 3.445** 10.608*** 1.657 Location 1.120 0.822 0.540 0.922 2.027* 1.661 Same Region 0.655 ** 0.860 0.485 0.609 ** 0.634 Duration 0.670** 0.826 0.500 0.923 0.583 0.720 Year-fix yes a no no no no no Industry-fix yes a no no no no no

Syndication Staging Lead-CVC

„ all

Dependent: IPO

Odds-ratios reported. Chi- squared p-values *** < O.Ol, ** < 0.05, *< 0.1 , a < O.Ol based on Type-3 effect analysis. Syndication=log2 (maximum number of investor in a syndicate). Staging=log 2 (number of financing rounds). LeadVC=1, if a CVC was lead-investor in a financing round, 0 otherwise. Lead-Experience=log2(most matured leadinvestor affiliated with the start-up). Firm-ag e=log 2 (exit date - founding date of the company), Profitable= 1, if company operates profitable at the exit date, 0 otherwise.

Table 4-4 : Logistic Regression:

118 Results

Chapter

4.

VCs as Intermediaries

in Exits

119

Second (S.4.2), if strong intermediaries are a valuable signal and efficiently mitigate information asymmetries in industries with high IPO rates, we expect that information about firm characteristics has low value, given strong VC-backing. This is based on the assumption that strong intermediaries certify firms quality in this case. In contrast, if strong intermediaries are no valuable signal to efficiently mitigate asymmetric information in industries with low IPO rates, information about VC backing has low value, given strong firm characteristics. We calculate the value of additional information as the absolute difference between the IPO ratios of two child nodes (in percentage points) relative to the IPO ratio of the parent node. The European sample is analyzed in figures 4.1 to 4.3. We illustrate the likelihood to exit to an IPO for the full sample in figure 4.1. The likelihood to exit to an IPO instead of a trade sale is 42.3% if the entrepreneurial firm is backed by more than four investors, and 13.3% otherwise. Further, if backed by more than four investors and the lead-investor has more than 14 years of experience in the venture capital market, the likelihood to exit to an IPO instead of a trade sale is largest at 51.2%. The criteria are met by 43 firms in our sample. The value of information about syndication size in the first node is 1.11. In the right branch of the tree, representing firms backed by strong intermediaries, additional information in the second node has a value of 0.44. In the left branch of the tree, in absence of strong intermediaries, additional information has a value of 2.3. Hence, the value of additional information is larger in absence of strong intermediaries. This indicates that information asymmetry is more severe in absence of strong intermediaries and hence that certification mitigates information asymmetries in the full sample. Figure 4.2 shows the classification tree for industries with low IPO rates, which is Consumer Services and Information Technology industries. In this tree, the firm's profitability is the most important splitting criteria. Hence, S.4.1 is confirmed. The likelihood to exit to an IPO is 20.3% for firms operating profitable, and 4.8% otherwise. The value of this information is 2.14. For firms operating profitable, additional information has a value of 0.65. For unprofitable firms, the value of additional information is 1.81.

120

4.6.

Results

Hence, the value of additional information is larger in absence of strong firm characteristics. This indicates that information asymmetry is more severe in absence of strong firm characteristics and hence that firm characteristics mitigate information asymmetries in industries with low IPO rates. Hence, S.4.2 is confirmed. Figure 4.3 shows the classification tree for industries with high IPO rates, which is Healthcare and Business & Financial Services industries. In this tree, syndication is the most important splitting criteria. Hence, S.4.1 is confirmed. The likelihood to exit to an IPO is 38.8% for firms backed by a syndicate, and 10.9% otherwise. The information value on syndication of investors is 1.42. For firms backed by a syndicate, the value of additional information is 0.67. For firms not backed by a syndicate, the value of additional information is 1.48. Hence, the value of additional information is larger in absence of strong intermediaries, hence intermediation mitigates information asymmetries in industries w i t h high IPO rates. Also, S.4.2 is confirmed. The U.S. sample is analyzed in figures 4.4 to 4.6. We illustrate the likelihood to exit to an IPO for the full sample in figure 4.4. For entrepreneurial firms backed by more than six investors and by lead-investors w i t h more than 25 years of experience in the venture capital market, the likelihood to exit to an IPO is 71.8%. The criteria are met by 71 firms in our sample. The information value of syndication is 3.07. The value of additional information is 0.61 if the firm is backed by strong intermediaries, and 2.29 otherwise. This shows that additional information is more valuable in the absence of strong intermediation. This also indicates that information asymmetry is more severe in absence of strong intermediaries and hence that intermediation mitigates information asymmetries in the full sample. Figure 4.5 shows the classification tree for industries with low IPO rates, which is also Consumer Services and Information Technology industries in the U.S.. In this tree, syndication is the most important splitting criteria. Hence, S.4.1 is rejected. The way splits are applied in this tree does not allow us to calculate the value of additional information. After the first split is applied, firms backed by strong syndicates are not further split. We can not test S.4.2

Chapter

4.

VCs as Intermediaries

in Exits

121

Figure 4.6 shows the classification tree for industries with high IPO rates, which is Healthcare and Business & Financial Services industries. In this tree, syndication is the most important splitting criteria. Hence, S.4.1 is confirmed. The likelihood to exit to an IPO is 31.9% for firms backed by a syndicate, and 17.1% otherwise. The value of information about syndication is 0.60. For firms backed by strong intermediaries, the value of additional information is 1.09. For firms backed by weak intermediaries, the value of additional information is 1.73. Thus, the value of additional information is larger in the absence of strong intermediaries, hence intermediation mitigates information asymmetries in industries with high IPO rates. S.4.2 is confirmed.

122

4.6.

Node Ν 0

Results

0 732 0.833

1 0.167 0 0.833 Syndication > 4

0 Node Ν 0

Γ "

1 1 651 0.866

0.133 1 0 0.866 — Experience > 9.1 —

0

Experience > 14.3η

Ί1 Iι

Node Ν 0

3 326 0.902

Node Ν 0

1 0

0.098 0.902

0.169 1 0.831 0 Profitable > 1

Node Ν 0

7 72 0.847

Node Ν 0

8 254 0.917

Node Ν 0

9 272 0.849

Node Ν 0

A 53 0.736

Node Ν 0

5 38 0.658

Node Ν 1

6 43 0.512

1 0

0.153 0.847

1 0

0.083 0.917

1 0

0.151 0.849

1 0

0.264 0.736

1 0

0.342 0.658

1 0

0.512 0.488

4 325 0.831

Figure 4-1: CART: European Sample Predictor variables and the cut points reported. Recursive partitioning of splits. Splitting is applied based on the chi-square test. Models are fit to minimum number of firms in a child node=30, pruning of trees=10. Child node labeled 1: observations that fulfill the splitting criteria. Child node labeled 0: observations that reject the splitting criteria. Information about sub-samples given in the node: 1 fraction of IPO exits, 0 residual fraction of trade sale exits.

Chapter

4.

VCs as Intermediaries

Node Ν 0 1 0

Node Ν 0

123

in Exits

Node Ν 0

0 405 0.926

1 0

0.074 0.926

2 336 0.952

Node Ν 0

0.048 0.952 Staging > 4

2 69 0.797

0.203 1 0 0.797 Syndication > 1 -

3 287 0.965

1 0.035 0 0.965 ρ Experience > 5.3 η

I 0 Node

I 1

0

0.925

Node Ν 0

1 0

0.075 0.925

1 0

Ν

7

80

1 8 207 0.981

Node Ν 0

0.019 0.981

0

1

0

1

4 49 0.878

Node Ν 0

5 35 0.857

Node Ν 0

6 34 0.735

0.122

1 0

0.143 0.857

1 0

0.265 0.735

0.878

Figure 4-2: CART : Low IPO Rates , European Sample Predictor variables and the cut points reported. Recursive partitioning of splits. Splitting is applied based on the chi-square test. Models are fit to minimum number of firms in a child node=30, pruning of trees=10. Child node labeled 1: observations that fulfill the splitting criteria. Child node labeled 0: observations that reject the splitting criteria. Information about sub-samples given in the node: 1 fraction of IPO exits , 0 residual fraction of trade sale exits.

124

4.6.

- Firm Age > 3.8 -

Experience > 8.7

Node Ν 0

4 72 0.942

1 0

0.058 0.942

3 32 0.781

1 0

0.219 0.781

Node Ν

0 1 0

7 37

1 0 1

Ί

Node Ν 0

6 109 0.532

0.468 1 0 0.532 • Firm Age > 5.4 -

ι- Experience > 9.5 η

Node Ν 0

Results

Node Ν 0

8 32 0.875

Node Ν 0

1 0

0.125 0.875

0

1

5 48 0.792

Node Ν 1

9 31 0.581

Node Ν 0

A 78 0.577

0.208

1 0

0.581 0.419

1 0

0.423 0.577

0.792

Figure 4-3: CART: High IPO Rates , European Sample Predictor variables and the cut points reported. Recursive partitioning of splits. Splitting is applied based on the chi-square test. Models are fit to minimum number of firms in a child node=30, pruning of trees=10. Child node labeled 1: observations that fulfill the splitting criteria. Child node labeled 0: observations that reject the splitting criteria. Information about sub-samples given in the node: 1 fraction of IPO exits , 0 residual fraction of trade sale exits.

Chapter

4.

VCs as Intermediaries

125

in Exits

Node Ν

0 1577

0

0.861

1

0.139

0

0.861

Syndication > 6

Node Ν 0 1 0

Node Ν 0

1 1419 0.901

Node Ν 1

0.099 0.901 Staging > 5

1 0.538 0 0.462 Lead-Exp. > 25.6

3 1174 0.945

Node Ν 0

0.055 1 0 0.945 Profitable >

2 158 0.538

1 0 I

5 87 0.609

0.391 0.609 Staging > 4

Node Ν 0 1 0

6 71 0.718

0.718 0.282 Staging > 6

I

Figure 44:

CART: U.S. Sample

Predictor variables and the cut points reported. Recursive partitioning of splits. Splitting is applied based on the chi-square test. Models are fit to minimum number of firms in a child node=30, pruning of trees=10. Child node labeled 1: observations that fulfill the splitting criteria. Child node labeled 0: observations that reject the splitting criteria. Information about sub-samples given in the node: 1 fraction of IPO exits , 0 residual fraction of trade sale exits.

126

4.6.

Node Ν 0

Results

0 876 0.938

1 0.062 0 0.938 Syndication > 6

Node Ν 0

1 825 0.956

0.043 1 0 0.956 Profitable > 1

0 Node Ν 0 1 0 ρ

Node Ν 0 1

0

3 786 0.968

0.032 0.968 Staging >5 -j

5 668 0.988

Node Ν 0

6 118 0.856

Node Ν 0

0.012

1 0

0.144 0.856

0

0.988

1

4 39 0.718

Node Ν 0

2 51 0.647

0.282

1 0

0.353 0.647

0.718

Figure 4.5: CART: Low IPO Rates , U.S. Sample Predictor variables and the cut points reported. Recursive partitioning of splits. Splitting is applied based on the chi-square test. Models are fit to minimum number of firms in a child node=30, pruning of trees=10. Child node labeled 1: observations that fulfill the splitting criteria. Child node labeled 0: observations that reject the splitting criteria. Information about sub-samples given in the node: 1 fraction of IPO exits, 0 residual fraction of trade sale exits.

Chapter

4.

VCs as Intermediaries

127

in Exits

Node Ν 0

0 625 0.752

1 0.248 0 0.752 Syndication > 6

Node Ν 0 1 0

Node Ν 0

Node Ν 1

1 531 0.829

0.171 0.829 Staging > 4

0.319 1 0 0.681 Lead-Exp. > 25.6η

3 1174 0.945

Node Ν 0

0.055 1 0 0.945 Lead-Exp. > 34.8-|

2 158 0.681

4 180 0.650

0.350 1 0 0.650 r-Lead-Exp. > 10.2-

1

I

ί Node Ν 0

7 312 0.936

Node Ν 0

8 39 0.795

Node Ν 0

1 0

0.064 0.936

1 0

0.205 0.795

0

1

9 31 0.774

Node Ν 0

A 149 0.624

Node Ν 1

5 50 0.520

Node Ν 1

6 44 0.864

0.282

1 0

0.376 0.624

1 0

0.480 0.520

1 0

0.136 0.864

0.774

Figure 4.6: CART: High IPO Rates , U.S. Sample Predictor variables and the cut points reported. Recursive partitioning of splits. Splitting is applied based on the chi-square test. Models are fit to minimum number of firms in a child node=30, pruning of trees=10. Child node labeled 1: observations that fulfill the splitting criteria. Child node labeled 0: observations that reject the splitting criteria. Information about sub-samples given in the node: 1 fraction of IPO exits , 0 residual fraction of trade sale exits.

128

4.7

4.7.

Conclusion

Conclusion

In exit transactions, VCs inherit an intermediary role between their portfolio firms and new investors to reduce information asymmetry and to minimize transaction cost. In this role, VCs signal their role as strong intermediaries through their experience, staging activity and syndication policy. The need for certification of the deal by the VC differs conditional on the type of transaction and is much more pronounced in IPOs than in trade sale exits. This is because IPO investors do not have the ability and the knowledge to assess the prospects of the firm properly and thus a strong intermediary is required to certify the transaction (Barry et al., 1990; Cumming and Johan, 2008a; Gompers and Lerner, 2004; Megginson and Weiss, 1991). In this chapter, we extend the research on the role of the VC's intermediation in exit transactions through an industry-level analysis. So far, no industry-level analysis of exit outcomes is performed. We find that the need for certification by the VC significantly differs across industries and is related to the industry's ratio of IPO to trade sale exits. Certification is much more important in industries w i t h high IPO rates, where IPO investors face high search and screening cost from a large number of potential investment opportunities. In these industries, VCs reduce information cost of IPO investors by signaling their role as a strong intermediary. The results further indicate that intermediation is less important in industries with low IPO rates, hence in absence of severe search and screening cost. Instead, we find that valuation uncertainty is a major source of concern in these industries, indicating that investors have difficulties to validate the pricing of an IPO if there is no representative peer group of IPO conducts. The results are consistent for exits of European and U.S. portfolio firms. We show that the role of the VC as an intermediary in exit transactions not only differs conditional on the type of transaction, but also on various external factors related to the transaction. This improves the understanding of VCs' role as intermediaries in exit transactions and provides scope for further research in this area. We identify IPO competition to be one external factor yearning for certification. We can think of a wide range of further factors to be analyzed in this context, including market condi-

Chapter

4.

VCs as Intermediaries

in Exits

129

tions, firm characteristics and popularity of the entrepreneur. Further, in this chapter, we address the question of what gives rise to different exit outcomes, but we do not seek to understand the role of the VC for underpricing in IPOs. However, these questions are in part related, as the VC tries to minimize transaction cost of the exit, and hence the likelihood to exit to an IPO will be related to expected underpricing. This is another probable investigation: according to our findings, we expect the impact of intermedation on underpricing of VC-backed IPOs to vary conditional on the need for certification by the VC.

131

Chapter 5

Conclusion 5.1

S u m m a r y of the M a i n Results

In this thesis, I analyze the VC's role as a monitoring agent and a certifying agent in the relationship with the entrepreneur. I focus on the impact of asymmetric information on optimal investments, learning, and exit outcomes in this context. The results provide some interesting new implications w i t h regard to the VC-entrepreneur relationship. In the second chapter, the impact of general market risk on the incentive of the entrepreneur is analyzed. It is formally shown that highly qualified entrepreneurs are more likely to leave a project than less qualified entrepreneurs if the project is treated by market risk. This is due to a higher relative value of private benefits if the entrepreneur can realize a high return on his human capital. As a consequence, VCs optimally accelerate investments to projects with a high relevance of the entrepreneur's human capital for the success of the project, relative to firms with a low relevance of the entrepreneur's human capital, if market risk is high. As a result, VCs allocate relatively more funds to projects w i t h a high relevance of the entrepreneur's human capital if market risk is high.

132

5.2.

Outlook

In the third chapter of the thesis, it is shown that precise and unbiased information about an entrepreneurial firm is required to efficiently learn about its prospects over time. However, learning is disrupted if this information exists, but is asymmetrically distributed between the VC and the entrepreneur. If entrepreneurs fear that the VC abandons the project because of bad prospects, entrepreneurs are found to manipulate negative information to reduce its information value. This makes it hard for the VC to identify the quality of the project early, and the likelihood of continuation of unpromising projects increases. Usually, continuation is in the interest of the entrepreneur (Cornelli and Yosha, 2003). In this way, the information premium earned by the VC in the case of efficient learning is transferred to the entrepreneur. In the forth chapter, it is shown that the need for the VC certifying an IPO exit is conditional on the exit conditions. Certifying an IPO is much more important in industries with high IPO rates, where IPO investors face high search and screening cost from a large number of potential investment opportunities. Therefore, only VCs that can signal their role as a strong intermediary can successfully exit their portfolio firms to IPOs in competitive IPO markets. However, in industries with low IPO rates, IPO investors have difficulties to validate the pricing of VC-backed IPOs. Therefore, in industries with low IPO rates, VCs can only successfully exit their portfolio firms to IPOs if the firm characteristics give rise to the application of traditional valuation approaches.

5.2

Outlook

The main results of my thesis provide scope for further research. In the second chapter, it is shown that interdependence of market risk and agency risk motivates the VC to accelerate investments to projects w i t h a high relevance of entrepreneur's human capital, relative to firm w i t h a low relevance of the entrepreneur's human capital, if market conditions worsen. Future research can extend this initial idea by analyzing the impact of market risk on VCs' portfolio weights. Given the results of my thesis, I expect that VCs' portfolio weights increase in firms with a high relevance

Chapter

5.

Conclusion

133

of the entrepreneur's human capital if market conditions worsen. Further, time-variant ageny risk motivates the time-variant use of explicit contractual claims. Given the results of my thesis, I expect a time-variant use of explicit contractual claims to increase entrepreneur's effort incentive with respect to certain environmental conditions. Furthermore, the results of my thesis suggest that the VC provides incentive to the entrepreneur by expanding the project through a follow-on investment. However, VCs also provide a wide range of non-monetary support to their portfolio firms (Gorman and Sahlman, 1989). Non-monetary support may be even more valuable than monetary support in difficult times. Hence, I expect that VCs increase their non-monetary support to firms w i t h a high relevance of the entrepreneur's human capital, relative to firms with a low relevance of the entrepreneur's human capital, when market conditions worsen. In the third chapter, it is found that learning patterns of VCs do, in principal, not substantially differ from those of other intermediaries. However, I have no comparative basis to figure out whether VCs in fact learn more efficiently about the prospects of a firm than other intermediaries. For this purpose, it is necessary to analyze failed projects more in detail. In general, I define efficient learning by the time needed to distinguish between a promising and an unpromising project. To analyze learning efficiency more in detail, one can compare the time to abandonment of initially similar projects if they are backed by different kinds of intermediaries. However, this requires detailed and private information from different types of intermediaries. This information will be almost impossible to access. In the forth chapter, it is shown that the need for certification by the VC in IPO exits is conditional on the exit conditions. There are two possible strands of research that can emerge from this chapter. First, only certification by the VC is analyzed. Further, I can think of certification by the entrepreneur as a complement or substitute of certification by the VC. Given the results of this chapter, I expect that certification by the entrepreneur is also more important in competitive IPO markets. Second, the exit conditions in my thesis only refer to the IPO competition within an industry. I can think of further conditions that require for strong certification by the VC, e.g. aggregate market uncertainty, or complexity of pricing the business model. Finally, the chapter addresses the question of

134

5.2.

Outlook

how certification impacts the exit outcome given different exit conditions. But, the role of certification for underpricing in IPOs given different exit conditions is not addressed. However, these questions are in part related as VCs try to minimize transaction cost of the exit, and the likelihood to exit to an IPO will be related to the expected underpricing. W i t h respect to the results of this chapter, I expect that certification by the VC impacts underpricing more in competitive IPO markets.

135

Bibliography Agarwal, R. and M. Gort (2002). Firm and product life cycles and firm survival. The American Economic Review 92(2), 184-190. Aghion, P. and P. Bolton (1992). A n incomplete contracts approach to financial contracting. The Review of Economic Studies 59 ( 3), 473-494. Akerlof, G. A. (1970). The market for "lemons": Quality uncertainty and the market mechanism. The Quarterly Journal of Economics, 488-500. Amer nie, J. (1984). A look at agency theory for the novice: Part i. CAmagazine , November 90, 97. Amit, R., J. Brander, and C. Zott (1998). Why do venture capital firms exist? theory and Canadian evidence. Journal of Business Venturing 13(6), 441-466. Amit, R., L. Glosten, and E. Muller (1990). Entrepreneurial ability, venture investments, and risk sharing. Management Science 36( 10), 1233-1246. Babcock-Lumish, T. L. (2009). Financing clusters of innovation: the geography of venture capital investment, US and UK. Working Paper. Baker, M. and P. A. Gompers (2003). The determinants of board structure at the initial public offering. The Journal of Law and Economics J^6( 2), 569-598. Baron, D. P. (1982). A model of the demand for investment banking advising and distribution services for new issues. The Journal of Finance 37(4), 955-976.

136

Bibliography

Barry, C. Β., C. J. Muscarella, J. W. Peavy, and M. R. Vetsuypens (1990). The role of venture capital in the creation of public companies: Evidence from the going-public process. Journal of Financial Economics 27(2), 447-471. Bascha, A. and U. Walz (2001). Convertible securities and optimal exit decisions in venture capital finance. Journal of Corporate Finance 7(3), 285-306. Bayar, O. and T. J. Chemmanur (2011). Ipos versus acquisitions and the valuation premium puzzle: A theory of exit choice by entrepreneurs and venture capitalists. Journal of Financial and Quantitative Analysis 46(6), 1755-1793. Bergemann, D. and U. Hege (1998). Venture capital financing, moral hazard, and learning. Journal of Banking & Finance 22(6), 703-735. Bergemann, D. and U. Hege (2005). The financing of innovation: Learning and stopping. BAND Journal of Economics, 719-752. Berger, A. N. et al. (1999). The "big picture" about relationship-based finance. In Federal B,eserve Bank of Chicago Proceedings , Number 761. Berglöf, E. (1994). A control theory of venture capital finance. Journal of Law, Economics , & Organization, 247-267. Berglund, T. and E. Johansson (1999). The entrepreneur's initial contact w i t h a venture capitalist: Why good projects may choose to wait. Swedish School of Economics and Business Administration Working Paper . Bienz, C. and T. E. Leite (2008). A pecking order of venture capital exits. Working Paper. Black, B. S. and R. J. Gilson (1998). Venture capital and the structure of capital markets: Banks versus stock markets. Journal of Financial Economics 47(3), 243-277. Boot, A. W. (2000). Relationship banking: What do we know? of Financial Intermediation 0(1), 7-25.

Journal

Bibliography

Bottazzi, L., M. Da Rin, and T. Hellmann (2008). Who are the active investors?: Evidence from venture capital. Journal of Financial Economics 89(3), 488-512. Boulton, T. J., S. B. Smart, and C. J. Zutter (2010). Ipo underpricing and international corporate governance. Journal of International Business Studies 41(2), 206-222. Brander, J. Α., R. Amit, and W. Antweiler (2002). Venture-capital syndication: Improved venture selection vs. the value-added hypothesis. Journal of Economics & Management Strategy 11(3), 423-452. Brush, C. G. and R. Chaganti (1999). Businesses without glamour? A n analysis of resources on performance by size and age in small service and retail firms. Journal of business venturing 14 ( 3), 233-257. Bruton, G., V. Fried, and R. D. Hisrich (1997). Venture capitalist and CEO dismissal. Entrepreneur ship: Theory and Practice 21(3), 41-55. Bruton, G. D., S. Chahine, and I. Filatotchev (2009). Founders, private equity investors, and underpricing in entrepreneurial IPOs. Entrepreneurship: Theory and Practice 33(A), 909-928. Burchardt, J., U. Hommel, D. S. Kamuriwo, and C. Billitteri (2014). Venture capital contracting in theory and practice: implications for entrepreneurship research. Entrepreneur ship: Theory and Practice . Campello, M. and R. Matta (2010). How are venture capitalists rewarded? The economics of venture capital partnerships. Working Paper. Casamatta, C. (2003). Financing and advising: Optimal financial contracts with venture capitalists. The Journal of Finance 58(5), 2059-2086. Casamatta, C. and C. Haritchabalet (2007). Experience, screening and syndication in venture capital investments. Journal of Financial Intermediation 16(3), 368-398. Chahine, S., I. Filatotchev, and M. Wright (2007). Venture capitalists, business angels, and performance of entrepreneurial IPOs in the UK and France. Journal of Business Finance & Accounting 34 ( 3-4), 505-528.

138

Bibliography

Chan, Y.-S. (1983). On the positive role of financial intermediation in allocation of venture capital in a market w i t h imperfect information. The Journal of Finance 38( 5), 1543-1568. Chan, Y.-S., D. Siegel, and Α. V. Thakor (1990). Learning, corporate control and performance requirements in venture capital contracts. International Economic B,eview , 365-381. Chandler, G. N. and S. H. Hanks (1994). Founder competence, the environment, and venture performance. Entrepreneur ship: Theory and Practice 18 ( 3), 77-90. Chen, H., P. Gompers, A. Kovner, and J. Lerner (2010). Buy local? The geography of venture capital. Journal of Urban Economics 67(1), 90102. Cherif, M. and K. Gazdar (2011). What drives venture capital investments in Europe? New results from a panel data analysis. Journal of Applied Business and Economics 12(3), 122-139. Choi, Y. R. and D. A. Shepherd (2004). Entrepreneurs' decisions to exploit opportunities. Journal of Management 30(3), 377-395. Chung, J.-W., B. A. Sensoy, L. Stern, and M. S. Weisbach (2012). Pay for performance from future fund flows: The case of private equity. The B,eview of Financial Studies 25(11), 3259-3304. Cochrane, J. H. (2005). The risk and return of venture capital. Journal of Financial Economics 75(1), 3-52. Colombo, M. G. and L. Grilli (2010). On growth drivers of high-tech startups: Exploring the role of founders' human capital and venture capital. Journal of Business Venturing 25(6), 610-626. Contai, C. and J. O'Quigley (1999). A n application of changepoint methods in studying the effect of age on survival in breast cancer. Computational Statistics & Data Analysis 30(3), 253-270. Cornelli, F. and O. Yosha (2003). Stage financing and the role of convertible securities. The B,eview of Economic Studies 70(1), 1-32.

Bibliography

Covin, J. G., D. P. S levin, and T. J. Covin (1990). Content and performance of growth-seeking strategies: A comparison of small firms in high-and low technology industries. Journal of Business Venturing 5(6), 391-412. Cox, J. C., S. A. Ross, and M. Rubinstein (1979). Option pricing: A simplified approach. Journal of Financial Economics 7(3), 229-263. Cumming, D. (2008). Contracts and exits in venture capital finance. The Review of Financial Studies 21 ( 5), 1947-1982. Cumming, D. and N. Dai (2010). Local bias in venture capital investments. Journal of Empirical Finance 17 ( 3), 362-380. Cumming, D. and S. Johan (2008a). Information asymmetries, agency costs and venture capital exit outcomes. Venture Capital: An International Journal of Entrepreneurial Finance 10( 3), 197-231. Cumming, D. and S. Johan (2008b). Preplanned exit strategies in venture capital. European Economic Review 52(1), 1209-1241. Cumming, D. and S. Johan (2010). Venture capital investment duration. Journal of Small Business Management 48 ( 2), 228-257. Cumming, D., D. Schmidt, and U. Walz (2010). Legality and venture capital governance around the world. Journal of Business Venturing 25(1), 54-72. Cumming, D. J. and J. G. Macintosh (2001). Venture capital investment duration in Canada and the United States. Journal of Multinational Financial Management 11(A), 445-463. Cumming, D. J. and J. G. Macintosh (2003a). A cross-country comparison of full and partial venture capital exits. Journal of Banking & Finance 27(3), 511-548. Cumming, D. J. and J. G. Macintosh (2003b). Venture-capital exits in Canada and the United States. The University of Toronto Law Journal 53(2), 101-199.

140

Bibliography

Das, S., M. Jagannathan, and A. Sarin (2003). Private equity returns: A n empirical examination of the exit of venture-backed companies. Journal of Investment Management 1( 1), 1-26. Davidsson, P. (1991). Continued entrepreneurship: Ability, need, and opportunity as determinants of small firm growth. Journal of Business Venturing 6(6), 405-429. Davila, Α., G. Foster, and M. Gupta (2003). Venture capital financing and the growth of startup firms. Journal of Business Venturing 18(6), 689-708. Denis, D. J. (2004). Entrepreneurial finance: an overview of the issues and evidence. Journal of Corporate Finance 10(2), 301-326. Dewatripont, M. and J. Tirole (1994). A theory of debt and equity: Diversity of securities and manager-shareholder congruence. The Quarterly Journal of Economics, 1027-1054. Dixit, A. K. and R. S. Pindyck (1994). Princeton university press.

Investment

under uncertainty.

Duffie, D. and D. Landò (2001). Term structures of credit spreads with incomplete accounting information. Econometrica 69(3), 633-664. Dushnitsky, G. and M. J. Lenox (2006). When does corporate venture capital investment create firm value? Journal of Business Venturing 21 (6), 753-772. Dushnitsky, G. and J. M. Shaver (2009). Limitations to interorganizational knowledge acquisition: The paradox of corporate venture capital. Strategic Management Journal 30 ( 10), 1045-1064. Eisenhardt, Κ. M. (1989). Agency theory: A n assessment and review. Academy of Management Review 14( 1), 57-74. Fama, E. F. and M. C. Jensen (1983). Separation of ownership and control. The Journal of Law & Economics 26(2), 301-325.

Bibliography

Félix, E. G. S., C. P. Pires, and M. A. Gulamhussen (2013). The determinants of venture capital in Europe - evidence across countries. Journal of Financial Services Research 44 ( 3), 259-279. Ferrary, M. (2010). Syndication of venture capital investment: The art of resource pooling. Entrepreneur ship: Theory and Practice 34 ( 5), 885907. Firth, D. (1993). Bias reduction of maximum likelihood estimates. Biometrika 80(1), 27-38. Folta, T. Β. (1998). Governance and uncertainty: The trade-off between administrative control and commitment. Strategic Management Journal , 1007-1028. Fried, V. H. and R. D. Hisrich (1994). Toward a model of venture capital investment decision making. Financial Management , 28-37. Giot, P. and A. Schwienbacher (2007). IPOs, trade sales and liquidations: Modelling venture capital exits using survival analysis. Journal of Banking & Finance 31(3), 679-702. Goldfarb, B., D. Kirsch, and D. A. Miller (2007). Was there too little entry during the dot com era? Journal of Financial Economics 86(1), 100-144. Gompers, P., A. Kovner, and J. Lerner (2009). Specialization and success: Evidence from venture capital. Journal of Economics & Management Strategy 18(3), 817-844. Gompers, P., A. Kovner, J. Lerner, and D. Scharfstein (2008). Venture capital investment cycles: The impact of public markets. Journal of Financial Economics 87(1), 1-23. Gompers, P. and J. Lerner (1996). The use of covenants: A n empirical analysis of venture partnership agreements. The Journal of Law and Economics 39(2), 463-498. Gompers, P. and J. Lerner (1999a). A n analysis of compensation in the US venture capital partnership. Journal of Financial Economics 51(1), 3-44.

142 Gompers, P. and J. Lerner (1999b). Press.

Bibliography

The Venture Capital Cycle.

MIT

Gompers, P. and J. Lerner (2000a). The determinants of corporate venture capital success: Organizational structure, incentives, and complementarities. In Concentrated corporate ownership , pp. 17-54. University of Chicago Press. Gompers, P. and J. Lerner (2000b). Money chasing deals? The impact of fund inflows on private equity valuation. Journal of Financial Economics 55(2), 281-325. Gompers, P. A. (1995). Optimal investment, monitoring, and the staging of venture capital. The Journal of Finance 50( 5), 1461-1489. Gompers, P. A. (1996). Grandstanding in the venture capital industry. Journal of Financial Economics 42(1), 133-156. Gompers, P. A. and J. Lerner (2004). The venture capital cycle. M I T press. Gorman, M. and W. A. Sahlman (1989). What do venture capitalists do? Journal of Business Venturing 4( 4), 231-248. Grossman, S. J. and O. D. Hart (1980). Takeover bids, the free-rider problem, and the theory of the corporation. The Bell Journal of Economics, 42-64. Hand, J. R. (2007). Determinants of the round-to-round returns to preIPO venture capital investments in us biotechnology companies. Journal of Business Venturing 22(1), 1-28. Hansen, E. (1992). Venture capital contracts with conversion option, reserve bank of new Zealand. Technical report, Discussion Paper 24. Harris, M. and A. Raviv (1979). Optimal incentive contracts w i t h imperfect information. Journal of Economic Theory 20(2), 231-259. Hart, O. (1995). Corporate governance: Some theory and implications. The Economic Journal 105 ( 430), 678-689.

Bibliography

Hart, Ο. and J. Moore (1994). Debt and seniority: A n analysis of the role of hard claims in constraining management. Technical report, National Bureau of Economic Research. Heckman, J. J. (1979). Sample selection bias as a specification error. Econometrica ^7(1), 153-161. Hellmann, T. (1998). The allocation of control rights in venture capital contracts. The Rand Journal of Economics, 57-76. Hellmann, T. and M. Puri (2002). Venture capital and the professionalization of start-up firms: Empirical evidence. The Journal of Finance 57(1), 169-197. Hochberg, Y. V. (2011). Venture capital and corporate governance in the newly public firm. Review of Finance 16(2), 429-480. Hochberg, Y. V., A. Ljungqvist, and Y. Lu (2007). Whom you know matters: Venture capital networks and investment performance. The Journal of Finance 62(1), 251-301. Hochberg, Y. V., A. Ljungqvist, and Y. Lu (2010). Networking as a barrier to entry and the competitive supply of venture capital. The Journal of Finance 65(3), 829-859. Hölmstrom, Β. (1979). Moral hazard and observability. The Bell Journal of Economics, 74-91. Hölmstrom, Β. (1982). Moral hazard in teams. The Bell Journal of Economics, 324-340. Hölmstrom, Β. and J. Tirole (1997). Financial intermediation, loanable funds, and the real sector, the Quarterly Journal of Economics 112(3), 663-691. Hopp, C. and F. Rieder (2011). What drives venture capital syndication? Applied Economics 43(23), 3089-3102. Hoppe, H. C. (2000). Second-mover advantages in the strategic adoption of new technology under uncertainty. International Journal of Industrial Organization 18(2), 315-338.

144

Bibliography

Hoque, H. (2014). Role of asymmetric information and moral hazard on IPO underpricing and lockup. Journal of International Financial Markets, Institutions and Money 30, 81-105. Hoskisson, R. E. and T. A. Turk (1990). Corporate restructuring: Governance and control limits of the internal capital market. Academy of Management Review 15 ( 3), 459-477. Houben, E. (2002). Venture capital, double-sided adverse selection, and double-sided moral hazard. Working Paper. Hsu, Y.-W. (2010). Staging of venture capital investment: A real options analysis. Small Business Economics 35 ( 3), 265-281. Inderst, R. and H. M. Müller (2004). The effect of capital market characteristics on the value of start-up firms. Journal of Financial Economics 72(2), 319-356. Innés, R. D. (1990). Limited liability and incentive contracting with exante action choices. Journal of Economic Theory 52(1), 45-67. Jain, B. A. and F. Tabak (2008). Factors influencing the choice between founder versus non-founder CEOs for IPO firms. Journal of Business Venturing 23(1), 21-45. Jeng, L. A. and P. C. Wells (2000). The determinants of venture capital funding: evidence across countries. Journal of Corporate Finance 6(3), 241-289. Jensen, M. C. and W. H. Meckling (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics 3(4), 305-360. Kaplan, S. N. and P. Stromberg (2001). Venture capitalists as principals: Contracting, screening, and monitoring. Technical report, National Bureau of Economic Research. Kaplan, S. N. and P. Stromberg (2003). Financial contracting theory meets the real world: A n empirical analysis of venture capital contracts. The Review of Economic Studies 70(2), 281-315.

Bibliography

Khanin, D., J. R. Baum, O. Turel, and R. V. Mahto (2009). Are some venture capitalists more likely than others to replace founder-CEOs? The Journal of Private Equity 12(2), 19-29. Kirilenko, A. A. (2001). Valuation and control in venture finance. The Journal of Finance 56(2), 565-587. Krishnan, C., V. I. Ivanov, R. W. Masulis, and A. K. Singh (2011). Venture capital reputation, post-IPO performance, and corporate governance. Journal of Financial and Quantitative Analysis J^6( 5), 1295-1333. Krohmer, P., R. Lauterbach, and V. Calanog (2009). The bright and dark side of staging: Investment performance and the varying motivations of private equity firms. Journal of Banking & Finance 33(9), 1597-1609. Lakonishok, J., A. Shleifer, R. Thaler, and R. Vishny (1991). Window dressing by pension fund managers. Technical report, National Bureau of Economic Research. Lee, P. M. and S. Wahal (2004). Grandstanding, certification and the underpricing of venture capital backed IPOs. Journal of Financial Economics 73(2), 375-407. Lerner, J. (1994). The syndication of venture capital investments. Financial Management , 16-27. Lerner, J. and A. Schoar (2004). The illiquidity puzzle: Theory and evidence from private equity. Journal of Financial Economics 72(1), 3-40. Li, Y. (2008). Duration analysis of venture capital staging: A real options perspective. Journal of Business Venturing 23(b), 497-512. Lieberman, M. B. and D. B. Montgomery (1988). First-mover advantages. Strategic Management Journal 9(SI), 41-58. Linnainmaa, J. (2006). Learning from experience. Technical report, working paper, University of Chicago. Lockett, A. and M. Wright (2001). The syndication of venture capital investments. Omega 29(5), 375-390.

146

Bibliography

Lundvall, Β. and Β. Johnson (1994). The learning economy. The Learning Economy and the Economics of Hope, 107. March, J. G. (1991). Exploration and exploitation in organizational learning. Organization Science 2(1), 71-87. Maula, M., E. Autio, and G. Murray (2005). Corporate venture capitalists and independent venture capitalists: What do they know, who do they know and should entrepreneurs care? Venture Capital: An International Journal of Entrepreneurial Finance 7(1), 3-21. McDonald, R. and D. Siegel (1986). The value of waiting to invest. The Quarterly Journal of Economics 101 ( 4), 707-727. McDougall, P. P., R, B. Robinson, and A. S. DeNisi (1992). Modeling new venture performance: A n analysis of new venture strategy, industry structure, and venture origin. Journal of Business Venturing 7(4), 267289. McGrath, R. G. (1997). A real options logic for initiating technology positioning investments. Academy of Management Review 22(4), 974996. McGrath, R, G., W. J. Ferrier, and A. L. Mendelow (2004). Real options as engines of choice and heterogeneity. Academy of Management Review 29(1), 86-101. Megginson, W. L. and K. A. Weiss (1991). Venture capitalist certification in initial public offerings. The Journal of Finance J^6( 3), 879-903. Metrick, A. and A. Yasuda (2010). The economics of private equity funds. The Review of Financial Studies 23(6), 2303-2341. Metrick, A. and A. Yasuda (2011). Venture capital and other private equity: A survey. European Financial Management 17(4), 619-654. Michaely, R. and W. H. Shaw (1994). The pricing of initial public offerings: Tests of adverse-selection and signaling theories. The Review of Financial Studies 7(2), 279-319.

Bibliography

Neher, D. V. (1999). Staged financing: A n agency perspective. The Review of Economic Studies 66(2), 255-274. Nikoskelainen, E. and M. Wright (2007). The impact of corporate governance mechanisms on value increase in leveraged buyouts. Journal of Corporate Finance 13(A) , 511-537. Norton, E. and Β. H. Tenenbaum (1993). Specialization versus diversification as a venture capital investment strategy. Journal of Business Venturing 8(b) , 431-442. Pastor, L. and P. Veronesi (2005). Rational IPO waves. The Journal of Finance 60(A) , 1713-1757. Phalippou, L. and O. Gottschalg (2008). The performance of private equity funds. The Review of Financial Studies 22(A) , 1747-1776. Pichler, P. and W. Wilhelm (2001). A theory of the syndicate: follows function. The Journal of Finance 56(6), 2237-2264.

Form

Pindyck, R. S. (1993). Investments of uncertain cost. Journal of Financial Economics 34( 1), 53-76. Reinganum, J. F. (1985). A two-stage model of research and development with endogenous second-mover advantages. International Journal of Industrial Organization 3(3), 275-292. Repullo, R. and J. Suarez (2004). Venture capital finance: A security design approach. Review of Finance £(1), 75-108. Ritter, J. R. and I. Welch (2002). A review of IPO activity, pricing, and allocations. The Journal of Finance 57(A), 1795-1828. Robbie, W. and K. Mike (1998). Venture capital and private equity: A review and synthesis. Journal of Business Finance & Accounting 25 (56), 521-570. Robinson, W. T., G. Kalyanaram, and G. L. Urban (1994). First-mover advantages from pioneering new markets: A survey of empirical evidence. Review of Industrial Organization 9(1), 1-23.

148

Bibliography

Sah, R. Κ. and J. E. Stiglitz (1986). The architecture of economic systems: Hierarchies and polyarchies. American Economic Review 76(A). Sahlman, W. A. (1990). The structure and governance of venture-capital organizations. Journal of Financial Economics 27( 2), 473-521. Sapienza, H. J., S. Manigart, and W. Vermeir (1996). Venture capitalist governance and value added in four countries. Journal of Business Venturing 11(6), 439-469. Schmidt, Κ. M. (2003). Convertible securities and venture capital finance. The Journal of Finance 58(3), 1139-1166. Schwienbacher, A. (2005). A n empirical analysis of venture capital exits in Europe and the United States. Working Paper. Schwienbacher, A. (2008). Innovation and venture capital exits. The Economic Journal 118(533), 1888-1916. Seppä, T. and M. Jääskeläinen (2002). How the rich become richer in venture capital: Firm performance and position in syndication networks. Frontiers of Entrepreneur ship Research 26. Shankar, V., G. S. Carpenter, and L. Krishnamurthi (1998). Late mover advantage: How innovative late entrants outsell pioneers. Journal of Marketing Research, 54-70. Shrader, R. and D. S. Siegel (2007). Assessing the relationship between human capital and firm performance: Evidence from technology-based new ventures. Entrepreneur ship: Theory and Practice 31(6), 893-908. Smith, J., R. L. Smith, R. Smith, and R. Bliss (2011). Entrepreneurial finance: strategy , valuation , and deal structure. Stanford University Press. Sorensen, M. (2008). Learning by investing: Evidence from venture capital. Working Paper. Sorenson, O. and T. E. Stuart (2001). Syndication networks and the spatial distribution of venture capital investments. American Journal of Sociology 106(6), 1546-1588.

Bibliography

Steeb, A. (2017). Profitability und Exit Strategie: Eine zweidimensionale Analyse europäischer VC-finanzierter Unternehmen. B.Sc. Thesis, University of Hohenheim. Suarez, F. F. and G. Lanzolla (2007). The role of environmental dynamics in building a first mover advantage theory. Academy of Management Review 32 ( 2), 377-392. Suchard, J.-A. (2009). The impact of venture capital backing on the corporate governance of australian initial public offerings. Journal of Banking & Finance 33 ( 4), 765-774. Trigeorgis, L. (1996). Real options : Managerial flexibility resource allocation. M I T press.

and strategy in

Ueda, M. (2004). Banks versus venture capital: Project evaluation, screening, and expropriation. The Journal of Finance 59(2), 601-621. Uhlaner, L., M. Wright, and M. Huse (2007). Private firms and corporate governance: A n integrated economic and management perspective. Small Business Economics 29(3), 225-241. Van Ees, H., J. Gabrielsson, and M. Huse (2009). Toward a behavioral theory of boards and corporate governance. Corporate Governance: An International Review 17(3), 307-319. Van Osnabrugge, M. (2000). A comparison of business angel and venture capitalist investment procedures: A n agency theory-based analysis. Venture Capital: An international Journal of Entrepreneurial Finance 2(2), 91-109. Weber, B. and C. Weber (2007). Corporate venture capital as a means of radical innovation: Relational fit, social capital, and knowledge transfer. Journal of Engineering and Technology Management 24( 1), 11-35. Wernerfeit, B. and A. Karnani (1987). Competitive strategy under uncertainty. Strategic Management Journal 8(2), 187-194. Wilson, R. (1968). The theory of syndicates. Econometrica: the Econometric Society , 119-132.

Journal of

150

Bibliography

Wright, M. and A. Lockett (2003). The structure and management of alliances: Syndication in the venture capital industry. Journal of Management Studies 40(8), 2073-2102. Zhang, J. (2007). Access to venture capital and the performance of venture-backed start-ups in silicon valley. Economic Development Quarterly 21(2), 124-147.