University of Southern California Dissertations and Theses

Essays in financial intermediation

(USC Thesis Other)

Essays in financial intermediation

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content ESSAYS IN FINANCIAL INTERMEDIATION by Yun Ling A Dissertation Presented to the FACULTY OF THE USC GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (FINANCE AND BUSINESS ECONOMICS) May 2017 Copyright 2017 Yun Ling Dedication To my beloved father Zhihao Ling and mother Ming Zhang, my husband Lian Liu, and daughter Chloe (Ling) Liu. ii Acknowledgments I am grateful to my advisor, Arthur Korteweg, and to the rest of my disserta- tion committee, Kenneth Ahern, John Matsusaka, Gordon Phillips, and Fernando Zapatero, for their kindness, guidance, and unwavering support. I am also grate- ful to Wayne Ferson and Anthony Marino. I would like to thank all USC FBE colleague and seminar participants at University of Buffalo, University of Illinois at Urbana-Champaign, Thomas Chemmanur and participants at the FMA 2016 Doctoral Consortium for helpful comments. iii Contents Dedication ii Acknowledgments iii List of Tables vi List of Figures viii Abstract ix 1 The Impact of Venture Capital on the Life Cycles of Startups 1 1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Economic Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 1.2.1 Economy and Agents . . . . . . . . . . . . . . . . . . . . . . 5 1.2.2 Sequence of Events . . . . . . . . . . . . . . . . . . . . . . . 6 1.2.3 Funding Decision . . . . . . . . . . . . . . . . . . . . . . . . 10 1.2.4 Estimation Strategy . . . . . . . . . . . . . . . . . . . . . . 13 1.3 Data and Measure . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 1.3.1 Sample . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 1.3.2 Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 1.4 Estimation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 1.4.1 Baseline Estimation . . . . . . . . . . . . . . . . . . . . . . . 23 1.4.2 Constrained Estimation . . . . . . . . . . . . . . . . . . . . 28 1.4.3 Expected VC Impact & Startup Future . . . . . . . . . . . . 30 1.4.4 Joint Effect of Selection and Influence . . . . . . . . . . . . . 32 1.5 Robustness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 1.5.1 Subsample Results . . . . . . . . . . . . . . . . . . . . . . . 35 1.5.2 Alternative Features . . . . . . . . . . . . . . . . . . . . . . 36 1.5.3 Alternative Models . . . . . . . . . . . . . . . . . . . . . . . 37 1.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 1.7 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 1.7.1 Estimation Procedure . . . . . . . . . . . . . . . . . . . . . 41 iv 1.7.2 Measure Construction . . . . . . . . . . . . . . . . . . . . . 49 1.7.3 Alternative Models . . . . . . . . . . . . . . . . . . . . . . . 54 2 Is Collaborative Activism Effective? 57 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 2.2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 2.3 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 2.4 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 2.4.1 Descriptive Statistics . . . . . . . . . . . . . . . . . . . . . . 70 2.4.2 Investment Capital in Co-Activism . . . . . . . . . . . . . . 72 2.4.3 Firm Characteristics Prior to the Filing date . . . . . . . . . 73 2.4.4 Market Response to 13D and 13D/A filings . . . . . . . . . 76 2.4.5 Abnormal Return by Objectives and the Degree of Hostility 78 2.4.6 Changes in Firm Characteristics prior and after the Filing Date . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 2.5 Alternative Explanations . . . . . . . . . . . . . . . . . . . . . . . . 80 2.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 2.7 Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 2.7.1 Variable Construction . . . . . . . . . . . . . . . . . . . . . 83 A Tables 87 B Figures 120 References 130 v List of Tables Table 1-1. Sample Statistics . . . . . . . . . . . . . . . . . . . . . . . . . 88 Table 1-2. Measure Statistics . . . . . . . . . . . . . . . . . . . . . . . . 89 Table 1-3. Estimation Result . . . . . . . . . . . . . . . . . . . . . . . . 91 Table 1-4. Previously Funded vs. Previously Unfunded Startups . . . . . 92 Table 1-5. Constrained Estimation . . . . . . . . . . . . . . . . . . . . . 93 Table 1-6. Expected VC Impact and Startup Future Growth . . . . . . . 94 Table 1-7. Selection on Private Information . . . . . . . . . . . . . . . . 95 Table 1-8. Simulation Results: Model 1 vs. Model 2 . . . . . . . . . . . . 96 Table 1-9. Estimation for Subsamples . . . . . . . . . . . . . . . . . . . . 98 Table 1-10. Estimation with Alternative Features . . . . . . . . . . . . . 100 Table 1-11. Estimation for Alternative Models . . . . . . . . . . . . . . . 102 Table 1A-1. Locations and Categories of Startup . . . . . . . . . . . . . 104 Table 1A-2. Educational Background of Startup Teams . . . . . . . . . . 105 Table 1A-3. Locations of VC and Funded Categories . . . . . . . . . . . 106 Table 1A-4. Educational Background of VC Teams . . . . . . . . . . . . 107 Table 2-1. Summary of Co-Activism Events . . . . . . . . . . . . . . . . 108 Table 2-2. Capital Commitment and Exit Strategy of Co-Activism . . . . 111 Table 2-3. Characteristics of Target Firms . . . . . . . . . . . . . . . . . 112 Table 2-4. Probit Analysis of Target Firms . . . . . . . . . . . . . . . . . 113 vi Table 2-5. Abnormal Performance of Target Firms around the Filings Dates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 Table 2-6. Performance of Target Firms by Objectives and the Degree of Hostility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 Table 2-7. Changes in Firm Characteristics Prior and After Co-Activism 117 vii List of Figures Figure 1-1. Sequence of Events . . . . . . . . . . . . . . . . . . . . . . . 121 Figure 1-2. Distribution of Imputed Values . . . . . . . . . . . . . . . . . 124 Figure 1-3. Distribution of Type Mixture . . . . . . . . . . . . . . . . . . 126 Figure 2-1. Abnormal Return around the Filing Date . . . . . . . . . . . 127 Figure 2-2. Abnormal Share Turnover around the Filing Date . . . . . . 129 viii Abstract My thesis consists of two essays in financial intermediation. The first chapter is “The Impact of Venture Capital on the Life Cycles of Star- tups”. It shows that venture capitalists (VCs) endogenously improve the quality of funded startups. In turn, this makes startups more likely to get subsequent investments. The resulting feedback effect amplifies VCs’ impact over time. To identifystartupqualityasacauseandaneffectoffundingseparately, Idevelopand estimate a dynamic model of funding that incorporates multiple funding rounds. Using CrunchBase, a novel database that includes unfunded startups, I estimate both the differential impact of being funded by one VC versus another, and the overall impact of being funded by a VC versus remaining unfunded. A simulation exercise shows that the feedback effect magnifies the overall impact of VCs by a factor of seven. The second chapter is “Is Collaborative Activism Effective?”. We define col- laborative activism (co-activism) as a set of independent activists that pursue the same objective and work together to influence corporate decisions. Using a hand- collecteddatasetfrom1994to2013, wefindthatco-activismtargetsunderperform- ingfirms, andismostlynonhostiledespiteitsinitialattempttocontrolaround10% of stocks. Co-activism succeeds in roughly 80% of the cases through discussions with the management. The market responds favorably to co-activism around the ix date of filing, but market expectation tapers off over the subsequent year. Our evidence suggests that co-activism aims for short-term gains at the expense of long-term shareholder value, and pressures the companies to take myopic actions that are harmful in the long run. x Chapter 1 The Impact of Venture Capital on the Life Cycles of Startups 1.1 Introduction Venture capitalists (VCs) add value to the startups that they fund (e.g. Sorensen, 2007). For example, VCs can help hire employees, market products, and develop sales. However, little is known about the dynamic interplay between VCs and the quality of funded startups. This paper shows that VCs change the underlying quality of startups, and in turn, improved quality gives funded star- tups an advantage in competing for additional funding. Consequently, changes in quality create a feedback effect that amplifies VCs’ impact over the life cycle of the startup. Overall, I find that funded startups have higher growth potential compared with unfunded ones. This is partially due to startups’ exogenous characteristics. Moreimportantly,VCsendogenouslyimprovethequalityoffundedstartups. Their impact depends not only on the characteristics of startups and VCs, but also on howwelltheyarematched. Withimprovedquality, fundedstartupsaremorelikely to exit through IPO or be acquired. Funded startups that haven’t exited yet are more likely to get subsequent investments compared with unfunded ones. Thus, the feedback effect amplifies small initial quality differences with the number of funding rounds. A random assignment of initial funding is associated with a return 1 spread between funded and unfunded startups that is seven times more than that with no feedback effect. A primary concern to establish the result is to identify startup quality as a cause and an effect of funding separately. On one hand, VCs prefer higher quality when they decide which startups to fund. On the other hand, the funding outcome determines how startup quality will change. Since startup quality is unobserved, there are two challenges to separate cause from effect. First, the effect of funding is observable only for matched pairs of VCs and startups. Quality thus becomes an omitted variable in the sample of matched pairs, a classical endogeneity problem in a regression framework. Second, startup quality is the cumulative result of the funding effects in all past rounds. When startup quality is missing, the identifica- tion bias from the previous round becomes an additional omitted variable in the identification for this round. In other words, the endogeneity problem compounds in a time series framework. I adopt a structural estimation approach to overcome the above challenges. Specifically, IgeneralizeSorensen’s(2007)modeltoadynamicsetting, withstartup quality changing over multiple funding rounds. Hence, the scope of VCs’ impact is not only on startups’ value at exit, but also on startups’ quality as an ongoing concern over multiple financing rounds. I include the dynamics of startup quality as a function of observable characteristics, conditional on the funding outcome in the previous round, to control for endogenous quality change. This is an important generalizationasitallowstheidentificationofthefeedbackeffect. Sorensen’smodel treats funding from seed to exit as one single round with startup quality fixed and exogenously given. His model is therefore unable to detect the feedback effect. Furthermore, I distinguish two types of VCs’ impact: differential and overall. The differential impact compares the effects of being funded by a VC A versus 2 another VC B, while the overall impact compares the effect of being funded by a VC A versus not being funded. The presence of competing startups makes a startups less attractive; however, the other startups don’t affect VC’s impact ons a priori. Hence, competing startups provide a source of exogenous variation. This variation helps identify the differential impact ifs is pushed down to match with a worse VC. If s is left unfunded, it helps to identify the overall impact. Sorensen’s model and data cannot identify the overall impact of VCs. Inthispaper, IuseCrunchBase, anoveldatabasewhichincludesunfundedstar- tups. An important advantage of this data is the inclusion of unfunded startups, which helps to identify the overall impact of VCs. CrunchBase is the database of TechCrunch, a leading online news publisher with the most comprehensive pub- lic coverage of global startup profiles. It provides company information such as products, industries, locations, key employees and funding rounds. It also provides human capital information such as educational background, investment experience and employment history. As of September 2016, the database has slightly fewer than 1 million visitors each week with thousands of corporate and organizational users including the Kauffman Foundation. The novel database allows me to estimate how various startup and VC features relate to the cause and the effect of funding after controlling for startup quality. More specifically, I find that startups’ products, VCs’ experience, and the existence of alumni ties between startups and VCs increase the likelihood of establishing a funding relationship. In contrast, startups’ locations and industries, as well as the distance between startups and VCs reduce the likelihood. These findings on the cause of funding illustrate the general idea in Bengtsson and Hsu (2010) and Bottazzi, Da Rin, and Hellmann (2011) that reduction in information and transaction costs leads to a higher probability of matching. Following funding, 3 the differential impact shows that startups’ quality is improved more by smaller and more compatible VC syndicates whose members have cooperated before. The overall impact shows that compared with unfunded startups, funded ones rely less on macroeconomic conditions and their own management teams, as they can resort to funding VCs for financial and human capital resources. These findings on the effect of funding complement those in Lerner (1995), Helmann and Puri (2002), and Hochberg, Lindsey, and Westerfield (2015). In general, the idea of feedback effect is central and pervasive in economics. Durlauf (1996) finds that local neighborhood selection, which is endogenous to par- ents’ income, affects children’s income through education and sociological mindset. The neighborhoodwide feedback effect creates an endogenous loop leading to per- sistent income inequality. More recently, Kahn (2010) finds that the disadvantage of graduating in a poor economy is persistent. This is because college graduates are more likely to suffer from underemployment or experience job mismatching in a poor economy, and in turn, less labor market experience such as promotions and training put them into further disadvantage. Conceivably, the feedback effect should play an important role in venture capital studies due to the dynamic aspect of sequential investments. However, little has been done thus far to explore this idea. The closest related paper to mine is Bergemann, Hege, and Peng (2011), who propose a theory of endogenous loops between VCs’ information acquisition and their degree of involvement in funding. My paper is the first to establish the feedback effect between startups’ quality and VCs’ investments, and is also the first to quantify the importance of feedback in venture capital studies. The rest of the paper proceeds as follows. Section 1.2 presents the formal dynamicmodelandbrieflydescribestheestimationstrategy. Section1.3introduces 4 CrunchBase and documents variable construction. Section 1.4 discusses empirical findings. Section 1.5 presents robustness results and Section 1.6 concludes. 1.2 Economic Model 1.2.1 Economy and Agents The economy has two types of agents: VC syndicates and startups. With discrete time periods, the set of VC syndicates is constant and is denoted by I. I assume each individual VC syndicate, denoted by i, is always there and ready to provide funding all the time. In contrast, each individual startup, denoted by j, enters the market at birth (i.e., when it is founded), and exits the market either as a success (e.g. going public or getting acquired) or as a failure (e.g. going out of business). I use a list of notations to characterize the different time-varying sets of star- tups. E t is the set of existing startups in the market by the end of t. It is also the set of existing startups at the beginning oft + 1. N t is the set of newborn startups that enter the market at t. The entrance occurs at the end of t after the funding decision is made. IPO/MA t andD t denote the sets of successful and unsuccessful exits att, respectively. The exiting startups do not need more funding, so the exits occur before funding decision is made. The startups that remain in the market need to compete for more funding. I call them âĂIJfunding candidatesâĂİ. How- ever, not all of them manage to get funded at t. The set of funding candidates is J t . There are two identities that relate the different sets of startups. First, each of the existing startups at the beginning of t will have exactly one of the three cases: exit successfully, exit unsuccessfully, or remain in the economy. Second, 5 the existing startups at the end of t consists of the newborn startups, and the old startups that do not exit (i.e., the funding candidates). Equivalently, the two identities can be written as follows using the notations defined above. E t−1 =IPO/MA t +D t +J t (K1) E t =J t +N t (K2) 1.2.2 Sequence of Events This subsection gives the sequence of events at timet. It highlights the dynam- ics of three key endogenous variables for startups. In the following, I first give the definitions of the three variables. I then describe the sequence of events, along with the dynamics of the key variables. Three Startup Variables The first variable, denoted byr j t , is equal to the cumulative return of a startup j up to time t. I call it the “growth variable”. By definition, the investment of $1 dollar in the startup at inception will worth expr j t at timet. 1 The second variable, denoted by s j t , is a signal perceived by the public on the well-being of a startup j at time t. I call it the “implicit exit value”. It will determine whether a startup exit at timet and in which way if it exits (e.g. IPO/MA, or Death). The evolution of both variables depend on whether a startup is funded by a VC syndicate in the previous time period. As a result, they reflect the direct influence of VC funding. The third variable, denoted by v ij t , is the expected value created jointly by a VC syndicate i and a startup j, if they choose to have a funding relationship at 1 The cumulative return is continuously compounded to calculate the value of investment (net of dilution). 6 time t. 2 The expected value is subjective and different across all pairs of i and j. Therefore, I call it the “subjective value added”. An important assumption is that a pair of startup and a VC syndicate share a common perspective toward the expected value they would create jointly. Together, the set of all concurrent expected values will determine the selection of VC funding. Sequence of Events [Insert Figure 1-1] A sequence of events occur at timet, as illustrated in Figure 1-1. First, the growth variable at timet is determined. It depends on the lagged growth variable att−1, the previous funding relations, and various current features. More specifically, if a startup is not funded previously, its return depends only on its own features. In contrast, if it is funded by a VC syndicate previously, its return also depends on the features related to the VC syndicate. Eq(1) gives the law of motion for the growth variable. The various current features include: macroeconomic variables X t , startup features X j t , VC syndicate features X i t , and startup-VC syndicate pair features X ij t . I use φ r,y and φ r,n to denote the coefficients associated with 2 Mathematically,v ij t ≡E ij t h r j t+1 −r j t i is funding j from t to t + 1 i , namely the value added r j t+1 −r j t from this period to the next period in the common perspective of i and j at time t, given that i is funding j from t to t + 1. 7 relevant features separately for the two cases. The noises are independent and follow Normal distributions with variances σ 2 r,y and σ 2 r,n . 3 r j t −r j t−1 =          h 1,X t ,X j t ,X i t ,X ij t i φ r,y +σ r,y j t if j is funded by i at time t− 1 h 1,X t ,X j t i φ r,n +σ r,n j t if j is unfunded at time t− 1 (1) Second, given the growth variable, the implicit exit value is determined. It also depends on a similar set of features according to whether a startup is previously- funded. In addition, the implicit exit value also depends on the concurrent growth variable. This assumption is to capture the intuition that a startup is more likely to have a successful exit, or equivalently the implicit exit value is higher, when its cumulative return is higher. Eq(2) gives the determinants of the implicit exit value. As before, I use φ s,y and φ s,n to denote the coefficients separately for the previously-funded and previously-unfunded cases. The noises are independent and follow Normal distributions with variances σ 2 s,y and σ 2 s,n . s j t =          h 1,X t ,X j t ,X i t ,X ij t ,r j t i φ s,y +σ s,y η j t if j is funded by i at time t− 1 h 1,X t ,X j t ,r j t i φ r,n +σ s,n η j t if j is unfunded at time t− 1 (2) Third, given the implicit exit value, a startup’s status is determined. For a startup j, the status denotes whether j exits the market at t and in which way if it exits. The status takes three values, “IPO/MA”, “Death”, and “Survival”, according to the implicit exit value. Eq(3) gives the correspondence. 4 As a result, 3 The first subscript “r” indicates that the equation’s dependent variable is r. The second subscript, “y” or “n”, indicates the answer (“yes” or “no”) to the question of whether the startup is previously VC-funded. 4 I setδ to 3. It could an arbitrary positive value. This is because the equation fors is uniden- tified without the specification of δ. The change of δ only shifts and re-scales the distribution of 8 the set of existing startups at the beginning of time t breaks into three groups. IPO/MA t contains the startups with status “IPO/MA”. D t contains the startups with status “Death”. J t contains the startups with status “Survival”. Recall that the startups inJ t are funding candidates and will compete for VC funding at time t. status j t =                    “IPO/MA” if δ≤s j t “Survival” if −δ≤s j t <δ “Death” if s j t <δ (3) Fourth, given the set of funding candidates, the subjective value added is deter- mined, for each pair of startup and VC syndicate in the economy. The pair-wise value depends on a set of startup features, VC syndicate features, and startup-VC syndicate pair features. Again, it also depends on a startup’s growth variable. Eq(4) gives the determinants of the subjective value added. 5 I use φ v to denote the coefficients. The noises are independent and follow standard Normal distribu- tions. 6 Fifth, given the subjective value added, the equilibrium funding at time t is determined as a two-sided matching between the set of VC syndicates and the s. Consequently, the estimated coefficients will have its intercept shifted, and other coefficients multiplied by common factor. 5 The equilibrium funding decision is determined through the comparison of the subjective value added. This will be discussed later. However, a common shift of all concurrent subjective value added will not change the equilibrium funding. Thus, the coefficients associated with the intercept and the macroeconomic variables are unidentified. Therefore, those variables are not included in the equation for v. 6 I set the variance to 1. It can be any arbitrary positive value. Again, this is because the equation forv is unidentified without the specification of it. The change of variance only re-scales the distribution of v and will not change the equilibrium funding decision. 9 set of funding candidates. I use μ t to denote the funding. More details will be given in the following subsection. v ij t = h X j t ,X i t ,X ij t ,r j t i φ v +ξ ij t , for all i∈I, j∈J t (4) Lastly, after the funding relationship is established, newborn startups enter into the market at the end of t. Their company value is set to $1, or equivalently their growth variable is set to zero. Thus, the existing startups at the end of t include newborn startups and the old startups that have not exited. 1.2.3 Funding Decision The funding decisionμ t lasts for one period from timet to timet + 1. I use the two-sided matching model in Sørensen (2007) as the prototype for the one-period funding decision. Two-sided means that both the VC syndicates and the startups are active in the search of a funding relationship. However, a startup can only be matched to one VC syndicate, while a VC syndicate can be matched with multiple startups. The number of startups a VC syndicatei funds att is denoted byq i t , and it will be calibrated to equal the actual number in estimation. As discussed before, the set of subjective value added n v ij t :i∈I,j∈J t o determines the equilibrium matching μ ∗ t . In the following, I first give the utility maximization problem for both agents in terms of the subjective value added, then discuss the equilibrium. Preferences and Choice Variable Given a funding relationshipμ t , the utility is the sum of subjective value added thatanagentcreates. Forastartupj,U j t equalsv ij t ifthereisafundingrelationship betweeni andj. For a VC syndicatei,U i t equals the sum ofv ij t if there is a funding 10 relationship betweeni andj. The choice variable is the indicator whether the pair (i,j) is in μ t . Both startups and VC syndicates can propose to the counterparty for the establishment of a pair. However, the pair (i,j) ends up in μ t if and only if both i and j want it to exist. Eq(5) and Eq(6) define the utility function and state the utility maximization problem. U i t = X j:ij∈μt v ij t , s.t.|i :ij∈μ t |≤q i (5) U j t = X i:ij∈μt v ij t , s.t.|j :ij∈μ t |≤ 1 (6) There are two assumptions behind the definition of utility. First, for each pair of VC syndicate and startup, I assume they agree with the subjective value added v ij t . Second, I assume there is a common fraction, say λ, of the value added goes to a VC syndicate, and the remaining fraction, 1−λ, goes to a startup. This assumption allows me to ignore the differences in bargaining power and agency problem. Thus, the ranking among the set of expected value added is sufficient to determine the equilibrium funding. Pairwise Stability and Equilibrium The equilibrium matching μ ∗ t is pairwise stable. Pairwise stability means that there exists no pair that can gain from pairwise deviation. A pairwise deviation would occur for a pair (i,j) not in μ t that both i and j prefer each other to their current matched counterparties. As a result, i and j would break up with their existing matched counterparties and form a new pair (i,j) between them. In the equilibrium matching μ ∗ t , such a pairwise deviation is not profitable for any pair. 11 The equilibrium matching exists and is unique if all v ij t are distinct. 7 The equilibrium condition can be characterized by a set of inequalities as in Sørensen (2007). Eq(7) and Eq(8) give the inequalities. For a pair (i,j) not inμ t ,v ij t is not greater than the two opportunity costs of i and j to break up with their matched counterparties. The two opportunity costs are equal to min ij 0 ∈μt v ij 0 t and v μt(j)j t respectively. Here, I use μ t (j) to denote the matched VC syndicate for j at t. For a pair (i,j) inμ t ,v ij t is greater than allv i 0 j t thati 0 wants to deviate toj and allv ij 0 t thatj 0 wants to deviate toi. LetS t (i) denote the set ofj 0 that want to deviate to i, and let S t (j) denote the set of i 0 that want to deviate to j. Eq(11) and Eq(12) give the expressions for S t (i) and S t (j) respectively. (i,j) / ∈μ t ⇔v ij t <v (7) (i,j)∈μ t ⇔v ij t ≥v (8) v = max min ij 0 ∈μt v ij 0 t ,v μt(j)j t (9) v = max " max j 0 ∈St(i) v ij 0 t , max i 0 ∈St(j) v i 0 j t # (10) S t (i) = n j 0 ∈J t :v ij 0 t >v μt(j 0 )j 0 t o (11) S t (j) = i 0 ∈I :v i 0 j t > min ij 0 ∈μt v i 0 j 0 t (12) 7 This is because i and j share the same perspective of v ij t . A proof is given in Sørensen (2005). In general, it is not true. The stable matching problem can be solved by the Gale- Shapley algorithm. 12 1.2.4 Estimation Strategy The parameter estimation is performed jointly on the main system of equations Eq(1), Eq(2), andEq(4). Theparametersinclude (φ r,y ,σ 2 r,y ), (φ r,n ,σ 2 r,n ), (φ s,y ,σ 2 s,y ), (φ s,n ,σ 2 s,n ), and φ v . Eq(1) and Eq(2) give the direct influence of VC funding on startup growth and implicit exit value. The direct influence is shown in a com- parison between the previously-funded and previously-unfunded startups. Eq(4) describes subjective value added as the determinant of the selection of VC fund- ing. The selection corresponds to a two-sided matching between VC syndicates and funding candidates. Not all variables in the main system of equations are observed. The three dependent variables are latent or only partially observed. Thus, the estimation involves the imputation of these variables. 8 The observed data is composed of four pieces. The first piece is the partially observed growth variable. It is observed at birth and exit of a startup, or when the startup gets VC funding. The second piece is the startup status (e.g. IPO/MA, Death, Survival) at each time period. It helps the imputation of the implicit exit value. The third piece is the equilibrium funding. It helps the imputation of the subjective value added. The last piece includes the macroeconomic variables, and the various features of startups, VC syndicates, and startup-VC syndicate pairs. They are the independent variables for the three equations. I use Gibbs Sampler to estimate the parameters in a Bayesian framework. In fact, it is simpler to estimate in this way. First, regarding the interdependence among the key variables, it is impossible to use regressions alone to accomplish the joint estimation. Also, the existence of latent variables makes it implausible to 8 The imputation relies on the observed data, the last-updated parameter estimates, and most importantly, the interdependence among the three key variables. 13 estimate using a GMM/SMM strategy. Therefore, it is easiest to use the Bayesian framework because it can handle the interdependence in the presence of latent variables. The Bayesian framework gives tractable posterior distributions for all the parameters and variables given proper priors. Thus, a tractable algorithm can be implemented using the Gibbs Sampler. The Gibbs Sampler iterates between parameter updates and the latent variable imputations. Section 1.7 gives the detailed algorithm for the estimation strategy. The distribution assumption is given as follows. I assume that the noises in Eq(1), Eq(2), and Eq(3) are independent. 9 As defined before, they follow Normal distributions with different variances. I also assume the parameters have conjugate priors. In Eq(1) and Eq(2), the priors of the joint distributions (φ,σ 2 ) follow Normal-Inverse-Gamma distributions. In Eq(3), the prior of φ v follows a Normal distribution. Section 1.7 gives the detailed assumption for the priors. For post- estimation analysis, I use the posterior t-statistics for hypothesis testing. 1.3 Data and Measure Most research in venture capital uses proprietary databases (e.g. VentureOne, VentureXpert). ForthestudyofVC’simpactonstartups, thebiggestdisadvantage of these databases is that they lack a control group of non-VC-funded startups. In this paper, I hand-collect a novel database from a leading startup platform, Crunchbase, that provides information for both VC-funded and non-VC-funded startups. Founded in 2007, Crunchbase has 1.5 million unique visitors each month in 2013. By 2014, it has a record of more than 290,000 companies (e.g. startups, 9 Later on, in an extended model, the three noises are assumed to be correlated. 14 VCs, incubators, accelerators, etc.) and 310,000 individuals (e.g. entrepreneurs, venture capitalists, angel investors, etc.) across 176 countries. For companies, a typical record includes founding information, current status (IPO, acquired, alive, dead), acquisitions and investments history 10 , funding history 11 , products and categories information 12 , contact information, and company news. There is also human capital information that relates companies to individuals. The individ- uals include founders, angel investors, board members and advisors, and personnel on the current and past teams of management for a company. For individuals, a typical record includes name, gender, primary location, employment history, and educational background. One unique feature about Crunchbase is that it collects information by crowd- sourcing. The advantage is that the database can be built very quickly at an exponential speed with insiders, especially entrepreneurs, feeding detailed infor- mation. Another database that is built in this way is Wikipedia which has more than 120,000 regular contributors and 12,000 editors by now. Like Wikipedia, the disadvantage of Crunchbase is that it may contain some inaccurate information. The Crunchbase team combines human and machine reviews to prevent it. 13 To check the credibility, I manually compare the Crunchbase profiles for a subsample of startups with the information from major business journals and pro- prietary databases. The subsample includes 250 startups with successful exits (IPO/MA) and 790 startups that have funding records from VentureXpert. More specifically, I compare the numbers for the money raised in IPO, the transaction 10 The company is the acquirer or investor. 11 The company is the investee. 12 The category is classified by Crunchbase denoting the sub-industry for the company. The categories are not mutually exclusive. 13 For more details, please see https://info.crunchbase.com/about/faqs. 15 value for MA, and the money invested in the funding rounds. Those numbers are similar for the information collected from Crunchbase and from other sources. In addition, I apply a number of filters to select the startups with the most accu- rate information for model estimation. The filtering procedure will be discussed in details in the construction of sample. Finally, one prevalent concern on any startup database is that it may contain some zombie companies. A zombie company shows as alive on the record but is actually out of business. Thus, I need to change the final status from “Survival” to “Death” for zombie companies in my sample. To do that, I visit each startup’s website in the sample if its final status is “Survival” according to the Crunchbase profile. Itturnsoutthatover65%ofthedeadcompaniesinthesamplearedetected in this way. 1.3.1 Sample The sample period is from 1998 to 2014. I apply a number of filters to select a sample that is suitable for the study and has the most accurate information. The first set of filters is on startups. A startup needs to have a birth year equal to or greater than 1998 to be included in the sample. Moreover, I include a startup if it has available website, category, headquarter, founder, and current team of management information on its Crunchbase profile. This gives a total number of 29,184 startups. The second set of filters is on VCs and funding rounds. For VCs, I only include experienced VCs that have participated in at least ten funding rounds in the sam- ple. It is due to the model assumption that VCs are always there ready to provide funding. This gives a total number of 765 VCs. For funding rounds, a filter is applied on the type of funding. It excludes angel-investing, debt-financing, equity 16 or product-crowdfunding, grant, non-equity-issuance, post-ipo-debt or equity, and secondary-market-investing. A funding round also needs to have available infor- mation on investment amount and post-investment valuation to be included in the sample. This gives a total number of 21,483 funding rounds. The third set of filters is on IPOs and MAs. For IPOs, I delete a record if a company’s market value at IPO cannot be calculated or otherwise obtained from other sources. Fortunately, no record of is dropped in this way. For MAs, I delete a record if either transaction value or acquired proportion of a company is missing and unavailable from the SDC Platinum Mergers & Acquisitions database. This gives a total number of 323 IPO and 1,728 MA records. Finally, the resulted datasets of startups, VCs, funding rounds, IPOs and MAs need to be consistent with one another. For instance, I delete the whole record of a startup if its corresponding IPO, MA or funding-round record is excluded due to missing information. I also delete the whole record of a VC if all of its funding records have been deleted given the above filters. Accordingly, VC-funded startups account for a smaller proportion in the sample than in the resulted dataset from the first set of filters. To keep the proportion roughly the same, I randomly drop non-VC-funded startups with a preference for those with the least information in the database. [Insert Table 1-1] Table 1-1 gives a descriptive summary of the final sample. It contains 9,303 startups and 755 VCs that have formed 2,844 distinct VC syndicates. Among the 9,303 startups, only 2,350 (25.26%) have been funded by VCs. For the VC-funded ones, about 22.47% go public or get acquired and 17.62% finally die. For the non-VC-funded ones, the proportions are 1.57% and 27.02% respectively. Getting VCs’ funding can increase the IPO/MA rate by 14 times. Regarding the number 17 of rounds for VC-funded startups, both the median and the mode are 2 rounds per startup. Regarding the size of VC syndicates, the mean and the median are 2.93 and 3 VCs per syndicate. 1.3.2 Measures There are two groups of variables for which I need to construct measures. The first group consists of startups’ birth, final status (e.g. IPO/MA, Survival, Death), and funding status (e.g. VC-funded or not, and by which VC syndicate if funded). It is straightforward to construct these measures as they are directly recorded in the database. They will be used to update the posterior distributions of latent variables. The second group consists of the dependent and independent variables in the main system of equations for estimation (Eq(1), Eq(2), and Eq(4)). The three dependent variables are the cumulative return, the implicit exit value, and the subjective value added. Among them, the implicit exit value and the subjective value added are latent variables to be imputed. The cumulative return is observed sporadically. Based on these observations, interim values are imputed during esti- mation. The independent variables include various observed features of startups and VC syndicates. The following details the construction of the cumulative return and those various features and Section 1.7 gives a summary. Cumulative Return By definition, the cumulative return is the logarithm of a startup’s valuation. I assume that all startups have a valuation equal to 1 at birth. Later on, a startup has its valuation revealed at exit or funding rounds. I estimate the valuation in three ways. First, I calculate the valuation at exit for startups that finally 18 exit during the sample period. These are the startups with final a status equal to “IPO/MA” or “Death”. For “IPO/MA” ones, I set the valuation to be the reported market value at IPO or the deal price divided by the percentage acquired in MA. For “Death” ones, I sample the valuation from a triangle distribution with a mode of 0.1. For the exact time of death, I collect additional information to determine when is the last time these finally-dead startups have events or news. I then sample from a uniform distribution spanning from 6 to 30 months after that time for the exact time of death. Second, I calculate the valuation at funding rounds for startups that have received VC funding during the sample period. The valuation is net of the pure money effect of investment, namely “anti-diluted” as defined in Cochrane (2005) and Korteweg and Sørensen (2010). More specifically, the valuation can be calcu- lated by compounding the “anti-diluted” period-to-period return. The period-to- period return from the last funding round to this funding round is equal to the ratio of the pre-investment value at this round to the post-investment value at last round. 14 As a result, the “anti-diluted” valuation is equal to the product of the period-to-period returns between neighboring funding rounds. This definition measures the growth rate of a VC-funded startup by excluding the dollar amount of investment. Third, I estimate the valuation for startups whose final status is “Survival” by the end of the sample period. The estimation is based on current startup performance. In particular, I visit their websites and use their Crunchbase profiles (e.g. company description, current team of management, offices, investments and 14 Section 1.7 gives the formula for the “anti-diluted” period-to-period return and company value. 19 acquisitions 15 ) to evaluate the performance. I then classify the startups into three groups according to their performance ranked in a decreasing order. Next, for each group, I find comparable startups with similar features and non-missing valuations at exit. Finally, I impute the valuation for each group by drawing samples from a smoothed distribution of the exit values of comparable startups. 16 Table 1-2 Panel A gives the summary statistics for the observed cumulative return. As implied by the percentile information, the cumulative return is very dispersed and has a bimodal distribution. [Insert Table 1-2] Macroeconomic Variables The macroeconomic variables include the risk-free rate (r f ), the Fama-French three factors ((r m −r f ), smb, hml), and a proxy for the cost of long-term borrowing (Y baa −Y us10 ). (Y baa −Y us10 ) is equal to the spread of the yield of Moody’s seasoned Baa corporate bond over the yield of 10-year Treasury bond. Both the risk-free rate and the spread are obtained from the Federal Reserve Bank of St. Louis. The Fama French three factors are from Ken French’s website. (r m −r f ) is excess market return over risk-free rate; smb is the factor return on the small-minus-big portfolio; and hml is the factor return on the high-minus-low portfolio. Table 1-2 Panel B gives the summary statistics. 15 The startups are the investors and acquirers. 16 Admittedly, the estimation is subjective. Nevertheless, given the huge variation in startups’ performance, it is better off to have a rough estimate than leave the valuation missing. In the latter case, the distribution of imputed interim returns would be unrealistically flattened. Also, the returns for successful and unsuccessful startups should follow very different distributions. 20 Startup Features The startup features are either constant or time-varying. The constant startup features are along three dimensions: location, category, and product. # locations is the number of cities that a startups headquarter or offices are located in. # categories is the number of categories a startup is classified into by Crunchbase. # products is the number of products that a startup has. In addition, I construct a set of dummies 1(LOC) to indicate whether a startup has a headquarter or offices located in a specific place LOC. LOC takes “ny” for New York, “ca” for California, “ous” for other places in U.S., “ona” for other places in North America, “as” for Asia, and “eu” for Europe. Table 1A-1 gives the list of top 20 cities and categories with the most startups. The time-varying startup features characterize funding history and human cap- ital information. For funding history, t from last round is the time in years since last funding round; t2 from last round is the square of it. # rounds is the num- ber of funding rounds experienced in the past. For human capital information, # startups founded is the number of companies that the startup founder has built in the past. 1(top20 school) is the dummy variable indicating whether a startup has people on its management team who graduated from a top school at a specific time. Table 1A-2 gives the top school list. 17 Table 1-2 Panel B gives the summary statistics for the startup features. 17 Both lists in Table 1A-2 and Table 1A-4 exclude some best schools but include some schools with lower ranks. There are two reasons. First, I give priority to schools that provide the best education in some technical fields (e.g. engineering, biochemistry). Second, I select schools that appear most frequently in the educational background of VC and startup personnel, since these schools should have very strong alumni network. 21 VC Syndicate Features The VC syndicate features also include the constant and the time-varying. The constant features characterize geographical and size information. # of VCs is the number of VCs in a syndicate and it measures the size of the syndicate. # locations if the number of cities that a VC syndicate has at least one VC member that has an office or headquarter located in it. Table 1A-3 gives the lists of top 20 cities with the most VCs (Panel B) and with the most VC syndicates (Panel C). A comparison between the two lists shows that the cooperation is prevalent among VCs in different locations. For instance, the percentage of VCs that has an office in New York, San Francisco, Menlo Park, and Palo Alto is at most 15%. In contrast, the percentage of VC syndicates that has an office in these places is at least 44%. The time-varying features characterize investment and human capital informa- tion. For a VC syndicate, 1(cooperated) is a dummy variable indicating whether any member VCs have cooperated in the past. # categories is the number of categories that a VC syndicate has at least one member VC that has investment experience in it in the past. # rounds is the median funding rounds that VC mem- bers have participated in before. This variable measures the average experience of a VC syndicate. Finally, 1(top20 school) is the dummy variable indicating whether a VC syndicate has people who graduated from a top school on its members’ man- agement teams at a specific time. Table 1A-4 gives the top school list. Table 1-2 Panel C gives the summary statistics for the VC syndicate features. Startup-VC Syndicate Features Last, I construct measures for each pair of startup and VC syndicate. There are 26,457,732 startup-VC syndicate pairs in total. The constant features measure the 22 pairwise geographical distance. Distance is the closest distance in miles between a startup and a VC syndicate. Based on that, days of travel is the number of (additional) days for a round travel. It equals 0 if distance is within 100 miles, 1 if distance is between 100 and 1,000 miles, 2 if distance is between 1,000 and 10,000 miles, and 3 otherwise. 18 The percentage of all pairs that has a distance with 100 miles is around 25%. For comparison, the percentage of pairs that has a funding relationship and has a distance within 100 miles is around 75%. The time-varying features describe the existence of funding relationship and alumni ties. For a startup-VC syndicate pair, 1(funding tie) is a dummy variable indicating whether any VC member has funded the startup in the past. 1(alumni tie) is a dummy variable indicating whether any VC member and the startup have people graduated from the same school. Table 1-2 Panel D gives the summary statistics for the startup-VC syndicate features. 1.4 Estimation Results 1.4.1 Baseline Estimation Table 1-3 presents the joint estimation result for the baseline model. In the table, the first four columns show the direct influence of VC funding on startup growth and implicit exit value as a comparison between previously-funded and previously-unfunded startups. Among them, columns 1 and 2 give the parameters for the law of motion of the growth variable in Eq(1); columns 3 and 4 give the parameters for the dynamics of the implicit exit value in Eq(2). The last column 18 For distance within 100 miles, I assume a one-day round travel by car. For distance between 100 and 1,000 miles, I assume a two-day travel by flight. For distance between 1,000 and 10,000 miles, I assume a three-day travel by flight. For distance greater than 10,000 miles, I assume an intercontinental travel. 23 shows the determinants of the selection of VC funding. It gives the parameters for the dynamics of the subjective value added in Eq(4). Influence of VC Funding The influence of VC funding on startup growth comes from two sources. First, there is a direct impact from the funding VC syndicate features and the pair fea- tures. Notethatthesefeaturesarenon-missingonlyforpreviously-fundedstartups. Second, there is an indirect impact given the presence of the funding VC syndicate. The indirect impact changes the effects of the macroeconomic variables and the startup features. It is shown as a comparison between the previously-funded and unfunded startups. For the direct impact, various VC syndicate and pair features show significance. For instance, funded startups’ growth is negatively correlated with VC syndicate size (# of VCs) and positively correlated with the educational background of indi- vidual venture capitalists (1(top20 school)). In addition, past cooperation among any VC members (1(cooperated)) also helps funded startups grow faster. Regard- ing the pairwise features, both the existence of alumni ties (1(alumni tie)) and the existence of past funding relations (1(funding tie)) correspond to a higher growth rate for the funded startups. It might result from a reduction in asymmetric information and agency cost facilitated by learning through social network or past cooperation. For the indirect impact, previously-funded startups are less dependent on the macroeconomic conditions. For instance, the cost of alternative funding, indicated by the difference between the BAA corporate bond yield and the 10-year Treasury yield (Y baa −Y us10 ), has a more significant negative effect on startup growth when the startup is previously-unfunded. Likewise, startup growth depends more on the 24 Fama French three factors ((r m −r f ), smb, hml) as well as the risk-free rate (r f ) without funding. Regarding the startup features, one surprising result is that the effects of startup human capital features exhibit opposite signs for the previously-funded and unfunded cases. Founding experience of entrepreneurs (# startups founded) has a positive effect on a startup growth when it is not funded. However, the effect becomes negative in the presence of VC funding. Similarly, the educational back- ground of a startup’s management team (1(top20 school)) promotes growth with- out funding but impedes growth with funding. The positive impact of a startup’s humanresourceseemstobesupplantedbythefundingVCsyndicate’s. Thechange in the sign indicates a power struggle between the top executives of the funded startups and funding VCs. The influence of VC funding on startup implicit exit value is mainly through its effect on startup growth. Faster startup growth implies better startup quality (i.e., higher cumulative return) and corresponds to a higher implicit exit value. As a result, a startup is more likely to exit through IPO or MA and less likely to run out of business. For the previously-funded case, a one standard deviation increase in the cumulative return (7.408) is associated with an increase of 0.207 in the implicit exit value. It corresponds to a 49.6% relative increase in the probability of IPO/MA and a 35.4% relative decrease in the probability of Death compared with the benchmark mean values. 19 For the previously-unfunded case, the relative increase and decrease in the probabilities of IPO/MA and Death are 46.3% and 45.2% respectively. 19 For the previously-funded case, the mean and standard deviation for the implicit exit value are 0.044 and 1.307. The mean probabilities of IPO/MA and Death are 1.19% and 0.99%. Thus, an increase of 0.207 in the implicit exit value changes the IPO/MA probability to Φ ((0.647−δ)/1.307) = 1.78%, and changes the Death probability to Φ ((−0.647−δ)/1.307) = 0.64%. Φ is the c.d.f. of standard Normal. δ equals 3. 25 [Insert Table 1-3] Selection of VC Funding The selection of VC funding depends on the subjective value added. One of the determinants for the subjective value added is startup quality. A one standard deviation increase in the cumulative return (7.172) 20 is associated with an increase of0.244inthesubjectivevalueadded. Itcorrespondstoamarginalincreaseof6.9% in the probability of getting funded, relative to a 50% benchmark probability. 21 In addition to startup quality, a wide range of startup- and VC-related features enter into the consideration of funding selection. Regarding the startup features, the expected value added grows with the number of a startup’s products (# prod- ucts)butdeclineswiththestartup’sgeographicalspan(# location)andcategorical span (# categories). It indicates a preference of more productive but focused star- tups. Also, startups that have experienced more funding rounds is preferred, as they have been selected by other savvy VCs in the past. Interestingly, better startup human capital (# startups founded, 1(top20 school)) lowers the chance of receiving VC funding. It is not surprising given their negative impact on funded startup’s growth. Regarding the VC-related features, smaller VC syndicates (# of VCs) with more funding experience (# rounds) correspond to a higher expected impact. However, diversity in the categories of funding (# categories) is associated with a lower expected impact. Also, better educational background of venture capitalists (1(top20 school)) is less preferred. Between a pair of VC syndicate and startup, the 20 This is the standard deviation of the cumulative return over the whole sample. 21 The benchmark 50% is the probability that some v ij is greater than v i 0 j 0 given the same distribution assuming all their determinants in Eq(4) are the same. Thus, an increase of 0.244 in v ij changes this probability to Φ 0.244/ √ 2 = 56.86%. Φ is the c.d.f. of standard Normal. 26 geographical distance (days of travel) decreases the expected value added while the existence of alumni ties (1(alumni tie)) and previous funding relationship (1(fund- ing tie)) increase it. The expected value added seems to be consistent with the VC syndicate’s impact on funded startups following the funding decision. Previously Funded vs. Previously Unfunded The overall effect of VC funding is measured by a comparison of startup growth and implicit exit value between previously-funded and previously-unfunded star- tups. It is to compare the dependent variables in Eq(1) and Eq(2). Table 1-4 com- pares the mean values using different control groups. The different control groups are all of the previously-unfunded startups and subgroups of them including only comparable ones. I use the Nearest Neighbor method and the Propensity Score Matching method to choose comparable unfunded startups. Identifying covariates include startup features and macroeconomic variables. Figure 1-2 compares the distributions. [Insert Table 1-4] [Insert Figure 1-2] In general, previously-funded startups exhibit faster growth and higher implicit exit value. For the period return 22 , the previously-funded group has a raw and trimmed mean both equal to 3.4%. 23 The previously-unfunded group has a raw and trimmed mean equal to 5.1% and 1.7% respectively. Growth of unfunded startups seems to have huge variance. However, on average, it is smaller than the growth of funded startups after ignoring the outliers. 22 Period return is the first difference in the cumulative return. One period is one month. 23 The trimmed mean is the mean of the subsample winsorized between 1 and 99 percentiles. 27 The mean exit value is 0.044 for previously-funded startup and -0.197 for previously-unfunded ones. It implies a higher unconditional probability of suc- cessful exit and a lower probability of failure exit for funded startups. This is consistent with the results on IPO&MA rate and Death rate. The two rates are the proportions of startups that go IPO/MA and Death this period following the previous funding decision. They represent the conditional probabilities of success- ful and failure exits given that a startup still exists in the economy. The IPO&MA and Death rates are 0.5% and 0.4% for previously-funded startups. In contrast, they are 0.02% and 0.49% for previously-unfunded startups. A Chi2 statistics of 164.9 shows that the two conditional distributions of successful and failure startups are significantly different across the funded and unfunded groups. The results stay qualitatively the same using comparable unfunded startups as alternative controls. The difference in the mean trimmed period return is around 5.2% between the funded and comparable unfunded startups. The difference in the mean implicit exit value is around 0.2. By large, VC funding improves the startup return and the chance of successful exit in the following period after controlling the determinants of funding selection. 1.4.2 Constrained Estimation Both startup growth and implicit exit value have separate dynamics across funded and unfunded startups. One concern is that the separate dynamics might mask the first-order effects of VC-related features. To address this concern, I impose the constraints such that funded and unfunded startups have the same dynamics of evolution. In particular, the constraints make the coefficients of the commontermsthesameforthetwogroupsofstartups. Thecommontermsinclude the intercept, the macroeconomic variables, and the startup features. By imposing 28 the constraints, it shuts down the indirect impact of VC funding. Section 1.7 gives the estimation strategy for the constrained model. Table 1-5 presents the constrained estimation results. [Insert Table 1-5] For the selection of VC funding, the coefficients stay almost the same. One exception is the effect of days of travel. It is tripled (-0.696) compared with the baseline model (-0.210). Now, a one standard deviation decrease in days of travel is associated with 0.176 increase in the expected value added. It corresponds to a marginal increase of 5.0% in the probability of getting funded, compared with a 50% benchmark probability. For the influence of VC funding, the new coefficients have some interesting patterns. First, the effects of the VC-related features are similar to those in the baseline model. It implies that the direct impact of VC funding is not sensitive to the assumption of separate dynamics for funded and unfunded startups. Second, the effects of the common terms (the intercept, the macroeconomic variables, and the startup features) are close to those for the unfunded case in the baseline model. This might be due to the large proportion of unfunded startups in the panel data. Moreover, the effect of startup quality on implicit exit value is now much larger. Now, a one standard deviation increase in the cumulative return corresponds to a 282.5% relative increase in the probability of IPO/MA and a 76.2% relative decrease in the probability of Death. 24 However, the boost in the coefficient is due to the fact that VC funding works as a confounding factor that improves startup quality and implicit exit value at the same time. In fact, it is more appropriate to have different dynamics for funded and unfunded startups to separate out the 24 The associated increase in the implicit exit value is 0.581. It increases the probability of IPO/MA from 0.40% to 1.53%. It decreases the probability of Death from 0.84% to 0.20%. 29 funding effect. Therefore, the following analysis is performed on the estimation results from the baseline model. 1.4.3 Expected VC Impact & Startup Future Given the estimation result, I then investigate how the subjective value added lines up with the true value added following funding. That is, within the group of previously-funded startups, whether the expected VCs’ impact correctly reflects startups’ future performance (startup growth, implicit exit value), as a measure of the true value added. Expected VC Impact & Startup Future For a direct test, I look at the correlations between the subjective value added and startup future performance. The sample of the test only includes funded star- tups at each funding round. Note that only on funded startups, venture capitalists haveachancetorealizetheirexpectedimpact. Also, theimpactmightlastformul- tiple periods. Therefore, I use each funding round as an observation, to facilitate the study of the impact’s duration. Table 1-6 presents the results. [Insert Table 1-6] The correlation is positive and significant for startup future performance one- period ahead. Within the group of funded startups, a one standard deviation increase in the expected VC impact is associated with a 0.114 standard deviation increase in the startup return one-period ahead. It also relates to a 0.056 standard deviation increase in the implicit exit value next period, which corresponds to a 15.1% relative increase in the probability of IPO/MA and a 13.7% relative decrease in the probability of Death. However, the positive correlation quickly fades away 30 beyond one period. It seems that the expected VC impact is better actualized in the near future of funded startups. cov v ij t ,r j t+1 −r j t =cov h X j t ,X i t ,X ij t ,r j t i φ v +ξ ij t , h X t+1 ,X j t+1 ,X i t+1 ,X ij t+1 i φ r,y +σ r,y j t+1 (13) The strong positive correlation one-period ahead is actually hinted in the esti- mation results. To see this, we can write out the covariance between the subjective value added and funded startups’ return next period as in Eq(13). One thing to notice is that a large proportion of the covariates are slow-moving, with half of them time-invariant. Therefore, the sign of the correlation depends on whether the signs of the corresponding coefficients in the estimated parameters (φ v and φ r,y ) are the same. In Table 1-3, most of these coefficients take the same signs. For instance, the existence of alumni ties has a positive effect on both subjective value added and funded startups’ growth. Thus, the source of the positive covariance is the observable startup and VC-related features. Selection on Private Information The above correlation is not accounted by any private information shared between a pair of startup and VC syndicate. In fact, the original model lacks a channel for private information to play a role. This is because all unobservable factors, absorbed by the three noise terms (,η,ξ), are assumed to be independent in the baseline model. To incorporate the effect of private information, I extend the model by revising the noise terms as follows. For startups funded at t− 1, the noise term in the growth variable att includes an additional termρ r,y ξ t−1 , and the noise term in the implicitexitvalueincludesanadditionaltermρ s,y ξ t−1 . Thecoefficientsρ r,y andρ s,y 31 are the covariances among the errors. They give the dependence of the subjective value added on the private information that will drive the startup growth and the implicit exit value one period ahead. Section 1.7 details the extended model and gives the estimation strategy. Table 1-7 presents the estimation results. [Insert Table 1-7] Now, a unit increase in the expected VC impact driven by the private infor- mation correlates with 0.9% increase in startup return following funding. It also correlates with 0.073 increase in the implicit exit value, which corresponds to a 48.74% relative increase in the probability of IPO/MA and a 90.90% relative decrease in the probability of Death. Therefore, private information could be an additional source to be added to the positive correlation between the expected VC impact and funded startups’ future performance. 1.4.4 Joint Effect of Selection and Influence The results of the joint estimation highlight one fact: selection and influence of VC funding are directly linked over time. On one hand, following the funding decision, VCs’ influence improves funded startups’ quality and their probabilities of successful exits. On the other hand, the selection of funding in turn is dependent on the expected VCs’ impact in the future, as well as the startups’ quality resultant from past VCs’ influence if funded before. A natural question then follows. How large is the joint effect of VC funding, given the direct linkage of selection and influence over time? To answer this question, I conduct a simulation experiment to study the joint effect of funding selection and influence over multiple rounds. More specifically, I compare the simulation results from two models. In Model 1, I break down the 32 interdependence between funding selection and influence, by making the funding decision random for all periods. In Model 2, the funding decision is random only for the first period. Model 1 : Selection is random for all periods. Influence follows the same dynamics as in the baseline model. Model 2 : Selection is random for the first period, then determined in equilibrium. Influence follows the same dynamics as in the baseline model. Forparametervalues,IusetheestimationresultsfromTable1-3. Thedynamics of the three key variables are the same as in the baseline model. I assume the economy has 100 startups born at time 0, with exactly the same features ex ante. There is only one VC syndicate, which can fund up to 50 startups each period. I assume all the startup and VC-related features take the mean values of the sample used in the above analysis. Note that there is no variation across the startups in the beginning. The only source that makes them different later on comes from the randomness in the funding decision in both models. Table 1-8 compares the simulated results for the two models. [Insert Table 1-8] The randomness in the initial funding decision magnifies significantly in Model 2, i.e., under the joint effect of selection and influence of VC funding. In Model 2, previously-funded startups have significantly higher period return and implicit exit value. In contrast, there is no significant difference between previously-funded and unfunded startups in Model 1. To show how the initial randomness accumulates over time in Model 2, I com- pare the transition matrix of VC funding. In Model 1, each startup has around 33 50% chance to get funded each period, whether it gets funded or not previously. However, in Model 2, a previously-funded startup has 93.1% chance to get funded this time, but the chance is only 4.1% for a previously-unfunded startup. Thus, the funded subsample in Model 2 does not change a lot over time – it always con- sists of a large proportion of startups which happen to get the funding in the first period, so they get the subsequent fundings as well. As a consequence, initial randomness might have a long-term effect given the interdependence between funding selection and influence. To verify the conjec- ture, I further compare the differences in future performance between funded and unfunded startups for the two models. Funded startups are not very different from unfunded ones in Model 1. However, they consistently over-perform in both short- term and long-term future in Model 2. To sum up, the joint effect of selection and influence of VC funding is essential. It helps accumulate the impact of some incidental factor over time which can become decisive in the end. 1.5 Robustness This section presents the results of some robustness checks. First, to account for the possible shifts in the parameter estimates, I perform the estimation on two subsamplesfrom1998to2006andfrom2007to2014. Second, tocurbthepotential multicollinearity problem, I extract factors from the original set of features and estimate the model using these factors. Third, I construct two alternative models to allow for the autocorrelation in the subjective value added and a hierarchical structure in startup returns driven by a hidden startup fixed-effect. 34 1.5.1 Subsample Results I perform the subsample estimation on two periods separately: from 1998 to 2006 and from 2007 to 2014. Both periods feature a flagging economy in the begin- ning and then a recovery afterwards. The first period captures the dot-com bubble, and the second period spans over the 2007-2008 financial crisis. Both subsamples only contain the startups that are born during the corresponding periods. For the first subsample running from 1998 to 2006, I adjust startups’ final status to “Survival” if they are going to be “IPO/MA” or “Death” in the second subsam- ple. Note that even if the model is correctly specified, there might be drift in the parameter estimates due to the evolution of the VC industry and changes in the startup ecosystem. Table 1-9 gives the estimation results. [Insert Table 1-9] Comparing the two periods, the effect of the cumulative return on the implicit exit value is much more significant in 2007-2014. It is true for both funded and unfunded startups: the coefficient jumps from 0.001 to 0.017 in the funded case, and it jumps from 0.016 to 0.048 in the unfunded case. Thus, the probability of IPO/MA depends more on startups’ quality in 2007-2014 and more so for the unfunded ones. Another difference is related to the number of locations covered by a VC syn- dicate. It has a significant effect on the selection of VC funding in 1998-2006 but the significance shifts to be on the influence of VC funding in 2007-2014. More specifically, the number of locations has a significant effect on the subjective value added in 1998-2006 and the effect fades away in 2007-2014. In contrast, its effect on startup growth becomes significant in the more recent subsample. It indicates 35 that VCs exert more efforts for funded startups more recently, while the vicinity to startups is more important for the investment decision in the early times. Additionally, there are two distinct differences for previously funded and unfunded startups separately. For the unfunded startups, the period returns rise with the risk-free rate in 2007-2014 but fall with it in 1998-2006. It implies a stronger substitutional effect in the later period but a strong income effect in the early period. For the funded startups, a top-school graduation dummy has a sig- nificant positive effect on startup growth only in 1998-2006. One explanation is that more entrepreneurs have to quit from schools in the early period in order to build their startups more seriously which got funding from VCs. Fewer have to do so more recently due to the advancement in the communication technology as well as the geographical expansion of the whole VC industry. By large, the parameter estimates do not shift a lot, and most importantly, the cumulative return continues to show positive effect on the implicit exit value as well as the subjective value added for both subsamples. 1.5.2 Alternative Features To curb the potential multicollinearity problem, I apply the principal factor method to extract common factors separately from the macroeconomic variables, the constant and time-varying startup features, the constant and time-varying VC features, and the constant and time-varying pair features. 25 Note that I include additional features (e.g. location, category dummies for the startups, and school, field, degree dummies for the VCs’ and startups’ personnel) but extract fewer factors to keep the total number of factors low. Table 1-10 Panel A gives the factor 25 The factor loadings are computed using the squared multiple correlations as estimates of the communality. By comparison, the principal component factor uses 1 for the communality for all pairs of variables. 36 loadings on the original features. The goal of this exercise is to double-check the interdependence between the selection and influence of VC funding rather than the effect of any individual feature. [Insert Table 1-10] As shown in Table 1-10 Panel B, the significance of many factors goes away for their effects on the growth of the funded startups. It results in a boost in the variance estimate. Both changes might be due to the reduction in the number of covariates. In contrast, the extracted factors still have significant effects on the growth of the unfunded startups and the subjective value added which determines the equilibrium funding decision. This might be due to the additional inclusion of many other features for the extraction of the factors. The effects of the cumulative return on the implicit exit value and the subjective value added remain positive and significant. 1.5.3 Alternative Models AR(1) in Subjective Value Added One extension of the model is to introduce autocorrelation in the subjective value added. Intuitively, the autocorrelation structure implies that how much a VC syndicate and a startup value the synergy they can create as a team today depends on how much they valued it yesterday. More specifically, I change the dynamics of v ij t given in Eq(4) to contain an additional term of the lagged value v ij t−1 . 26 Section 1.7 details the model and the estimation strategy. 26 Note that the introduction of the lagged value v ij t−1 is not supposed to substitute the effect of 1(funding tie). The latter dummy equals to one if any member VC has funded the startup before, and v ij t−1 is related to the probability of the syndicate as a whole to fund the startup in the previous period. 37 [Insert Table 1-11] Table 1-11 Panel A gives the parameter estimates. The AR(1) coefficient has an estimate of 0.009 and a t-value of 34.65. The effect is statistically significant but economically small. At the same time, the estimates barely change for Eq(4) which determines the subjective value added (e.g. the coefficient for the effect of the cumulative return on the subjective value added is 0.034 in Table 1-3 and 0.033 in Table 1-11 Panel A). Interestingly, the effect of the cumulative return on the implicit exit value decreases for both funded and unfunded cases even though Eq(2) remains exactly the same. The reason is that the system of equations (Eq(1), Eq(2), Eq(4)) needs to be estimated jointly. This again is due to the interdependence between the selection and influence of VC funding that the model intends to capture. A Hierarchical Model for Startup Returns It is well known that the startup return distribution has excessive kurtosis. Thus, several papers use Mixture-of-Normals to model the startup return distri- bution (Ewens 2009, Korteweg and Sørensen 2010). The underlying assumption is that there are different types of startups, e.g., winners, losers, and break-eveners, so the mean expected returns are different for different types. 27 On the other hand, theremightalsobeastartupfixed-effecttoaccountfortheinborndifferenceamong different startups. Therefore, I extend the baseline model to allow for a hierarchical structure on the startup returns such that the noise in the period returns samples from a Mixture-of-Normals. What’s new here is that the probability mixture is a hidden 27 The terms to describe the type of startups are borrowed from Ewens 2009. 38 startup fixed-effect and is different across startups. The extended model bor- rows from the Latent-Dirichlet-Allocation (LDA) model well-known in the area of machine learning. It belongs to the class of hierarchical models documented in Korteweg and Sørensen (2014) and more generally the class of Bayesian network models. Morespecifically, thehiddenstartupfixed-effectisastartup-specificprobability mixture. I denote it by p j ≡ p (1) j ,p (2) j ,p (3) j for startup j. The three elements of p j represent the probabilities that startup j is a winner, loser, and break-evener, respectively. For the funded case, the mean period returns for the three types are μ (1) y , μ (2) y , and μ (3) y ; for the unfunded case, they are μ (1) n , μ (2) n , and μ (3) n instead. Givenp j , startupj’s period return defined in Eq(C6) follows a Mixture-of-Normals distribution with the probability mixturep j . Section 1.7 details the model and the estimation strategy. Table 1-11 Panel B gives the parameter estimates. Compared with Table 1-11, the estimates for the old parameters stay qualitatively the same. For the new parameters, the mean growth rate for the previously funded case (i.e.,μ y ’s) for the winners, losers, andbreak-evenersis0.011, -0.005, and0.006respectively. However, none of them is significantly different from 0. In contrast, the mean growth rate for the previously unfunded case (i.e.,μ n ’s) for the three types is 1.564, -1.055, and 0.056 respectively. The mean returns for the winner and the break-evener types are significantly above 0, while for the loser type it is significantly below 0. The difference across the three types is economically big and more so for the unfunded case. [Insert Figure 1-3] Figure 1-3 plots the distribution of the hidden startup fixed-effect p j . The triangle represents the sample space, and each vertex on the triangle represents 39 a certain type. 28 As shown in the figure, there is a large mass concentrated on the line between the loser vertex and the break-evener vertex, and only a few startups have a probability greater than 0.5 to be become a winner. Even the one with the highest winning probability has a non-zero chance to have mediocre performance. Thedistributionmirrorsthefactthatmoststartupsendupmediocre or unsuccessful even though the returns for the successful ones are stunningly high. 1.6 Conclusion In this paper, I study how VCs select startups to fund over multiple rounds in a dynamic setting. The model I build highlights the interdependence between fund- ing decision and funding impacts. Using a hand-collected database including both VC-funded and non-VC-funded startups, I develop an estimation strategy using Gibbs sampler to jointly estimate the model parameters in a Bayesian framework. The results show that selection and influence of VC funding are directly linked over time. The selection of funding depends on both startups’ quality and VCs’ expected influence in the future. Following funding, VCs’ influence improves star- tups’ quality and their probabilities of successful exits. The results suggest a joint effect of selection and influence of VCs’ investments in a dynamic setting. Namely, funded startups have better quality, and thus they are more likely to get funded again in the future. A simulation experiment shows that under this joint effect, initial random differences in startups can magnify significantly over time. This paper illustrates the dynamic aspect of investment, for which post- investment activities are relevant for decision making. Using venture capital as an example, the paper highlights the importance of a joint consideration of both 28 The top vertex represents the winner type; the rightmost vertex represents the loser type; the leftmost vertex represents the break-evener type. 40 investee’s value ex-ante and investor’s influence ex-post. A frequently invested item becomes a valuable asset to the whole economy as it capitalizes the impacts of its past investors. While absent from a static setting, these new insights shed light on some fundamental issues of strategic investment behavior. 1.7 Appendix 1.7.1 Estimation Procedure Prior Distributions The prior distribution assumptions are as follows. The parameters in Eq(1) and Eq(2), i.e., φ r,y ,σ 2 r,y , φ r,n ,σ 2 r,n , φ s,y ,σ 2 s,y , and φ s,n ,σ 2 s,n , have Normal- Inverse-Gamma priors (with unknown variances σ 2 to be estimated). φ k,m |σ 2 k,m ∼N 0,σ 2 k,m A −1 k,m , σ 2 k,m ∼ Γ −1 (a k,m ,b k,m ) (A1) where k = “r” or “s” to denote the dependent variable, and m = “y” or “n” to denote the answer to whether the startup is previously funded. The parameter in Eq(4), i.e., φ v , has a Normal prior (with known variance equal to 1). φ v ∼N 0,A −1 v (A2) Note that the prior means ofφ’s are assumed to be zero so that the null hypoth- esis is that all coefficients are insignificant. A’s are diagonal matrices with all ele- ments equal to 1/100, and a’s and b’s are set to be 2.0 and 1.0. The assumption on the prior distributions is to form a Bayesian Linear Regression setting which 41 gives tractable posterior distribution. Please see Korteweg (2013) for a detailed description of the setting as well as the rule for parameter estimation. Algorithm for Estimation (Baseline Model) For parameter learning, I develop a parallel Gibbs Sampler to draw from the posterior conditional distributions given the data and the prior distributions. In particular, I factorize the joint posterior distribution into a full set of conditional distributions of (1) the latent variables r,s,v, and (2) the parameters φ r,y ,σ 2 r,y , φ r,n ,σ 2 r,n , φ s,y ,σ 2 s,y , φ s,n ,σ 2 s,n , andφ v . The algorithm consists of the following six steps to be performed iteratively. For initial values, φ’s are set to 0 and σ 2 ’s are set to 1.0. Steps 1. Impute r j t given s j t , n v ij t ,i∈I o , parameters and data 2. Impute s j t given r j t , parameters and data 3. Imputev ij t (all together for eacht) given n r j t ,j∈J t o , the equilibrium condi- tion, parameters, and data 4. Update φ r,y ,σ 2 r,y and φ r,n ,σ 2 r,n given all r and data 5. Update φ s,y ,σ 2 s,y and φ s,n ,σ 2 s,n given all s, r, and data 6. Update φ v given all v, r and data The following paragraphs give the detailed information for the steps 1 to 6. I use some new notations to simplify the description. 42 Notations • φ r : equals φ r,y or φ r,n for previously funded or unfunded • σ 2 r : equals σ 2 r,y or σ 2 r,n for previously funded or unfunded • φ s : equals φ s,y or φ s,n for previously funded or unfunded • σ 2 s : equals σ 2 s,y or σ 2 s,n for previously funded or unfunded • φ s,1 : the vector of φ s except the last element • φ s,−1 : the last element of φ s • φ v,1 : the vector of φ v except the last element • φ v,−1 : the last element of φ v • X j t : equals h 1,X t ,X j t ,X i t ,X ij t i if j is funded by i at t− 1, and equals h 1,X t ,X j t i if j is not funded at t− 1 • X ij t : equals h X i t ,X j t ,X ij t i for a pair of i and j • μ t (i) ={j :ij∈μ t } • μ t (j) ={i :ij∈μ t } 1. Impute r As in Korteweg and Sørensen (2010), r is imputed using a FFBS (forward filtering and backward smoothing) method. Given infrequent observable values of r, this method samples interim values given the information that is cor- related with these interim values. Note that in Eq(2) and Eq(4), both the implicit exit value s and the subjective value added v depend on r. Therefore, s and v generate the information set for the conditional distribution of r. I use m j t|τ and P j t|τ to denote the conditional mean and variance ofr j t given information generated 43 bys andv up to timeτ. Below gives the forward and backward steps for the FFBS method. • Forward – Forecast m j t|t−1 =m j t−1|t−1 +X j t φ r (A3) P j t|t−1 =P j t−1|t−1 +σ 2 r (A4) – Update b =    φ s,−1 φ v,−1    (A5) e =    s j t −X j t φ s,1 P i v ij t −X ij t φ v,1 /I   −bm j t|t−1 (A6) K =P j t|t−1 b   bP j t|t−1 b 0 +    σ 2 s 0 0 σ 2 v /I       −1 (A7) m j t|t =m j t|t−1 +Ke (A8) P j t|t−1 = (1−Kb)P j t|t−1 (A9) • Backward 44 – Given the draw of r j∗ t+1 G =          P j t|t P j t|t +σ 2 r,y −1 if j is funded at t P j t|t P j t|t +σ 2 r,n −1 if j is unfunded at t (A10) M =m j t|t +G r j∗ t+1|t (A11) V =P j t|t (1−G) (A12) – Draw r j∗ t ∼N(M,V ) Notethatgivens,v, andthedata, theconditionaldistributionsof n r j t : 0≤t≤T o are independent across j. Therefore, the FFBS procedure can be performed in a parallel fashion for all startups. 2. Imputes The distribution ofs follows truncated Normal givenr and startup status (i.e., IPO/MA, Death, or Survival), and it is conditionally independent of v. By assumption, VC’s appearance also affects the startup’s status. Here, I let δ = 3. Using the same notation, s is sampled as follows. M = h X j t ,r j t i φ s , V =σ 2 s (A13) • Status j t = IPO/MA: draw s j t ∼N(M,V )× 1[δ≤s j t ] • Status j t = Survival: draw s j t ∼N(M,V )× 1[−δ≤s j t <δ] • Status j t = Death: draw s j t ∼N(M,V )× 1[s j t <−δ] Here, 1[.] denotes the indicator function. The imputation of s can be performed parallely for all startups and for all time periods. 45 3. Impute v The distribution of v ij t is also one-dimensional truncated Normal given r j t and v −ij t which is defined as the collection n v i 0 j 0 t :i6=i 0 or j6=j 0 o . As in Sørensen (2007), the conditional distribution of v ij t depends on whether j is matched with i at time t. More specifically, v is sampled as follows. M = h X ij t ,r j t i φ v (A14) • Matched Draw v ij t ∼N(M, 1)× 1 h v≤v ij t i , where v is given in Eq(10). • Unmatched Draw v ij t ∼N(M, 1)× 1 h v ij t <v i , where v is given in Eq(9). Again, the imputation of v can be performed in a parallel fashion for all time periods. However, within a specific time period, the v’s for all matched pairs need to be imputed first (either parallely or sequentially) since the values for the unmatched pairs will depend on those of the matched pairs. 4. Update φ r and σ 2 r • Update φ r,y and σ 2 r,y The subsample includes all previously-matched funding candidates. Let N y be the size of the subsmaple. Using the Bayesian Linear Regression rule, we can sample φ r,y and σ 2 r,y as follows. Here, X and y represent the stacks of 46 the independent and dependent variables for the previously-matched case in the linear regression in Eq(1). a =a r,y +N y /2 (A15) b =b r,y + h y 0 y−G 0 (X 0 X +A r,y ) −1 G i /2 (A16) G = (X 0 X +A r,y ) −1 X 0 y (A17) First draw σ 2 r,y ∼ Γ −1 (a,b), then draw μ r,y |σ 2 r,y ∼ N G,σ 2 r,y (X 0 X +A r,y ) −1 . • Update φ r,n and σ 2 r,n The subsample includes all previously-unmatched funding candidates, letN n be the size of the subsample. Using the Bayesian Linear Regression rule, we can sample φ r,n and σ 2 s,n as follows. Here, X and y are the stacks of the independent and dependent variables for the previously-unmatched case in the linear regression in Eq(1). a =a r,n +N n /2 (A18) b =b r,n + h y 0 y−G 0 (X 0 X +A r,n ) −1 G i /2 (A19) G = (X 0 X +A r,n ) −1 X 0 y (A20) First draw σ 2 r,n ∼ Γ −1 (a,b), then draw μ r,n |σ 2 r,n ∼ N G,σ 2 r,n (X 0 X +A r,n ) −1 . 5. Update φ s and σ 2 s • Update φ s,y and σ 2 s,y The subsample is the same as above for the update of φ r,y and σ 2 r,y . The 47 difference is that X and y now represent the stacks of the independent and dependent variables for the previously-matched case in Eq(2). a =a s,y +N y /2 (A21) b =b s,y + h y 0 y−G 0 (X 0 X +A s,y ) −1 G i /2 (A22) G = (X 0 X +A s,y ) −1 X 0 y (A23) First draw σ 2 s,y ∼ Γ −1 (a,b), then draw μ s,y |σ 2 s,y ∼ N G,σ 2 s,y (X 0 X +A s,y ) −1 . • Update φ s,n and σ 2 s,n Again, X and y now are the stacks of the independent and dependent vari- ables for the previously-unmatched case in Eq(2). a =a s,n +N n /2 (A24) b =b s,n + h y 0 y−G 0 (X 0 X +A s,n ) −1 G i /2 (A25) G = (X 0 X +A s,n ) −1 X 0 y (A26) First draw σ 2 s,n ∼ Γ −1 (a,b), then draw μ s,n |σ 2 s,n ∼ N G,σ 2 s,n (X 0 X +A s,n ) −1 . 6. Update φ v The subsample includes all pairs of VCs and funding candidates. The Bayesian Linear Regression now does not include the noise variance. Here X and y represent the stacks of the independent and dependent variables in Eq(4). G = (X 0 X +A v ) −1 X 0 y (A27) Draw μ v ∼N G, (X 0 X +A v ) −1 . 48 Algorithm for Constrained Estimation (Baseline Model) Now I impose the constraint that in Eq(1) and Eq(2), the parameters that are not associated with the VC-related features are the same for funded and unfunded startups. Equivalently, Eq(1) and Eq(2) become the following. 29 r j t −r j t−1 = h 1,X t ,X j t ,X i t ,X ij t i φ r,y +σ r,y j t , for all j∈E t−1 (A28) s j t = h 1,X t ,X j t ,X i t ,X ij t ,r j t i φ s,y +σ s,y η j t , for all j∈E t−1 (A29) With X i t = 0 and X ij t = 0 if j is not funded at t− 1. It is straightforward to change the algorithm in Section 1.7 for the estimation here. Themodeldoesnothave φ r,n ,σ 2 r,n and φ s,n ,σ 2 s,n , soallthesubsamplesin E t−1 are used for the estimation of the parameters φ r,y ,σ 2 r,y and φ s,y ,σ 2 s,y , with the above augmentation of the independent variables (i.e., X i t = 0 and X ij t = 0) for the previously-unfunded startups. 1.7.2 Measure Construction Dependent Variables • r: Cumulative Return • s: Implicit Exit Value • v: Subjective Value Added Independent Variables • Macroeconomic Variables 29 Recall thatE t−1 represents the set of existing startups at the end oft−1, or at the beginning of t. 49 • Startup Features • VC Syndicate Features • Startup - VC Syndicate Pair Features Cumulative Return r = cumulative return = log(V ) • For newborn startups: r = log(V ) = 0, V = 1. • For existing startups: – IPO: r = log(V ), with V = market value at IPO. – MA: r = log(V ), with V = acquired price at MA. – Death: r = log(V ), with V ∼ triangle distribution with a = 0.05, b = 0.1, and c = 0.8. • At funding round: r = log(V ) = P t log V PRE t /V POST t−1 , so V = Q t V PRE t /V POST t−1 , where V POST = I +V PRE . Here, V is the anti-diluted valuation of the startup at t, I is the investment amount. Macroeconomic Variables • (Y baa −Y us10 ) = (Moody’s seasoned Baa corporate bond yield)− (10-year Treasury bond yield). • (r m −r f ) = monthly market excess return over risk-free rate. • smb = monthly factor return for the small-minus-big portfolio. • hml = monthly factor return for the high-minus-low portfolio. • r f = monthly risk-free rate. 50 Startup Features Constant • # locations = number of cities that a startup’s headquarter or offices are located in. • # categories = number of categories that a startup is classified into. • # products = number of products that a startup has. • 1(LOC) = dummy variable indicating whether the startup has its headquar- ter or offices in LOC, where LOC is – ca: state of California – ny: state of New York – ous: other places in U.S. except from New York and California – ona: other places in North American except from U.S. – as: Asia – eu: Europe Time-varying • t from last round = time since last funding round in years at a specific time t. • t2 from last round = square of (t from last round) at a specific time t. • # rounds = number of funding rounds experienced in the past at a specific time t. 51 • # startups founded = number of companies the founder of the startup has built in the past prior to a specific time t. • 1(top20 school) = dummy variable indicating whether the startup at a spe- cific time t has people on the management team who graduated from a top school. The list of top schools (for startups) is given in Table 1A-2. VC Syndicate Features Constant • # of VCs = number of VC members in a VC syndicate. • # locations = number of cities that a VC syndicate has at least one member VC that has an office or headquarter. • 1(LOC) = dummy variable indicating whether the VC syndicate has at least one member VC has its headquarter or offices in LOC, where LOC defined as above in startup features. Time-varying • 1(cooperated) = dummy variable indicating whether any member VCs have cooperated in the past prior to a specific time t. • # categories = number of categories that the VC syndicate has at least one member VC that has funding experience prior to a specific time t. • # rounds = median funding rounds that VC members have participated in prior to a specific time t. It measures the average experience of the VC syndicate. 52 • 1(top20 school) = dummy variable indicating whether a VC syndicate at a specific time t has people on its member VCs’ management teams who graduated from a top school. The list of top schools (for VCs) is given in Table 1A-4. Pair Features Constant • distance = The closest distance in miles between a startup and a VC syndi- cate. • days of travel = number of days for a round travel between a startup and a VC syndicate using the closest distance defined above, where the number of days equals – 0: if distance∈ [0, 100], indicating a round travel by driving within one day – 1: if distance∈ [100, 1000], indicating a round travel by flight within two days – 2: if distance∈ [1000, 10000], indicating a round travel by flight within three days – 3: if distance∈ [10000,∞), indicating an intercontinental travel Time-varying • 1(funding tie) = dummy variable indicating whether any VC member in the syndicate has funded the startup prior to some specific time t. • 1(alumni tie) = dummy variable indicating whether any VC member and the startup have any alumni ties at some specific time t. 53 1.7.3 Alternative Models Autocorrelation in Subjective Value Added The subjective value added at t might depend on its value at t− 1. Therefore, one extension of the baseline model is to includev ij t−1 in the expression ofv ij t . With Eq(1) and Eq(2) unchanged, Eq(4) changes to the following. v ij t = h X i t ,X j t ,X ij t ,r j t ,v ij t−1 i φ v +ξ ij t , for all i∈I,j∈J t (C1) The new parameter introduced is the last element inφ v that is associated with v ij t−1 . For estimation, there are some small changes in the imputation of v and in the update step of the FFBS method for the imputation of r. Funding Decision Incorporating Private Information When making the funding decision, venture capitalists may have some private information on startups that are unobservable to an outside economist. This pri- vate information will drive startup growth and implicit exit value att+1, and thus is incorporated in the subjective value added at t to make the funding selection. Therefore, the model can be extended as follows. With Eq(4) unchanged, I modify Eq(1) and Eq(2) to the follows for startups that are previously funded at t− 1 to incorporate the “private information” ξ ij t−1 in the subjective value added at t− 1. r j t −r j t−1 = h 1,X t ,X j t ,X i t ,X ij t i φ r,y +ρ r,y ξ ij t−1 +σ r,y j t , if j is funded by i at t− 1 (C2) s j t = h 1,X t ,X j t ,X i t ,X ij t ,r j t i φ s,y +ρ s,y ξ ij t−1 +σ s,y η j t , if j is funded by i at t− 1 (C3) 54 Note that the difference here is that we now includeξ ij t−1 as the last independent variable for both equations. The associated parameters are denoted by ρ r,y and ρ s,y respectively. The correlations between v ij t−1 and r j t −r j t−1 , and between v ij t−1 and s j t now also capture the private information shared between startups and VC syndicates which is not loaded on publicly observable features. The changes in the estimation strategy stem from the imputation of r and s for the previously funded case, the update of φ r,y ,σ 2 r,y and φ s,y ,σ 2 s,y , and the imputation of v for matched pairs given next-period r and s for those pairs. Hierarchical Model for Returns The extended model has a hidden startup fixed-effect p j ≡ p (1) j ,p (2) j ,p (3) j , which is a startup-specific probability mixture. It consists of the probabilities that a startup belongs to a specific type: winner, loser, and break-evener. The probabilities are denoted by p (1) j , p (2) j , and p (3) j . The period return for a specific type follows a Normal distribution with different means: μ (1) y , μ (2) y , and μ (3) y for the three types with funding, and μ (1) n , μ (2) n , and μ (3) n for the three types without funding. More specifically, Eq(1) changes to the following. r j t −r j t−1 =          γ j y,t + h X t ,X j t ,X i t ,X ij t i φ r,y +σ r,y j t if j is funded by i at t− 1 γ j n,t + h X t ,X j t i φ r,n +σ r,n j t if j is unfunded at t− 1 (C6) with γ j y,t ≡ n μ (k) y : with probability p (k) j o , for k = 1, 2, 3 (C7) γ j n,t ≡ n μ (k) n : with probability p (k) j o , for k = 1, 2, 3 (C8) 55 Here, γ j y,t and γ j n,t are i.i.d. and follow categorical distributions with the com- mon probability mixture p j . Equivalently, both noises in the period returns γ j y,t +σ r,y j t and γ j n,t +σ r,n j t follow Mixture-of-Normals with the common probability mixture p j . I assume the priors of p j follow a Dirichlet distribution as follows. p j ∼ Dirichlet (α) (C9) Let z j t denote the realized type (winner, loser, break-evener) for a startup j at time t then P z j t =k =p (k) j . Eq(C6) can be re-written as follows. r j t −r j t−1 =          μ (z j t ) y + h X t ,X j t ,X i t ,X ij t i φ r,y +σ r,y j t if j is funded by i at t− 1 μ (z j t ) n + h X t ,X j t i φ r,n +σ r,n j t if j is unfunded at t− 1 (C10) Consequently, the estimation for the extended model includes the imputation of the two additional variables, p j and z j t , and the update of the two additional sets of parameters μ y = μ (1) y ,μ (2) y ,μ (3) y and μ n = μ (1) n ,μ (2) n ,μ (3) n . 56 Chapter 2 Is Collaborative Activism Effective? 2.1 Introduction “Southeastern Asset Management and Carl C. Icahn Urge Special Committee and Board of Directors to Act in Best Interest of Stockholders and Move Dell Forward.” — PR Newswire 7/18/2013 “Fresh off the disappointment of its attempt to thwart Michael Dell’s privati- zation plan with Carl Ichan, Memphis-based investment firm Southeastern Asset Management has made another huge play on News Corp.” — PR Newswire 9/13/2013 Icahn’s unsuccessful joint attempt with Southeastern Asset Management to take Dell public raises several important questions on activism. Do activists and fund managers team up to change corporate governance? Is activism through teamwork more likely to succeed? How is the operating performance of target firms improvedbyactivists’collaborativeeffort? Thepastliteraturehasconcentratedon the effects of activism initiated by individuals or investment funds. None identifies the collaboration between activists and activist funds and their impact on target firms. 57 Blockholders have strong incentives to collaborate because working together may be more powerful to enhance negotiation stance and force target firms to meet demands. Whenactivistshaveconsolidatedholdingsonthesamefirm, theysharea common incentive to monitor and influence target firms’ corporate governance and performance. For an activist or a fund manager who holds only a few percent of shares in the firms where she would like to pressure corporate boards and managers for change, teaming up with other activists helps accomplish her objectives, which is implausible if demanding alone. However, it is unclear whether working collaboratively is effective. Co-activism may be more likely to succeed in maximizing shareholder wealth than a single activist due to closer monitoring, more efficient resource allocation, more reduc- tion in agency costs, lower monitoring costs due to information sharing, and a stronger stance to negotiate with firms’ management. An activist might be less interested in working alone because she only enjoys a fraction of benefits from mon- itoring but has to bear the full cost, while co-activists can share the cost. Because co-activists share common capital at stake and common goals to enhance share- holder value, they intend to act as effective monitors, vote in support of the same proposals, and share resources and information. Moreover, co-activism can be a means of power. Aligning with other activists has more power to demand change in management, board structure, and operating efficiency. The exploited power might better mitigate the agency conflict between managers and shareholders and help succeed in turning poorly performing firms around. Conversely, potential costs of co-activism are likely to outweigh the benefits. Co-activism can cause three layers of agency problems. First, co-activism can further increase the separation between ownership and managers. Collaborative individuals or entities may attempt to exert suboptimal effort and act in their 58 own best interest other than siding with collaborators to maximize shareholder value. Also, collaborators are not necessarily effective monitors. Some activists vote their own shares while others typically delegate their proxy voting to outside fund managers. Second, agency issues can be worsened by misaligned incentives between activists. For instance, fellow activists may free ride on the lead activist’s fil- ing (Grossman and Hart (1980)), and thus, no additional value is added on co- activism. Information asymmetry may worsen when working with another activist due to barrier to transfer information and each activist’s private protocols and tactics to improve managements’ decisions and firms’ operations. This communi- cation problem, along with multi-layered moral hazard and asymmetric informa- tion, would make it harder to extract benefits from a larger information pool. If one activist has to meet redemptions from investors regularly, selling the underper- formingfirms’sharesinthesecondarymarketismoreeffectivethancommunicating with its collaborator and monitoring firms’ management for months. Furthermore, since activists face different levels of regulatory restrictions and compensation structure and need to deliver stated objectives to investors, their incentives may be ill-aligned and each activist ends up overprotecting their own large stake on investment. Third, co-activism can be viewed as undesirable from small shareholders’ perspective, but co-activists could extract private benefits from it. Collaboration could fail because prolonged fight and news coverage by multiple activists could signal their own greed and manager’s reluctance to change. The increased conflict of interest could trigger anger from fellow shareholders. They may feel uneasy about whether co-activists have sufficient expertise, experience, or skills to put their hands on management. 59 We find that co-activism targets poorly performing firms. Target firms experi- encesignificantlynegativestockreturns, lowoperationalperformanceandaccount- ing profitability, cash flow constraint before the intervention by co-activists, com- pared with industry/size/book-to-market matched firms. Payout policies at these firms are more restricted than that of matched firms. A small fraction of co- activists intends to take control of target firms, but the majority holds around 20 percent of stocks with the maximum capital at 25 percent. Given that co-activists may desire to acquire a large stake quickly, we do not find that targets carry higher trading liquidity and institutional ownership. This might suggest that target firms are oversold by retail investors, but co-activists are confident in turning these firms around. Despite joint effort by activists, we find that most co-activism cases are non- confrontational. Only about 15% of co-activism involves proxy fights, litigation, or a hostile bid to take control of the company. Over 50% of the co-activism cases initiate open communication with the board and management on a regular basis. This finding is consistent with McCahery et al. (2014), who document that com- municating with discussions with management are the most widely used channel to directly intervene the management. Item 4 in 13D filings states the intended goal of an activist. About one third of co-activists intend to demand seats on the board of directors and lobby for better board governance, and another one third attempt to maximize shareholder value. Although co-activism often adopts a soft stance to negotiate with target firms’ management, we also observe that the success and partial success rates of achieving its stated goal are quite high. It has been a long debate whether shareholders influence firm management. Smith (1996) shows that shareholder activism successfully changes corporate gov- ernance structures and improves shareholder values. On the other hand, Wahl 60 (1996) examines pension fund activism and concludes that activism adds little or no values to target firms. More recently, researchers have found that hedge fund activism improves shareholder wealth (e.g. Brav et al. 2008 and Klein and Zur 2009). The conflicting results between pension funds and hedge funds come from the regulation, compensation structure, incentives, and conflicts of interest (see Brav et al. 2010). Given the long history of our sample, we are able to study both short- and long- term effect of co-activism. One criticism on hedge fund activism is that its positive impact on target firms is short-lived. Contestants argue that the long-term costs of activism are not reflected in stock prices yet one or two years after the filing date and thus market inefficiency drives initial stock prices up and stock return should underperform in the long term. Policy makers have been lobbying for limiting the role, rights, and involvement of shareholder activists on these grounds. The market expects co-activism to succeed. We observe that abnormal stock returns during the (-20,+20) announcement window around the Schedule 13D file date range from 6% to 7%. Although we do not observe evidence of a negative abnormal drift during the one year period subsequent to the announcement, the abnormal returns do drop off gradually, except for the market-adjusted case. One year after the attack, cash flow, total assets, revenues and market equity are sig- nificantly lower than that of matched firms and other financial and accounting measures signal no significant improvement. Five-year buy and hold abnormal returns are significantly negative and operation performance continues to decline five years subsequent to the intervention. In particular, declining cash flows over time provide evidence that co-activists affect the cash flows that eventually accrue to all shareholders, which in turn induces negative abnormal stock returns in the 61 longrun. Weconcludethatinvestorsoverestimatethepositiveeffectofco-activism, that myopic or short-sighted co-activism is supported by the data. Our findings on target firms’ diminishing operating and stock return perfor- mance are somewhat contrast to past studies on activism. These studies typically provide short term evidence on stock reaction and changes in operating perfor- mance prior and after the 13D filing date and the effect of activism persists over time. Earlier studies show that activism agenda initiated by pension funds and mutual funds fail to increase shareholder value (e.g. Black (1998), Karpoff (2001), and Gillan and Starks (2007)). Brav et al. (2008) and Klein and Zur (2009) docu- ment positive market reaction to hedge fund activism after the Schedule 13D filing date. Bebchuk et al. (2013) show that hedge fund activism improves operating performance during the five-year period during the intervention. In contrast, we find that co-activism is short term in focus as the operating performance deterio- rates and abnormal returns drop one year after activist investors work side by side to demand changes on target companies. Note that one key element in our definition of co-activism is independent indi- viduals or entities. Under Exchange Act Section 13(d) 1 , individuals or entities are required to be “acting in concert” or filing as group when they are related. 2 For example, father and son, who both hold shares more than 5%, have to file as a group in a single filing. A blockholder is required to file jointly with the firms that she has partnership with if these firms also hold shares in the same company. These cases are not considered collative activism based on our definition. In other words, we do not simply claim a filing with multiple filers as collative activism. We require collaborates to be independent, such as Carl Icahn and Southeastern 1 More details on SEC rules are explained in Section II. 2 Roughly 47% of the 13D and 13D/A filings from 1994-2013 have multiple filers. 62 Asset Management, so that conflicts between them can arise and possibly exac- erbate multi-layer agency problems among management, co-activists, and small shareholders. Another distinction between our definition of group membership and that of the Section 13(d) is that a co-activism case may involve an individual or entity that does not own shares of a subject company more than 5% or an individual or entity that is required to file separately under 13(d). Section 13(d)(3) provides that a “group” is formed when “two or more persons act as a partnership, lim- ited partnership, syndicate, or other group.”, but does not explicitly require any beneficial ownership. Therefore, a person may hold less than 5% of the shares of a subject company and is not required for a 13D filing, but act collectively as a group to monitor management and optimize firm value. We contribute to the literature in the following ways. First, we identify co- activism in a large sample dataset from 1994 to 2011. The failure of activists’ activities in the past might inspire the collaboration between activists and fund managersinrecentyears. Thelongtimewindowofoursampleallowsustoexamine any pattern of co-activism over time and its short- and long-term impact on target firms. We document a new way that large-percentage shareholders interact with one another and a distinct pattern of market reaction on co-activism between short run and long run. Second, to our knowledge, no paper has studied the effect of co-activism on target firms. Most studies focus on individual or institutional shareholder activism and their influence on target firms. Our evidence concludes that the efficacy of co-activism is different from other institutional activists and quite limited. This result might explain that no co-activism activities come in waves. 63 Third, except for hedge funds, the literature shows that institutional activists target weak firms, but have little impact on their operating efficiency and stock performance. We find that co-activism is mostly nohostile and has a roughly 80% success rate. The market responds to it positively in the short run, suggesting that the market expects the success of co-activism. Fourth, our evidence has important implications for the policy debate about the definition of “group”. If co-activism aims for short-term gains at the expense of long-term shareholder value, it should be classified as a “group” in Section 13(d) in a joint filing so that investors can be aware of this collaborative effort. But some of our co-activism cases do not file jointly as a group. This might create an agency problem for investors as they demand more transparency in firm actions. Last, a long debate over shareholder activism is whether shareholders should be granted for more power to intervene corporate decisions and hold directors accountable (e.g. Bebchuk (2005)). The elected representatives on the board of directors, alongwithdispersedownership, mightcontributetoweakshareholdersin the U.S. companies (e.g. Roe (1994)). The long-term loss in co-activism questions whether demanding changes in governance arrangements through the board is sufficient, but more shareholder power to intervene corporate decisions directly may be harmful. 2.2 Literature Review Shareholder activism has gained prominence since 1980. Institutional investors attempt to change the structure and decisions of the firms by proxy proposals in order to maximize shareholder values. However, earlier shareholder activism is 64 handicapped by regulations and institutional investors can only rely on proxy pro- posals to communicate with other shareholders. In 1992, Securities and Exchange Committee (SEC) remove the constraints on communication among shareholders. Because the cost of communication is reduced, large shareholders begin to negoti- ate with firm management directly instead of proxy proposals. Pension funds start the early wave of shareholder activism by institutional investors in the early 1990s. Pension funds support retirement plans, and thus, have a long-term investment horizon and penalize early withdrawals. As such, pension fund managers have an incentive to submit shareholder proposals to pres- sure the management. However, the effective of pension fund activism is mixed. Wahal (1996) and Karpoff et al. (1996) find little evidence on abnormal operating and stock performance of the target firms. Guercio and Hawkins (1999) study the largest and most active pension funds and find pension fund activism effective in share value maximization. Mutualfundsarelesslikelytoengageinshareholderactivismbecausemanagers often face liquidity demand from investors. Instead of asking for the management to adopt value-increasing policies or restructure the board, mutual fund managers can quickly sell the underperforming stocks in the open market. In addition, mutual funds are required to report investment holdings quarterly. Since voluntary disclosure increases transparency in investments, fund managers may prefer to unloading underperforming stocks rapidly. Yermack (2010) conclude that mutual fund activism has limited impact on firm governance and performance. Due to redemption restrictions, hedge funds can invest in illiquid assets, use leverage, adopt dynamic trading strategies, concentrate investments on a small set of firms, and take relatively long-term positions in underperforming companies. Since hedge fund managers face high water mark provisions and sometimes invest 65 personal wealth in the fund, they have stronger performance incentives than other institutional types. Hedge funds have lower conflicts of interests than other insti- tutional types because they normally have no or limited business relations with the target firms. No holdings disclosure requirements in hedge funds may allow hedge fund managers to hide trades from investors. Based on the these reasons, hedge funds are more effective in reducing agency conflicts between managers and shareholders, monitoring firm operations and performance, and forcing manage- ments and boards to change. For example, Brav et al. (2008) and Klein and Zur (2009) find that hedge fund activism achieves higher abnormal return and leads to permanent improvement in shareholder value. The literature on collaboration is mostly restricted to its effect on firm produc- tivity. However, studies on collaboration among institutional investors are limited. Hamilton et al. (2003) provide evidence that teamwork improves worker produc- tivity on average, but its effect diminishes as more workers join due to free-riding. Bonatti and Horner (2011) find that free-riding leads to a reduction in effort and the declining adoption of teams over time. However, there are limited applications in finance. But interestingly, in the past year or so, it becomes popular in the media that some institutional investors align themselves with other institutional investors or individual activists like Carl Icahn, Bill Ackman, David Einhorn, Daniel Loeb, and Ralph Whitworth in demanding any reform on business strategies, organizational spin-offs, firm governance, board independence, and stock performance. An individual or entity that files 13D filings with others for the same pur- poses needs to act as a group under the Exchange Act. Section 13d-3 defines a beneficial owner as “any person who, directly or indirectly, through any contract, arrangement, understanding, relationship, or otherwise has or shares voting power 66 which includes the power to vote, or to direct the voting of, such security” and/or “investment power which includes the power to dispose, or to direct the disposi- tion of, such security.” Section 13(d)(3) further defines a “group” as “person” and focuses on collective efforts by a group of individuals or entities for the purpose of acquiring holding, or disposing of securities. It provides that: “When two or more persons act as a partnership, limited partnership, syndicate, or other group for the purpose of acquiring, holding, or disposing of securities of an issuer, such syndicate or group shall be deemed a ‘person’ for the purposes of this subsection”. The purpose of Section 13 (d) is to help investors acquire pertinent and trans- parent information that a company can be influenced or controlled by a group of large shareholders and make investment decisions accordingly. The requisite disclosure as a group is designed to prevent a group of persons from colluding to influence management in favor of their common interests. However, acting in con- cert as a group is not necessarily equivalent to filing jointly under the Exchange Act. Some of our co-activism cases indeed file separately, which may imply lack of transparent information on collectively owned beneficial ownership to investors. This evidence challenges the definition of group membership in the 13D filing rules. 2.3 Data We collect activism data from 13D and 13D/A filings from 1994 to 2013. Investors are required to file a schedule 13D with the SEC within 10 days after acquiring more than 5% of any publicly traded company if they have an intention to influence the firm’s management. A Schedule 13D/A is an amendment filing by the same investor for the same firm and is filed subsequent to the initial Schedule 13D. Our data period overlaps with the periods in Brav et al. (2008) and Klein and 67 Zur (2009). Our long data offer an advantage to study the time series of activism such as the success of activism over time. There are no specific flags in the filings to identify co-activists, and thus the only way to identify co-activism cases is to hand collect data from all 13D and 13D/A filings. We first screen out filings that include the following strings: “in concert”, “joint solicitation”, “file jointly”, and “act jointly”. 3 We read through these filings and check them with websites, News sources, Factiva, and newspaper and magazine articles to identify co-activism cases. We have 195 target firms in our sample, which is comparable to the number of hedge fund activists in Klein and Zur (2009). From 13D and 13D/A filings, we record the identity of the filer, filing date, ownership and its changes, cost of purchase, invested capital and percentage of ownership (the size of the activists’ stakes in their target firms, both in dollar value (at cost)), and as a percentage of outstanding shares of the target.), and the purpose of the investment (from Item 4 “Purpose of Transaction”). To find the motives and results of co-activism and the target firm’s response, we check with Factiva, news sources, and newspaper and magazine articles for the following text strings: “fund and shareholder activist” or “fund and shareholder activism” or “fund and 13D”. To remove blocktrades due to investment purpose only, we first remove 13D and 13D/A filings specifying the purpose of investment as “investment only” in Item 4. We also remove events from 13G. Fund managers are required to file 13G if they invest in more than 5% on a firm and hold no intention to influence 3 These are common phrases appearing in legal documents or news articles. For example, on May 10, 2013, the 13D filed by Carl Icahn states the following: The Reporting Persons have agreed to act in concert with Southeastern solely for the purposes of promoting the proposals contained in the May 9 Letter, including urging shareholders to vote against the proposed going private transaction, and the joint solicitation of proxies for the Issuer’s 2013 annual meeting. 68 the firm or its management. Following Brav et al. 2008, we further drop filings whose primary purpose is related to bankruptcy reorganization or the financing of a distressed firm, a merger and acquisition-related risk arbitrage, and investment purpose only 4 , or the filings whose target is a closed-end fund or other nonregular corporation. To determine whether entities are independent, we look up activists’ company information from Item 2 in Schedule 13D. We classify active investors as indi- viduals, hedge funds, mutual funds, pension funds, private equity/venture capital funds, banks, and others from Item 2 in Schedule 13D. The identification codes of reporting person we pick include IV (Investment Company), EP (Employee Ben- efit Plan or Endowment Fund), PN (Partnership), IA (Investment Advisor), IN (individual), or BD (Broker-dealer). To remove 13D and 13D/A filings associated with Merger and Acquisition, we further remove filings whose filer identifications with CO (Corporation) only, both CO and PN, or both CO and IN. Klein and Zur (2009) group individuals, venture capital firms, and private equity or asset management groups that invest for wealth investors as other private investors. We restrict the target firms to be US firms. We check the identity of the activists and activist funds through the website. We acknowledge our search process may not be perfect, but we try our best effort to correctly classify almost all, if not all, co-activism cases. 4 However, if a filing changes its primary purpose from investment only after in the initial 13D filing, we will consider such an event as long as it satisfies our definition of collaborative activism. In addition, an investment fund might claim that the purpose of transaction in the initial 13D filing is solely for investment, but the fund might actually partner with other large shareholders to demand desired changes on firm management. For instance, Southeastern Asset Management states their investment in Dell as “investment only”. 69 2.4 Empirical Results 2.4.1 Descriptive Statistics Table 2-1 shows the number of co-activism cases over time and the types of filers. Co-activism is common throughout the years. It occurs in approximately 10 companies per year since 1997 with a slight decline after the financial crisis. To avoid double-counting in the type of filers, we first identify a filer as Hedge Fund if one of the activists is hedge fund. Then we associate the remaining filings with institutional funds. If a filing has no filers from hedge funds or institutional funds, we classify it as individuals. For instance, the collaboration from Carl Icahn and Southeastern Asset Management is classified as Mutual Fund. Panel B of Table 2-1 indicates that most collaborations are through individuals and hedge funds. It is less common that institutional activists work with pension funds and mutual funds, which is consistent with the relatively low number of events initiated by both fund types. The diverse distribution of the events across fund type suggests that our results are not simply driven by one specific category. Panel C of Table 2-1 show activists’ stated objectives from Item 4 of schedule 13D and 13D/A filings and their respective success and partial success rates. Following Brav et al. (p.14 2008), we focus on the following main and sub objectives: 5 1. General undervaluation/maximize shareholder value 2. Capital Structure 3. Business Strategy 4. Sale of Target Company 5 For detailed descriptions of these objectives, please see Brav et al. (2008). 70 5. Governance (Repeal classified board, eliminate poison pill, supermajority, cumulative voting, audit related, board related, executive compensation) Two main objectives of co-activism are shareholder value maximization and board independence and fair representation. With the combined consolidated shares of stocks, joint collaboration has more shareholder power to remove board directors whom they are displeased with. Thus, if activists are dissatisfied with the board’s decision and the management is unwilling to initiate any changes, electing and replacing directors offers a direct mechanism to influence the management team and vote in alignment with shareholders’ interests over charter amendments and reincorporations. 25 co-activism cases are related to selling a company to a third party or taking a company private. Co-activists also propose business strategies to the firm’s management, a total of 21 cases, such as restructuring the business, oppose a merger, pursue alternative strategies, or improve operational efficiency. The success and partial success rates are quite high for co-activism. We define success as achievement of the activist’s main stated goal throughout the life of a 13D filing from news sources, including the amendments. A partial success is defined when activists and the company reach some settlement that partially meets the activist’s original goal. Most objectives exhibit more than 50 percent success and partial success rates. Co-activists are able to convince the management to sell the target company with a 100 percent success rate. They also enjoy around an 80 percent success rate of taking the firm private, restructuring the capital, opposing a merger and acquisi- tion, and gaining representation on the board. Only the objectives involving excess cash and takeover defenses fail. The high success and partial success rates may be reflected in the stock prices of target firms. 71 Activist’ tactics in Panel D of Table 2-1 follows Brav et al. (2010). Although joint collaboration may increase shareholder power, most co-activism adopts non- hostile tactics to monitor and demand changes on target firms. More than 50 per- cent initiate communication with the board and management on a regular basis to deliver their requests. Only about 15 percent of the events are hostile (proxy fight, litigation, and takeover). About 9 percent of the tactics involves a threatened or actual proxy contest. Co-activists resort to litigation and taking the company pri- vate to accomplish the stated objectives in about 1 percent and 5 percent of the events, respectively. 2.4.2 Investment Capital in Co-Activism To measure the economic magnitude of co-activism, we report its capital com- mitment in Panel A of Table 2-2. Initial capital is retrieved from the initial Sched- ule 13D filings. Maximum capital is the maximum stake that a co-activist holds in the target firms, which is retrieved from the subsequent amendments to the initial 13D filings. Note that we do not combine the capital committed by the co-activists who target the same firm with the same stated objectives. Co-activism does not always intend to take control of the company. Fewer than 10 percent of the co-activists seek control in target firms and the 90th percentile of the maximum ownership falls below 50 percent. A median co-activist invests 10.7 percent of stocks, and accumulates to 13.4 percent at the maximum. The median dollar amount of the maximum capital is about 1850 million. The economic size of co-activism is much larger than that of hedge fund activism. According to Brav et al. (2008), at the 50th percentile of the sample, hedge funds hold 9.1 percent and 15.8 million in the targetcompanies. We also observe a more dispersed distribution of the share percentage and dollar amount of committed capital. The interquartile 72 rangeofco-activism’sinitialstakesrangesfrom6percentto21.1percent, compared with 5.4 percent to 8.8 percent in hedge fund activism. Brav et al. (2008) document that hedge fund activism takes about two years before exiting the intervention on average. 6 They attribute the long duration of the attack to the possible long-term success in hedge fund activism. We rely on either news sources or the amendments to the initial 13D filings to determine the exit strategy. We use the exit strategies specified by Brav et al. (2008) and report them in Panel B of Table 2-2. Most of the events fall into the “Still holding/no information” category. Another dominating exit strategy is to sell or merge the target company to another company. 2.4.3 Firm Characteristics Prior to the Filing date Table 2-3 shows the firm characteristics of the target firms prior to co-activism. Bethel et al. (1998) document that blockholders target poorly performing compa- nies. Klein and Zur (2009) report that hedge funds target more profitable firms than other private activists, such as individuals, private equity funds, venture cap- ital firms, and asset management groups. We compare the firm characteristics of the target firms on the year prior to the target to a matched sample of firms. Following Daniel et al. (1997), Brav et al. (2008), and Klein and Zur (2009), we match each target firm by industry, size, and, book-to-market ratio. We first find firms that match target firms based on the Fama-French (1997) 48 industry classification. From these possible industry matches, we choose the one firm with the closest size and book-to-market ratio. 6 On average, the exit time for our sample is about 245 days. We use the date of the last amendment to a 13D filing as the exit date. This approach is more conservative than Brav et al.’s because they also use the actual exit date of a filing if they are able to identify it in the news. 73 We follow Fama and French (1993) to use NYSE breakpoints to assign 5-by-5 size and book-to-market ratio sorted portfolios. Columns 1 and 2 in Table 2-3 show the cross-sectional mean and median of the targets’ firm characteristics. To report the cross-sectional mean (Avg Diff) of the differences on firm characteristics in column 4 in Table 2-3, for each target firm, we first calculate the average firm characteristics of the matched firms, and then take the difference between the target and the matched mean. To evaluate the statistical significance of the difference, we adopt two methods. Column 5 presents the t-statistics associated with the difference in means, and column 6 provides the Wilcoxon signed-rank statistics without assuming the data to have normal distribution, for the difference in medians. The Wilcoxon test serves as a robustness check because it reduces noise from extreme observations with fat tails and skewness. Table 2-3 show the results on the comparisons between matched firms and tar- get firms. STKRET measures the 12 month buy-and-hold return from 1 year prior to the 13D filing date. ABRET is buy-and-hold return from 20 days prior to the 13D filing date to 20 trading days after the date. Both return based measures are significantly different between targets and matched firms, indicating underperfor- mance in target firms prior to the intervention and positive market reaction after the intervention. Operational performance metrics such as EBIDTA to assets (ROA), CF to assets (CF/ASSETS), sales growth (GROWTH), revenue (REV), and Altman’s (1968) Z-score (ZSCORE) are all significantly lower than matched firms. Co- activism also targets “value” firms as evidenced by low Market equity (ME) and Toqin’s q (Q) although the book-to-market ratio is insignificant. 74 The variables related to targets’ capital structure suggest that co-activists pre- fer firms with restricted dividend policy. Target firms show significantly lower dividend per share (DIV/SHR), dividend yield (DIVYLD), and dividend payout ratio (PAYOUT) than matched firms. In addition, Herfindahl-Hirschman Index (HHI), a measure of market concentration, is low in target firms. This indicates that they are in industries with less concentrated market shares. Past studies document that activism supports Jensen’s (1986) free cash flow theory, which states an inverse relationship between the amount of firm debt and the level of cash flow because future interest payments reduce cash flow. However, we do not find free cash flow issues and lower leverage with the target firms as cash and cash equivalents (CASH) and leverage (LEV) are not significantly different. The literature documents high institutional ownership, analyst coverage, and trading liquidity in the targets by hedge fund activism. However, we do not find similar evidence. Given that co-activists need to accumulate a higher stake than hedge fund activists in a short time, it will be interesting to explore how and where they acquire shares in future studies. We next extend our univariate analysis to probit multivariate regressions to assess the marginal effect of each variable. Table 2-4 shows the results from probit analysis. TheresultsaremostlyconsistentwiththoseinTable2-3. Aone-standard deviation decrease in Q and ROA increases the probability of being targeted by 50 percent and 7 percent, respectively. The coefficients on DIVYLD and ME are significantly negative, suggesting that target firms object to pay dividend and experience a sharp decline in equities. The differences with the univariate anal- ysis are research and development expenses (RND), Z-score (ZSCORE), capital expenditures, and investor sophistication. The loading on RND is negative and significant, suggesting that co-activism is more likely to accuse a company with 75 low innovations. Z-score is a bankruptcy indicator and its coefficient is 0.067, implying that co-activists do not simply bet on a bankrupt firm to turn it around although their targets all well underperform prior to the filing date. In view of capital structure, firms with lower leverage (LEV) and higher total debt to total assets (DEBT/ASSETS) than their matched firms are more likely to be targeted by co-activists. Two leverage ratios have opposite sign because we calculate equity differently between these two measures. DEBT/ASSETS reflect degrading market value of the equity, which is possibly much lower than matched firms, and thus falsely inflating this ratio. In addition, multivariate analysis points out that ana- lyst coverage and institutional ownership likely to be higher in target firms. This confirms the sophistication of clientele for target firms. Overall, standard measures of accounting and financial health suggest that target firms poorly underperform prior to the Schedule 13D filing date. This findings are consistent with Bethel et al. (1998), but in contrast with studies on hedge fund activism (Brav et al. (2008) and Klein and Zur (2009)). 2.4.4 Market Response to 13D and 13D/A filings Given the long time-series of our sample, we investigate both short-term and long-term abnormal performance of co-activism. We use the following event win- dows: [-20 Day, +5 Day], [-20 Day, +20 Day], [-20 Day, +1 Year], [-20, +5 Year]. We define the filing date as day 0. For example, [-20 Day, +5 Day] denotes 20 days before and 5 days after the filing date. We start with 20 days prior to the filing date to capture possible information leakage and the 10 day window to file 13D and 13D/A filings. We extend the event window from 5 days to 20 days to reflect any subsequent news coverage. We further extend the event window to one year and five years to investigate whether co-activism is short-lived. 76 We adopt three measures of abnormal stock returns. Size-adjusted return is the difference between its buy-and-hold return over a selected time period and the buy-and-hold return for the same time period on the Fama-French size-matched portfolio of firms. The market-adjusted return is the difference between the tar- get’sbuy-and-holdreturnandthevalue-weightedNYSE/Amex/Nasdaqindexfrom CRSP. The industry-adjusted return is the difference between the target’s buy- and-hold return and the return for all firms in the target’s Fama-French (1997) 48-industry code. Table 2-5 provides the cross-sectional distribution of abnormal stock return on target firms. Target firms earn statistically significant average market-, industry-, and size-adjusted returns of 6%, 6.8%, and 6.9%, respectively over the [-20 Day, +5 Day] window. The medians are 2.3%, 3.2%, and 2.3%, respectively, and also statistically different from zero. Over the [-20 Day, +20 Day] window, the mean (median) of market-, industry-, and size-adjusted returns are 5% (2.8%), 6.6% (3.9%), and 7.3% (3.5%), and statistically different from zero. These abnormal returns are consistent with those of Bethel et al. (1998), Brav et al. (2008), and Klein and Zur (2009), but in contrast to studies in pension fund and mutual fund activism (e.g. Karpoff (2001)). Most strikingly, our findings on long-term abnormal stock performance are opposite to previous studies. Figure 2-1 shows that 20 days after the filing date, the abnormal return shows a pattern of trending down. Over the [-20 Day, 1 Year] window, the mean (median) of market-, industry-, and size-adjusted returns are insignificantly different from zero. This pattern exacerbates over the long [-20 Day, 5 Year] window. Both mean (median) industry- and size-adjusted abnormal return drop below zero (-30.7% (-37%), -35% (-45.3%), respectively) and are significantly 77 different from zero. The mean market-adjusted return of -8.8% is not significant, but its median of -35.4% is comparable to industry- and size-adjusted return. To summarize, our results on abnormal stock return echo policy makers’ con- cern that activism can be short-lived. Our evidence that market responds pos- itively over the short event window but long-term market reaction turns quite negative implies that co-activism aims for short-term gains at the expense of long- term shareholder value. This might explain why we do not observe a wave of co-activism over time and the duration of co-activism is relatively short. 2.4.5 Abnormal Return by Objectives and the Degree of Hostility We further break down abnormal performance by activists’ main objectives (Panel C, Table 2-1) and the degree of hostility. We use main objectives because a couple of sub objectives have only one firm and its t-statistics cannot be derived. Hostility is classified according to tactics in Panel D of Table 2-1. Tactics including a threatened or actual proxy fight, takeover and lawsuit are considered hostile. Panel A and B of Table 2-4 show abnormal performance of target firms by objectives and the degree of hostility over the [-20 Day, +20 Day] window. We focus our interpretation on size-adjusted return as three measures of abnormal per- formance yield similar results. Business strategy (operational inefficiency, restruc- turing, objection to M&A, and alternative strategies) and sale of target company are two predominate drivers for the abnormal performance of co-activism. 7 Mean differences in buy-and-hold abnormal return between target firms and matched 7 If we winsorize our data at the 1% and 99% levels, median abnormal performance for max shareholder value will also be significantly different from zero. 78 firms are 29.2% and 8.8% for business strategy and sale of target company, respec- tively. Themediansare20.7%and2.5%. Bothmeansandmediansaresignificantly different from zero. All other objectives exhibit positive adjusted returns despite being insignificant. The breakdown of hostile and non-hostile co-activism shows that the market responds to hostile events more favorably than non-hostile events. Hostile events show strong statistical significance on mean (median) adjusted returns, ranging from 7.7% (4.2%) to 10.14% (5.3%). Non-hostile events exhibit insignificant abnor- mal returns and they are mostly negative. Allinall, marketrespondsdifferentlytothetypeofobjectivesandtactics. High abnormal returns by a specific objective and whether a tactic is hostile might imply their costs of carrying out an attack by co-activism are higher. To compensate such high costs, investors demand high premium as evidenced by positive abnormal return over the [-20 Day, 20 Day] announcement window. 2.4.6 Changes in Firm Characteristics prior and after the Filing Date Tofacilitateourstudyontheshort-termandlong-termeffectontargetfirms, in this section, we examine the changes in accounting performance over the following time periods: [0 Year to 1 year], [-1 Year to 1 Year], and [0 Year to 5 Year]. The year of the filing date is 0 Year. The accounting variables are all available as of the December in year T. Panel A and B in Table 2-7 present the operating performance one year after the attack from the filing year and one year prior to the filing date, respectively. Revenue and market equity are significantly lower one year after the intervention, suggesting that target firms are not turned around by co-activism. A big drop in 79 total assets may come from a significant decline in cash. It seems that target firms do not increase interest obligations or loose dividend policies to reduce cash as the changes in leverage, dividend yield, and dividend per share are not statistically different from zero. Also, target firms cut back investment, such as research and developmentexpenses. Assuch, thereductionincashmightimplythatco-activists extractrentsfromthetargetfirms,suchasincreasingpayrolltotheCEOandboard members. Interestingly, the long-term [0 Year to 5 Year] window shows a similar pattern. Target firms’ revenue and cash continues to drop, causing market equities and total assets to plunge. This result is consistent with our finding that abnormal performance of co-activism in the long run is significantly negative. 2.5 Alternative Explanations One possible explanation for short-term abnormal returns, but not long-term, is due to demand pressure from co-activists. 8 A collective stake on target firms might be so substantial that stock prices are pushed up by the initial capital during the [-20 Day, +20 Day] announcement window. In Figure 2-2, we plot the average abnormal share turnover over the event window. Following Brav et al., we use the turnover over the [-100 Day, -40 Day] window preceding the filing date as a benchmark to measure abnormal turnover. We can see an increasing trend in abnormal trading volume during the 10-day period before the filing, reflecting the required 10-day window to file a Schedule 13D filing if purchasing more than 5% of stocks. However, we find little abnormal share turnover relative to [-100 Day, 8 Previous studies also show no direct relationship between ownership concentration and firm value (e.g. Demsetz and Lehn (1985), Short (1994), Himmelberg et al. (1999), Song and Walking (1993), and Slovin and Sushka (1993)). 80 -40 Day] after the filing date, suggesting abnormal returns are not driven by price pressure from co-activists. Another reason that institutional activists work side by side may be attributed to common trading strategies. For example, we might simply capture activists targeting value firms at the same time as target firms appear to have a low q. We may reject this claim on several grounds. One, Panel A of Table 2-1 shows no wave inco-activismandthenumberofeventsstaysaboutthesameovertime. Ifactivists team up because of a shared common strategy, we should observe way more events over the years as many institutions or individual investors follow similar trading strategies, such as momentum or value investing. A low number of co-activism events might imply more competitions among activists than cooperation. Two, Panel B of Table 2-6 indicates that target firms attacked by hostile tactics earn significant positive abnormal return, but those by non-hostile intervention has insignificant and negative abnormal return. The distinct market response between hostile and non-hostile events may imply that there is differential information between these two types of events. Carrying out a hostile attack by activists is more costly. As a result, when co-activism takes place, investors digest information fromtheeventanddemandhighexpectedreturnforbearinghighcostsaccordingly. Third, momentum effect lasts for about one year. Our pattern of abnormal stock performance shows reversal in roughly 9 months after the filing. This rejects that short-term positive market reaction is simply driven by momentum. Our results are less likely to be driven by market overreaction. Market over- reaction states that the market overreacts towards new information and causes temporary price impact rather than reflecting firm value changes. Consequently, return should revert to negative territory shortly after the event to reflect firm’s fundamental value. However, target firms’ return in co-activism takes about 9 81 months to revert to the original level and more than three years to stay below zero, which suggests price reflection fundamental changes in firm value over time. Moreover, activists may possess private information on target firms or own better stock picking skills so that price appreciates after the filing date. However, it is more likely that an institutional investor will exploit private information to maximize her own profits rather than sharing with other fund managers. Also, the long-term reversal on stock performance refutes co-activists’ superior stock picking skills. If co-activists are superior stock pickers, target firms’ performance should persist at least one year. 2.6 Conclusion We construct a unique dataset on co-activism from 1994 to 2013 from SEC 13D and 13D/A filings. The occurrences of co-activism over time suggest that large shareholders have incentives to work together to improve the governance of the firm from within, by taking steps to protect their own investments in the face of potential managerial agency conflicts. We document that co-activism enjoys a high success rate of accomplishing its stated goals without taking control blocks of stocks. The high success rate is expected by the investors, as we find that market responds positively surrounding the filing date. However, abnormal performance turns significantly negative in the long run, implying that co-activism aims for short-term gains at the expense of long-term shareholder value and pressures the companies to take myopic actions that are costly in the long term. Consistent with this argument, our results on the changes of operating performance show that financial and operational measures such as revenue, cash, and market equity start to deteriorate about one month 82 after the filing date. Overall, in the long run, the costs of co-activism outweighs its benefits, resulting in negative abnormal stock return and declining operating performance in target firms five years after the filing date. Our results imply that activists or activist funds with co-activists are more likely to act in the interest of their investors at first, but the costs associated with collective monitoring of their target firms gradually catch up and adversely affect their return performance. This has two important implications. If co-activism hurtsshareholdervalueoverthelongrunanditisnotidentifiedunderthedefinition of “group” in Schedule 13(d), policymakers should impose activists to disclose such collaboration. Moreover, the failure to enhance long-term firm value in co-activism may imply that shareholders should be disallowed to directly initiate and vote to adopt more proposals on corporate governance and structure, and shareholder power should be limited. In short, our evidence implies that co-activism is short-term in focus and dam- age shareholder value in the long run. While the efficacy of institutional activism continues to be the subject of debate, our evidence that co-activism is effective in pressuring the management to change and maximize shareholder value might start a new form of shareholder activism in the coming years. 2.7 Appendix 2.7.1 Variable Construction Below gives the list of variable construction. All fundamentals are in million dollars. • AbnormalReturn(ABRET):Buy-and-hold[-20Day,+20Day]returnaround filing date - Buy-and-hold market return during the same event period. 83 • Amihud Illiquidity (AMIHUD): Yearly average (using daily data) of 1000 * Sqrt(|Return| / (Dollar Trading Volume)). • Number of Analysts (ANALYST): Number of analysts covering the company from I/B/E/S (in million). • Total Asset (AT): Total asset from Compustat. • Book-to-MarketRatio(BM):BookValueofEquity/MarketValueofEquity. • Cash and Cash Equivalents (CASH): Cash and cash equivalents from Com- pustat. • Dividend Share (DIV/SHR): Dividend per share (Ex-Date). • Dividend Yield (DIVYLD): (Common Dividend + Preferred Dividends) / (Market Value of Common Stocks + Book Value of Preferred). • Growth of Sales (GROWTH): Sales / Lagged Sales. • Herfindahl index (HHI): Herfindahl index of sales in different business seg- ments using two-digit of SIC codes. • Institutional Holdings (INST): The proportion of shares held by institutions. • Leverage (LEV): The book leverage ratio defined as Debt / (Debt + Book Value of Equity). • Market Value of Equity (ME): Stock Price * Total Shares Outstanding. • Payout Ratio (PAYOUT): Total Dividend Payments / (Net Income before Extraordinary Items). 84 • Tobin’s q (Q): (Book Value of Debt + Market Value of Equity) / (Book Value of Debt + Book Value of Equity). • Revenue (REV): Sales from Compustat. • R&D Expenses (RND): R&D expenses from Compustat. • Return on Assets (ROA): EBITDA (from Compustat) / Total Assets. • Buy-and-hold Past 12-month Return (STKRET): Buy-and-hold Past 12- month return before the filing date. • Z-Score (ZSCORE): 1.2X 1 + 1.4X 2 + 3.3X 3 + 0.6X 4 +X 5 withX 1 =(Current Asset - Long-term Asset) / Total Asset, X 2 =Retained Earnings / Total Asset, X 3 =EBIT / Total Asset, X 4 =Book Value of Equity / Total Asset, X 5 =Sales / Total Asset. • (Normalized) Cash Flow (CF/ASSETS): (Net Income + Depreciation and Amortization) / Total Asset. • (Normalized) Cash Flow from Operations (CFO/ASSETS): Net Cash Flow From Operating Activities/Total Asset. • (Normalized)CapitalExpenditures(CAPX/ASSETS):(CapitalExpenditure - Sales of Property)/Total Asset. • (Normalized) Cash and Cash Equivalents (CASH/ASSETS): Cash and Cash Equivalents/Total Asset. • (Normalized) Current Asset (CA/ASSETS): Current Asset/Total Asset. • (Normalized) Total Debt (DEBT/ASSETS): Total Debt/Total Asset. 85 • (Normalized) Current Debt (CDEBT/ASSETS): Current Debt/Total Asset. • (Normalized) Long-Term Debt (LTDEBT/ASSETS): Long-term Debt/Total Asset. 86 Appendix A Tables 87 Table 1-1. Sample Statistics This table presents summary statistics for the sample. Panel A gives information on the overall sample. Panel B gives information on the final status of startups thatareVC-fundedandVC-unfunded. PanelCgivesinformationonthesizeofVC syndicates. The size refers to the number of VCs in a syndicate. Panel D describes the distribution of the number of funding rounds experienced by the startups in the sample. Panel E describes the distribution of funding by funding rounds. The time period is from January 1998 to December 2014. A. Sample # Startups 9,303 # VC Syndicates 2,844 # VCs 755 # Time Periods 204 B. Startup Final Status Status Death Survival IPO&MA Total Unfunded 1,879 4,965 109 6,953 Funded 414 1,408 528 2,350 Total 2,293 6,373 637 9,303 C. VC Syndicate # of VCs 1 2 3 4 5 6 7 8 9 10 11 Freq. 505 839 661 406 234 99 50 25 18 3 4 Percent. 17.76 29.5 23.24 14.28 8.23 3.48 1.76 <1 <1 <1 <1 D. Startup Funding # of funding 0 1 2 3 4 5 6 7 8 9 10 11 Freq. 6,953 452 1,015 479 236 96 44 13 11 1 1 2 Percent 74.74 4.86 10.91 5.15 2.54 1.03 <1 <1 <1 <1 <1 <1 E. Funding by Rounds Rounds of funding 1 2 3 4 5 6 7 8 9 10 11 Freq. 2350 1898 883 404 168 72 28 15 4 3 2 Percent 40.33 32.57 15.15 6.93 2.88 1.24 <1 <1 <1 <1 <1 88 Table 1-2. Measure Statistics This table presents summary statistics for the variables in Eq(1), Eq(2) and Eq(4). Panel A gives information on the cumulative return r in Eq(1). Panel B to Panel E give the information on the right-hand-side variables X in the three equations. Panel B describes the macroeconomic variables X t . Panel C describes startup- specific features X j t separately for constant ones and time-varying ones. Panel D describes VC syndicate-specific features X i t . Panel E describes startup-VC syndicate-pairwise-specific features X ij t . There are 204 months, 9,303 startups, and 2,844 VC syndicates. In total, there are 520,715 startup-month observations, 580,176 VC syndicate-month observations, 26,457,732 startup-VC syndicate pairs, and 1,650,020,544 startup-VC syndicate-month observations. The detailed con- struction of these variables is given in Section 1.7. A. Cumulative Returns Mean Std Skew Kurt min p5 p10 p25 p50 p75 p90 p95 max 6.6 9.5 0.3 1.2 -8.2 -2.3 -2.3 -2.3 0.5 17.1 18.8 19.6 26.2 B. Macroeconomic Variables N Mean Std min p25 p50 p75 max (Y baa −Y us10 ) (%) 204 2.62 0.8 1.56 2.11 2.57 2.98 6.01 (r m −r f ) (%) 204 0.48 4.68 -17.23 -2.09 1.19 3.49 11.35 smb (%) 204 0.28 3.56 -16.41 -1.63 0.19 2.28 22.02 hml (%) 204 0.21 3.4 -12.61 -1.49 0.08 1.71 13.89 r f (%) 204 0.18 0.17 0 0.01 0.13 0.37 0.56 C. Startup Features N Mean Std min p25 p50 p75 max Constant # locations 9,303 1.21 1.01 1 1 1 1 66 # categories 9,303 1.85 1.3 1 1 1 2 14 # products 9,303 0.43 1.97 0 0 0 0 134 Time-varying t from last round 520,715 3.02 3.01 0 0.83 2 4.25 16.92 t2 from last round 520,715 18.14 34.91 0 0.69 4 18.06 286.17 # rounds 520,715 0.32 0.81 0 0 0 0 10 # startups founded 520,715 0.17 0.47 0 0 0 0 5 1(top20 school) 520,715 0.3 0.46 0 0 0 1 1 89 D. VC Syndicate Features N Mean Std min p25 p50 p75 max Constant # of VCs 2,844 2.93 1.63 1 2 3 4 11 # locations 2,844 6.37 5.40 1 3 5 9 75 Time-varying 1(cooperated) 580,176 0.33 0.47 0 0 0 1 1 # rounds 580,176 5.16 9.65 0 0 1 6 116 # categories 580,176 8.11 12.87 0 0 1 11 87 1(top20 school) 580,176 0.86 0.35 0 1 1 1 1 E. Startup-VC Syndicate Features N Mean Std min p25 p50 p75 max Constant distance 2.65× 10 7 2894 3522 0 79 1200 4313 19933 [0, 100] 2.65× 10 7 0.25 0.43 0 0 0 1 1 (100, 1000] 2.65× 10 7 0.21 0.4 0 0 0 0 1 (1000, max] 2.65× 10 7 0.54 0.5 0 0 1 1 1 days of travel 2.65× 10 7 2.29 0.84 0 1 2 2 3 Time-varying 1(funding tie) 1.65× 10 9 0.61% 1(alumni tie) 1.65× 10 9 20.76% 90 Table 1-3. Estimation Result This table presents the estimation result for the baseline model described by the system of equations Eq(1), Eq(2) and Eq(4). The first two columns give the estimates forφ r,y andφ r,n in Eq(1) for the law of motion of the cumulative return r. The third and fourth columns give the estimates for φ s,y and φ s,n in Eq(2) for the determinants of the implicit exit value s. The last column gives the estimates forφ v in Eq(4) for the determinants of the subjective value addedv. Details of the estimation are given in Section 1.7. Numbers in the brackets are the t-statistics. Significance at 10%, 5%, and 1% levels are denoted by *, **, and ***. Influence Selection Growth (r) Exit Value (s) Value Added (v) Funded Unfunded Funded Unfunded [1] [2] [3] [4] [5] r 0.028 *** 0.063 *** 0.034 *** (30.78) (40.17) (178.94) sigma2 0.01 *** 0.092 *** 1.124 *** 1.373 *** (103.37) (1287.12) (11.79) (28.35) intercept 0.004 *** 0.004 *** -0.251 -0.447 *** (2.91) (44.61) (1.35) (3.25) Macro Variables (Y baa −Y us10 ) -0.001 -0.013 *** 0.005 0.024 (0.90) (197.74) (0.16) (1.02) (rm−r f ) 0.007 0.032 *** -0.045 0.573 (0.72) (41.62) (0.10) (1.62) smb -0.023 0.006 *** -1.332 -0.929 * (0.95) (5.03) (1.40) (1.92) hml 0.036 * 0.099 *** -0.045 -0.250 (1.67) (48.54) (0.04) (0.54) r f 0.017 0.196 *** 0.072 0.370 ** (0.21) (57.13) (0.30) (2.48) Startup Features # locations 0.003 *** 0.017 *** 0.009 -0.004 -0.853 *** (5.26) (406.38) (0.43) (0.29) (6.23) # categories -0.009 *** -0.004 *** -0.007 -0.006 -0.273 *** (4.90) (171.65) (0.29) (0.75) (9.10) # products 0.004 *** 0.004 *** 0.007 0.015 0.024 *** (2.67) (85.13) (0.34) (1.06) (7.71) t from last round 0.000 0.041 (0.35) (1.66) t2 from last round -0.040 -0.046 *** (0.11) (3.00) # rounds 0.008 0.001 *** 0.226 -0.001 0.022 *** (0.00) (187.20) (0.05) (1.35) (35.88) # startups founded -0.005 0.005 *** -0.012 0.013 -0.158 *** (1.49) (11.48) (0.36) (0.45) (12.09) 1(top20 school) -0.019 *** 0.154 *** 0.005 -0.037 -0.434 *** (10.38) (372.63) (0.09) (0.91) (35.30) VC Syndicate Features # of VCs -0.003 *** 0.006 -0.342 *** (18.15) (0.22) (10.27) # locations 0.000 *** 0.002 0.001 *** (8.26) (0.49) (6.60) 1(cooperated) 0.001 *** 0.002 0.030 *** (12.59) (0.46) (12.99) # rounds -0.001 -0.068 0.101 *** (0.64) (0.53) (15.29) # categories 0.000 *** -0.001 -0.019 *** (8.08) (0.59) (11.38) 1(top20 school) 0.039 *** -0.001 -0.647 *** (18.08) (0.02) (24.72) Pair Features days of travel -0.146 0.295 -0.210 *** (1.66) (0.86) (28.02) 1(funding tie) 0.004 *** -0.251 4.486 *** (4.02) (1.35) (10.36) 1(alumni tie) 0.003 *** -0.003 0.117 *** (18.62) (0.24) (35.28) 91 Table 1-4. Previously Funded vs. Previously Unfunded Startups This table compares the means of the period return (r t+1 −r t ), the implicit exit value s t , IPO&MA rate and death rate for the previously funded and unfunded startups. Period return* is winsorized between 1 and 99 percentiles separately for the subsamples of previously funded and unfunded startups. Panel A uses all previously unfunded startups as control. Panel B uses only comparable star- tups identified by the Nearest Neighbor (NN) method (with 1 or 5 neighbors) or the Propensity Score (PS) method (with probit or logit treatment model). The covariates for the identification of the comparable startups include startup-specific variables (e.g. # locations, # categories, # products, # startups founded), and macroeconomic variables (e.g. (Y baa −Y us10 ), (r m −r f ), smb, hml, and r f ). Num- bers in the brackets are the t-statistics. Significance at 10%, 5%, and 1% levels are denoted by *, **, and ***. A. Previously Funded vs. Previously Unfunded Period return* Period return Exit value IPO&MA(%) Death(%) Funded 0.034 0.034 0.044 0.5 0.4 Unfunded -0.017 0.051 -0.197 0.02 0.49 Difference 0.05 *** -0.017 0.241 *** 0.48 *** -0.09 t or chi2 10.501 0.901 16.232 164.91 B. Previously Funded vs. Comparable Previously Unfunded Period return* Period return Exit value Nearest Neighbor (nn1) 0.052 *** -0.039 0.206 *** (9.62) (0.57) (9.84) Nearest Neighbor (nn5) 0.052 *** -0.042 0.223 *** (10.59) (0.81) (12.64) Propensity Score (probit) 0.051 *** -0.024 0.201 *** (6.88) (0.58) (9.86) Propensity Score (logit) 0.051 *** -0.025 0.199 *** (8.97) (0.79) (9.74) 92 Table 1-5. Constrained Estimation This table presents the estimation results for the constrained model in which the coefficients associated with the intercept term, the macroeconomic variables, and the startup-specific features are constrained to be the same. This is equivalent to the assumption that funded and unfunded startups have the same law of motions for their cumulative return r and implicit exit value s, and at the same time, the VC-related features for the unfunded startups are set to be zeros. Therefore, Eq(1) and Eq(2) become Eq(A28) and Eq(A29). Details of the estimation are given in Section 1.7. Numbers in the brackets are the t-statistics. Significance at 10%, 5%, and 1% levels are denoted by *, **, and ***. Influence Selection Growth (r) Exit Value (s) Value Added (v) [1] [2] [3] r 0.081 *** 0.033 *** (43.10) (163.69) sigma2 0.085 *** 1.584 *** (595.14) (209.15) intercept 0.052 *** -0.934 *** (70.07) (7.63) Macro Variables (Y baa −Y us10 ) 0.052 *** -0.044 ** (70.07) (2.41) (rm−r f ) -0.011 *** 0.785 *** (24.03) (2.94) smb 0.008 *** -2.095 *** (16.17) (4.06) hml 0.099 *** -0.085 (6.35) (0.21) r f 0.226 *** 0.569 *** (8.44) (5.79) Startup Features # locations 0.02 *** 0.01 -0.849 *** (71.08) (0.63) (6.26) # categories -0.009 *** 0.000 -0.273 *** (62.09) (0.02) (8.93) # products 0.001 *** 0.014 0.023 *** (26.69) (1.18) (7.16) t from last round 0.052 *** (10.81) t2 from last round -0.026 *** (8.19) # rounds 0.002 *** -0.003 0.022 *** (454.42) (1.28) (44.80) # found startups -0.003 -0.066 -0.164 *** (0.30) (1.55) (10.54) 1(top20 school) -0.017 *** 0.042 -0.435 *** (32.69) (0.62) (34.76) VC Syndicate Features # of VCs -0.004 *** 0.048 * -0.344 *** (7.99) (1.67) (10.28) # locations 0.001 *** 0.006 0.001 *** (5.37) (1.35) (5.21) 1(cooperated) 0.001 *** 0.005 * 0.03 *** (9.46) (1.94) (12.58) # rounds 0.01 *** -0.258 ** 0.105 *** (2.41) (1.98) (4.76) # categories 0.000 *** -0.004 -0.02 *** (3.76) (1.35) (11.43) 1(top20 school) 0.042 *** -0.05 -0.648 *** (7.11) (0.38) (25.79) Pair Features days of travel -0.021 *** 0.591 -0.696 *** (3.74) (1.51) (12.16) 1(funding tie) -0.119 -0.652 4.52 *** (0.36) (1.38) (10.13) 1(alumni tie) 0.005 *** -0.024 0.117 *** (5.27) (1.16) (41.81) 93 Table 1-6. Expected VC Impact and Startup Future Growth This table presents the correlations between the subjective value addedv t and the future period return (r t+τ −r t+τ−1 ) as well as the future implicit exit value s t+τ for startups that are selected to be funded at t. The correlations are calculated as corr(v ij t ,r j t+τ −r j t+τ−1 ) and corr(v ij t ,s j t+τ ), with τ varies from 1 to 12 (months). Numbers in the brackets are the t-statistics. Significance at 10%, 5%, and 1% levels are denoted by *, **, and ***. Period return Exit value 1 0.114 *** 0.056 *** (30.00) (12.06) 2 0.089 * 0.125 ** (1.78) (2.26) 3 0.059 -0.025 (0.24) (0.04) 4 0.03 0.043 (0.06) (0.12) 5 0.019 -0.029 (0.03) (0.06) 6 0.059 -0.03 (0.23) (0.05) 7 0.065 0.103 ** (0.29) (2.50) 8 0.102 * -0.048 (1.59) (0.14) 9 0.126 ** 0.038 (5.33) (0.08) 10 0.106 * -0.02 (1.63) (0.03) 11 0.105 * 0.061 (1.45) (0.24) 12 0.07 0.07 (0.27) (0.37) 94 Table 1-7. Selection on Private Information This table presents the estimation results for the extended correlation model in which Eq(1) contains an additional term ρ r,y ξ ij t−1 for the funded case and Eq(2) contains an additional termρ s,y ξ ij t−1 for the funded case. The estimates forρ r,y and ρ s,y give the dependence of the subjective value added on the private information that will drive the startup growth and the implicit exit value one period ahead. They are denoted by “rho” in the table. Details of the extended model are given in Section 1.7. As before, the first two columns give the estimates forφ r,y andφ r,n , the third and fourth columns give the estimates for φ s,y and φ s,n , the last column gives the estimate forφ v . Numbers in the brackets are the t-statistics. Significance at 10%, 5%, and 1% levels are denoted by *, **, and ***. Influence Selection Growth (r) Exit Value (s) Value Added (v) Funded Unfunded Funded Unfunded [1] [2] [3] [4] [5] rho 0.009 *** 0.073 *** (42.32) (5.43) r 0.017 *** 0.044 *** 0.033 *** (37.90) (77.06) (327.79) sigma2 0.011 *** 0.092 *** 1.256 *** 1.284 *** (46.21) (1316.04) (49.91) (127.15) intercept 0.006 *** 0.005 *** -0.29 -0.464 *** (7.08) (50.29) (1.28) (4.47) Macro Variables (Y baa −Y us10 ) -0.001 -0.013 *** 0.008 0.027 (0.99) (183.77) (0.27) (1.19) (rm−r f ) 0.009 0.026 *** 0.103 0.61 (0.61) (37.22) (0.18) (1.26) smb -0.026 0.008 *** -1.644 -0.884 * (1.12) (7.27) (1.54) (1.77) hml 0.034 0.105 *** -0.095 -0.205 (1.60) (47.28) (0.10) (0.33) r f 0.013 0.195 *** 0.133 0.391 ** (0.96) (54.45) (0.50) (1.99) Startup Features # locations 0.002 *** 0.017 *** 0.007 -0.014 -0.849 *** (12.89) (440.09) (0.36) (1.43) (6.28) # categories -0.009 *** -0.004 *** -0.01 0.001 -0.271 *** (11.70) (175.47) (0.47) (0.10) (8.88) # products 0.004 *** 0.004 *** 0.013 0.008 0.024 *** (8.33) (92.83) (0.55) (0.85) (6.15) t from last round 0.000 -0.238 (0.12) (1.59) t2 from last round -0.078 -0.046 *** (0.20) (3.16) # rounds 0.05 0.001 *** 0.617 -0.001 0.022 *** (0.20) (184.14) (0.13) (1.19) (40.95) # found startups -0.004 0.005 *** -0.012 0.004 -0.164 *** (0.54) (11.75) (0.35) (0.14) (12.10) 1(top20 school) -0.021 *** 0.154 *** -0.003 -0.036 -0.434 *** (6.20) (403.86) (0.05) (1.41) (30.54) VC Syndicate Features # of VCs -0.005 *** 0.009 -0.347 *** (12.52) (0.36) (9.98) # locations 0.000 *** 0.002 0.001 *** (3.32) (0.57) (6.84) 1(cooperated) 0.001 *** 0.002 0.03 *** (60.80) (0.66) (12.53) # rounds 0.000 -0.012 0.105 *** (1.27) (0.10) (5.51) # categories 0.000 *** -0.002 -0.02 *** (16.51) (0.68) (11.34) 1(top20 school) 0.038 *** -0.011 -0.643 *** (14.66) (0.17) (23.59) Pair Features days of travel -0.029 *** 0.291 -0.703 *** (2.04) (0.68) (12.70) 1(funding tie) 0.006 *** -0.29 4.484 *** (7.08) (1.28) (9.75) 1(alumni tie) 0.003 *** -0.005 0.118 *** (63.38) (0.30) (27.67) 95 Table 1-8. Simulation Results: Model 1 vs. Model 2 Thistablecomparesthedifferenceinmeansoftheperiodreturn (r t+1 −r t ), implicit exit value s, IPO&MA rate and death rate for the two models in the simulation. Model 1 assumes random selection of funding for all periods to exclude the chain effect. Model 2 assumes random selection of funding only at the first period. I simulate 100 startups that have ex-ante homogeneous features which are set to the means of these features in sample. The startups also have synchronous births and same cumulative return in the beginning. I assume there is one VC that has a funding quota of 50 so each startup gets funded with a probability of 0.5 at any time. Panel A compares the values for previously funded and unfunded startups for the two models. Panel B compares the transition matrix of current funding given previous funding for the two models. Panel C compares the values τ period ahead for currently funded and unfunded startups, with τ ranging from 1 to 10. Numbers in the brackets are the t-statistics. Significance at 10%, 5%, and 1% levels are denoted by *, **, and ***. A. Previously Funded vs. Previously Unfunded 1. Random Selection for All Periods Period return Exit value IPO&MA(%) Death(%) Funded 0.515 0.577 1.327 0.207 Unfunded 0.319 0.326 1.067 0.249 Difference 0.196 0.251 0.26 -0.041 t-stat 0.74 0.61 1.002 0.363 2. Random Selection Only for the First Period Period return Exit value IPO&MA(%) Death(%) Funded 0.956 0.828 1.886 0.179 Unfunded 0.106 -0.126 0.811 0.516 Difference 0.851 *** 0.954 *** 1.075 *** -0.337 * t-stat 2.92 2.45 2.456 1.645 B. Transition Matrix 1. Random Selection for All Periods 2. Random Selection Only for the First Period Currently Funded Currently Funded Previously Funded N Y Previously Funded N Y N 0.502 0.498 N 0.931 0.069 Y 0.533 0.467 Y 0.041 0.959 96 C. Future Difference for Initially Funded vs. Unfunded 1. Random Selection for All Periods Period return Exit value IPO&MA(%) Death(%) 1 0.196 0.251 0.26 -0.041 2 0.278 0.082 0.094 -0.015 3 0.175 0.037 0.081 -0.015 4 0.082 0.17 -0.007 0.075 5 0.069 0.024 0.271 0.045 6 0.113 0.247 0.639 *** -0.017 7 -0.026 0.215 0.291 -0.016 8 0.02 -0.06 0.822 *** -0.241 *** 9 0.01 0.336 0.43 -0.209 10 0.034 0.087 0.222 -0.011 2. Random Selection Only for the First Period Period return Exit value IPO&MA(%) Death(%) 1 0.851 *** 0.954 *** 1.075 *** -0.337 * 2 0.906 *** 1.077 *** 1.124 *** -0.349 * 3 0.909 *** 1.214 *** 1.278 *** -0.284 4 0.881 *** 1.052 ** 1.256 *** -0.215 5 0.856 *** 1.023 ** 1.336 *** -0.225 6 0.879 *** 1.35 *** 1.312 *** -0.344 7 0.857 *** 1.218 ** 1.175 *** -0.358 8 1.024 *** 1.34 ** 1.256 *** -0.281 9 1.114 *** 1.677 *** 1.316 *** -0.243 10 1.359 *** 1.549 *** 1.4 *** -0.233 97 Table 1-9. Estimation for Subsamples This table presents the estimation results using two subsamples for the baseline model described by the system of equations Eq(1), Eq(2), and Eq(4) as in Table 1- 3. Panel A and Panel B give the results for 1998-2006 and 2007-2014 respectively. The subsamples only contain startups that are born during those sub-periods. The first two columns give the estimates forφ r,y andφ r,n . The third and fourth columns give the estimates for φ s,y and φ s,n . The last column gives the estimates for φ v . Significance at 10%, 5%, and 1% levels are denoted by *, **, and ***. A. 1998-2006 Influence Selection Growth (r) Exit Value (s) Value Added (v) Funded Unfunded Funded Unfunded [1] [2] [3] [4] [5] r 0.001 ** 0.016 *** 0.033 *** (2.08) (42.55) (123.33) sigma2 0.011 *** 0.108 *** 1.524 *** 1.473 *** (58.52) (131.291 (24.17) (61.29) intercept 0.052 *** 0.028 *** 0.136 -0.264 ** (16.17) (28.04) (0.53) (2.34) Macro Variables (Y baa −Y us10 ) -0.009 *** -0.051 *** -0.019 0.051 (2.92) (478.79) (0.15) (1.47) (rm−r f ) 0.056 0.04 *** 0.267 0.093 (0.60) (31.51) (0.15) (0.28) smb -0.036 -0.074 *** -0.379 0.349 (0.56) (40.50) (0.25) (0.82) hml -0.055 -0.048 *** 0.114 0.728 (0.39) (15.57) (0.06) (1.56) r f -0.024 -0.101 *** -0.111 0.072 (0.74) (84.64) (0.25) (0.56) Startup Features # locations -0.004 *** 0.013 *** 0.008 0.007 -0.154 *** (10.00) (473.09) (0.27) (0.86) (9.02) # categories 0.002 -0.002 *** 0.014 0.003 -0.44 *** (0.18) (69.35) (0.29) (0.20) (6.06) # products 0.002 *** 0.006 *** -0.002 -0.001 0.051 *** (24.60) (265.24) (0.15) (0.13) (10.27) t from last round 0.000 -0.099 (0.03) (0.53) t2 from last round -0.067 -0.008 (0.06) (0.27) # rounds 0.047 0.005 *** 1.013 0.003 0.08 *** (0.00) (1384.51) (0.08) (0.86) (14.70) # found startups -0.001 0.051 *** -0.002 -0.013 -0.125 *** (0.23) (73.39) (0.03) (0.32) (16.06) 1(top20 school) -0.012 *** 0.176 *** 0.026 0.021 -0.465 *** (3.47) (694.07) (0.22) (0.88) (11.80) VC Syndicate Features # of VCs -0.009 *** -0.021 -0.319 *** (13.36) (0.50) (7.60) # locations 0.000 0.001 0.023 *** (0.57) (0.06) (12.42) 1(cooperated) 0.002 *** 0.002 0.032 *** (5.21) (0.14) (7.30) # rounds -0.001 -0.043 0.34 *** (0.46) (0.29) (31.20) # categories 0.001 ** 0.000 -0.03 *** (2.54) (0.01) (2.84) 1(top20 school) -0.004 -0.151 -0.774 *** (0.76) (0.71) (49.66) Pair Features days of travel 0.004 -0.05 -1.13 *** (0.82) (0.22) (22.80) 1(funding tie) 0.052 *** 0.136 6.73 *** (16.62) (0.53) (10.51) 1(alumni tie) 0.003 *** -0.001 0.07 *** (16.05) (0.06) (6.62) 98 B. 2007-2014 Influence Selection Growth (r) Exit Value (s) Value Added (v) Funded Unfunded Funded Unfunded [1] [2] [3] [4] [5] r 0.017 ** 0.048 *** 0.035 *** (22.78) (56.00) (70.00) sigma2 0.036 *** 0.159 *** 1.361 *** 1.587 *** (36.96) (60.03) (36.58) (57.20) intercept 0.019 *** 0.085 *** -0.117 -0.147 (4.07) (40.93) (0.29) (1.37) Macro Variables (Y baa −Y us10 ) -0.005 -0.002 *** 0.001 -0.002 (0.20) (6.51) (0.01) (0.08) (rm−r f ) 0.072 0.008 * 0.138 -0.276 (0.11) (1.93) (0.15) (0.50) smb -0.094 -0.009 * -0.457 -0.235 (0.14) (1.90) (0.23) (0.27) hml -0.111 -0.034 *** -0.773 -0.113 (0.16) (8.71) (0.44) (0.17) r f 0.227 0.261 *** 0.181 -0.041 (0.03) (6.17) (0.15) (0.13) Startup Features # locations -0.001 0.049 *** -0.015 0.015 -1.024 *** (0.41) (400.97) (0.52) (0.66) (8.35) # categories -0.002 -0.012 *** 0.02 -0.004 -0.27 *** (0.29) (182.43) (0.71) (0.31) (10.29) # products 0.000 0.005 *** -0.002 0.004 0.035 *** (0.02) (63.69) (0.07) (0.40) (18.53) t from last round 0.001 0.351 * (0.47) (1.67) t2 from last round -0.04 -0.11 *** (0.05) (2.60) # rounds -0.062 0.011 *** 0.337 0.027 *** 0.13 *** (0.00) (944.51) (0.03) (3.43) (25.82) # found startups -0.017 0.016 *** 0.067 0.036 -0.001 (0.63) (25.75) (1.28) (1.29) (0.32) 1(top20 school) 0.007 0.178 *** 0.002 -0.041 -0.28 *** (0.81) (187.21) (0.02) (1.03) (11.46) VC Syndicate Features # of VCs 0.001 -0.004 -0.485 *** (1.22) (0.08) (15.05) # locations 0.002 *** -0.004 -0.002 (7.79) (0.32) (0.90) 1(cooperated) 0.000 0.003 0.042 *** (1.10) (0.45) (19.62) # rounds 0.008 0.037 0.047 *** (1.33) (0.16) (5.63) # categories 0.000 -0.002 -0.026 *** (0.21) (0.35) (16.58) 1(top20 school) 0.004 0.082 -0.458 *** (0.40) (0.53) (19.78) Pair Features days of travel -0.037 *** -0.164 -0.356 *** (4.13) (0.22) (3.97) 1(funding tie) 0.019 *** -0.117 6.179 *** (4.57) (0.29) (11.09) 1(alumni tie) 0.001 -0.004 0.065 *** (0.77) (0.12) (8.29) 99 Table 1-10. Estimation with Alternative Features This table presents the estimation results using the principal factors extracted from the various measures described in Section 3.2 as covariates. Panel A gives the factor loadings separately for the macroeconomic variables, constant and time- varying startup features, VC syndicate features, and pair features. I use f_t, f_j, f_tj, f_i, f_ti, f_tji to denote these factors. Note that there is no f_ji because âĂIJdaysâĂİ (days of a round travel between a VC syndicate and a startup) is the only startup-VC time-invariant feature. Panel B gives the estimation result. The first two columns give the estimates for φ r,y and φ r,n . The third and fourth columns give the estimates for φ s,y and φ s,n . The last column gives the estimates for φ v . Significance at 10%, 5%, and 1% levels are denoted by *, **, and ***. A. Factor Loadings Macro Variables Pair Features f_t f1_t f2_t Constant (Y baa −Y us10 ) 0.328 -0.277 f_ji: days of travel (r m −r f ) 0.135 0.247 Time-varying smb 0.204 0.286 f_tji f1_tji hml -0.187 -0.226 1(funding tie) 0.5095 r f -0.355 0.193 1(alumni tie) 0.0431 Startup Features VC Syndicate Features Constant Constant f_j f1_j f_i f1_i # locations 0.558 # of VCs 0.543 # categories 0.128 # locations 0.849 # products 0.308 Others: location dummies Others: location/category dummies Time-varying Time-varying f_ti f1_ti f2_ti f_tj f1_tj f2_tj 1(cooperated) 0.112 0.16 # rounds 0.085 0.107 # rounds 0.116 0.23 t from last round -0.387 0.495 # categories 0.127 -0.423 t2 from last round -0.044 0.23 1(top20 school) 0.26 -0.145 # startup founded 0.026 0.008 Others: category/school dummies 1(top20 school) 0.207 0.067 degree/major dummies 100 B. Parameter Estimates Influence Selection Growth (r) Exit Value (s) Value Added (v) Funded Unfunded Funded Unfunded [1] [2] [3] [4] [5] r 0.083 *** 0.077 *** 0.039 *** (7.85) (97.29) (141.25) sigma2 0.04 *** 0.096 *** 1.622 *** 0.893 *** (76.87) (2019.42) (50.34) (222.54) intercept 0.655 *** -0.025 *** -0.42 0.062 (42.47) (215.12) (0.22) (0.29) Macro Factors f1_t -0.005 -0.018 *** -0.189 -0.04 ** (0.17) (82.48) (0.70) (2.00) f2_t 0.001 0.005 *** -0.274 0.007 (0.03) (15.06) (0.68) (0.28) Startup Factors f1_j 0.002 0.014 *** 0.056 0.027 * -0.012 *** (0.27) (142.44) (0.66) (1.69) (3.54) f1_tj -0.008 -0.074 *** 0.171 0.894 ** 4.325 *** (0.72) (412.23) (0.09) (2.54) (9.60) f2_tj -0.009 -0.038 *** -0.154 0.38 *** 0.91 *** (0.17) (584.25) (0.04) (4.50) (10.54) f3_tj 0.008 0.262 *** -0.049 -1.248 ** -5.683 *** (1.49) (1106.98) (0.06) (2.51) (8.69) VC Syndicate Factors f1_i 0.006 0.075 0.11 *** (0.45) (0.52) (10.65) f1_ti 0.004 0.066 0.233 *** (0.17) (0.53) (30.77) f2_ti -0.006 -0.039 0.337 *** (0.65) (0.44) (33.17) Pair Factors days of travel 0.000 0.000 -0.042 *** (0.57) (0.65) (23.88) f1_tji 0.081 *** 0.381 1.096 *** (49.14) (0.15) (50.04) 101 Table 1-11. Estimation for Alternative Models This table presents the estimation results for the alternative models. Panel A presents the result for the extended model with an AR(1) structure in the subjective value added. The AR(1) coefficient is denoted by “lagged_v” in the table. Details of the model are given in Section 1.7. Panel B presents the result for the extend model with a hierarchical structure in startup returns. The fixed-effects in startup growth for the pure type of winner, loser, and break-evener are denoted by μ (1) y , μ (2) y , μ (3) y for the previously funded case and by μ (1) n , μ (2) n , μ (3) n for the previously unfunded case. Details of the model are given in Section 1.7. The first two columns give the estimates for φr,y and φr,n. The third and fourth columns give the estimates for φs,y and φs,n. The last column gives the estimates for φv. Significance at 10%, 5%, and 1% levels are denoted by *, **, and ***. A. Autocorrelation in Subjective Value Added Influence Selection Growth (r) Exit Value (s) Value Added (v) Funded Unfunded Funded Unfunded [1] [2] [3] [4] [5] lagged_v 0.009 *** (34.65) r 0.017 *** 0.04 *** 0.033 *** (34.90) (66.66) (48.87) sigma2 0.011 *** 0.092 *** 1.385 *** 1.338 *** (97.84) (773.63) (90.78) (95.24) intercept 0.004 *** 0.005 *** -0.407 * -0.466 *** (3.03) (54.83) (1.81) (4.41) Macro Variables (Y baa −Y us10 ) -0.001 -0.013 *** 0.012 0.029 (0.86) (18.59) (0.40) (1.22) (rm−r f ) 0.007 0.026 *** 0.104 0.548 * (0.74) (38.15) (0.21) (1.67) smb -0.023 0.008 *** -1.78 * -0.721 (0.88) (7.04) (1.82) (1.28) hml 0.036 0.105 *** -0.085 -0.177 (1.59) (4.68) (0.10) (0.33) r f 0.017 0.196 *** 0.145 0.381 *** (0.19) (5.60) (0.58) (3.39) Startup Features # locations 0.003 *** 0.017 *** 0.009 -0.002 -0.861 *** (5.40) (4113.36) (0.44) (0.12) (6.24) # categories -0.009 *** -0.004 *** -0.01 0.001 -0.282 *** (4.35) (1622.62) (0.52) (0.07) (8.87) # products 0.004 *** -0.004 *** 0.004 0.008 0.026 *** (2.33) (1011.76) (0.16) (0.78) (8.21) t from last round 0.000 0.242 * (0.11) (1.74) t2 from last round -0.047 -0.051 *** (0.13) (3.53) # rounds -0.025 0.001 *** 0.135 -0.002 ** 0.023 *** (0.11) (199.13) (0.03) (1.99) (39.64) # found startups -0.005 0.005 *** -0.012 0.002 -0.164 *** (1.55) (122.57) (0.39) (0.05) (11.06) 1(top20 school) -0.019 *** 0.154 *** -0.008 -0.039 -0.438 *** (10.80) (374.70) (0.16) (1.31) (44.48) VC Syndicate Features # of VCs -0.003 *** 0.005 -0.314 *** (18.39) (0.18) (11.45) # locations 0.000 *** 0.001 0.002 *** (8.36) (0.35) (5.77) 1(cooperated) 0.001 *** 0.002 0.028 *** (13.40) (0.59) (14.68) # rounds -0.001 -0.035 0.107 *** (0.69) (0.31) (6.35) # categories 0.000 *** -0.002 -0.019 *** (7.65) (0.69) (13.13) 1(top20 school) 0.039 *** 0.003 -0.638 *** (17.30) (0.04) (26.34) Pair Features days of travel -0.162 * 0.55 -0.72 *** (1.76) (1.39) (13.70) 1(funding tie) 0.004 *** 0.407 * 4.198 *** (4.17) (1.81) (10.15) 1(alumni tie) 0.003 *** -0.002 0.107 *** (19.84) (0.14) (43.43) 102 B. Hierarchical Model for Return Influence Selection Growth (r) Exit Value (s) Value Added (v) Funded Unfunded Funded Unfunded [1] [2] [3] [4] [5] mu1 0.011 1.564 *** (1.01) (33.16) mu2 -0.005 -1.055 *** (0.45) (65.20) mu3 0.006 0.056 *** (0.39) (46.03) r 0.024 *** 0.057 *** 0.036 *** (39.92) (57.85) (264.89) sigma2 0.007 *** 0.105 *** 1.403 *** 1.211 *** (304.83) (642.93) (28.02) (55.72) intercept -0.397 *** -0.686 *** (2.61) (4.40) Macro Variables (Y baa −Y us10 ) 0.000 -0.004 *** 0.021 0.088 *** (0.56) (4.44) (0.69) (3.35) (rm−r f ) 0.017 *** -0.018 *** 0.425 0.465 ** (0.51) (4.52) (0.59) (2.00) smb 0.013 0.043 *** -0.32 -2.655 *** (0.21) (4.42) (0.37) (6.61) hml -0.024 0.022 *** -0.856 -1.032 * (0.31) (2.59) (1.12) (1.95) r f 0.015 ** 0.163 *** 0.297 0.564 *** (2.06) (14.82) (1.19) (5.19) Startup Features # locations 0.000 0.023 *** 0.006 -0.015 -0.565 *** (1.53) (75.70) (0.27) (1.06) (7.60) # categories -0.007 *** -0.006 *** 0.008 -0.016 -0.305 *** (18.16) (26.65) (0.43) (1.62) (11.04) # products 0.002 *** 0.009 *** -0.01 0.022 *** 0.006 *** (5.33) (66.36) (1.04) (2.68) (15.08) t from last round 0.001 0.339 * (1.25) (1.90) t2 from last round -0.062 -0.062 *** (0.19) (3.55) # rounds 0.025 0.002 *** 0.301 -0.002 * 0.022 *** (0.13) (32.92) (0.08) (1.86) (30.32) # found startups -0.011 * 0.035 *** 0.044 0.003 -0.146 *** (1.81) (8.52) (1.00) (0.09) (20.84) 1(top20 school) 0.01 0.138 *** -0.017 -0.085 ** -0.464 *** (1.61) (268.67) (0.35) (2.55) (25.69) VC Syndicate Features # of VCs -0.006 *** 0.016 -0.417 *** (16.03) (0.57) (14.62) # locations 0.003 *** -0.003 0.032 *** (2.11) (0.37) (29.53) 1(cooperated) 0.000 * -0.001 0.03 *** (1.76) (0.42) (16.10) # rounds -0.007 *** 0.058 0.114 *** (4.17) (0.36) (5.63) # categories 0.000 *** 0.001 -0.023 *** (8.06) (0.51) (17.33) 1(top20 school) 0.004 * 0.048 -0.556 *** (1.86) (0.63) (55.21) Pair Features days of travel -0.014 *** 0.206 -0.913 *** (5.68) (1.03) (21.89) 1(funding tie) 0.012 -0.397 *** 4.113 *** (1.52) (2.61) (14.40) 1(alumni tie) 0.001 *** 0.004 0.11 *** (4.67) (0.33) (24.41) 103 Table 1A-1. Locations and Categories of Startup This table presents the location and category summary for the startups in sample. Panel A counts the number of startups that has an office in California (CA), New York (NY), other locations in U.S. except California and New York (OUS), other locations in North America except U.S. (ONA), Asia (AS) and Europe (EU). Panel B counts the number of startups that has an office in the top 20 cities. Panel C counts the number of startups that belongs to the top 20 categories. The classification of category is from TechCrunch. A. Locations by Area Area CA NY OUS ONA AS EU Freq. 2,542 945 2,630 348 1,184 2,085 B. Top 20 Cities C. Top 20 Categories City Freq. Category Freq. San Francisco 882 Software 1,561 New York 809 Mobile 1,222 London 480 Advertising 689 Los Angeles 211 Games 505 Chicago 177 Education 351 Palo Alto 173 Consulting 345 Seattle 165 Internet 340 Mountain View 141 Apps 320 Austin 141 Finance 296 Paris 139 Analytics 295 Toronto 123 Technology 263 Boston 113 Search 255 Bangalore 108 Video 247 San Diego 102 Startups 245 Berlin 99 Networking 238 Cambridge 98 Music 217 Tel Aviv 96 Android 210 Santa Monica 95 Fashion 208 Singapore 86 Design 206 San Jose 80 Travel 194 104 Table 1A-2. Educational Background of Startup Teams This table presents the educational background for the people associated with the startups in sample. The total number of people with valid educational background is 15,342. They are either ever-employed or currently-employed by the startups in sample, including the founders. Panel A counts the number of people who have completed a degree in the top 20 schools. A school is included if it is in the top 20 on the U.S. news rankings or it frequently appears in the startups’ personnel’s educationalbackground. PanelBcountsthenumberofpeoplewhohavecompleted a degree in the three broad fields of engineering, business & economics, and law & politics. Panel C counts the number of people by their highest degrees. A. Top School List School Freq. School Freq. Stanford 1,073 Dartmouth 128 Harvard 804 Oxford 124 NYU 712 Santa Clara 124 UC Berkeley 578 Cambridge 122 Upenn 494 Brown U 122 MIT 472 Boston U 121 Columbia 384 Caltech 118 Northwestern 310 Galtech 115 UCLA 274 UCSB 113 Cornell 273 Georgetown 108 UT Austin 221 U Colorado Boulder 107 U Tel Aviv 213 U Wisconsin Madison 103 USC 209 San Jose State U 93 Yale 199 U Maryland 92 U Chicago 193 INSEAD 89 Carnegie Mellon U 191 PSU 88 U Illinois 180 UC Davis 79 Duke 163 U Waterloo 78 Princeton 162 U Johns Hopkins 68 U Washington 157 B. Field C. Degree Field Freq. Degree Freq. Engineering 7,105 M.S. & M.A. 2,784 Business & Economics 4,967 M.B.A 2,948 Law & Politics 723 Ph.D. 845 105 Table 1A-3. Locations of VC and Funded Categories This table presents the location and funding category summary for the VCs in sample. Panel A counts the number of VCs and VC syndicates that has an office in California (CA), New York (NY), other locations in U.S. except California and New York (OUS), other locations in North America except U.S. (ONA), Asia (AS), and Europe (EU). Panel B and Panel C count the number of VCs and VC syndicates that has an office in the top 20 cities. Panel D counts the number of startups by categories that have received VCs’ funding. The classification of category is from TechCrunch. A. Locations by Area Area CA NY OUS ONA AS EU Freq. VC 343 126 253 21 113 145 VC Syndicate 2,330 1,278 1,650 111 1,278 846 B. Top 20 Cities by VC C. Top 20 Cities by VC Syndicate D. Top 20 Funded Categories City Freq. City Freq. Category Freq. New York 119 Menlo Park 2,175 Software 215 San Francisco 97 New York 1,806 Advertising 206 Menlo Park 84 San Francisco 1,504 Mobile 117 Palo Alto 82 Palo Alto 1,260 Consulting 77 London 70 Beijing 802 Biotechnology 61 Beijing 39 Shanghai 775 Games 55 Boston 39 Herzliya 676 Education 45 Shanghai 30 London 654 Internet 43 Cambridge 29 Cambridge 639 Finance 40 Paris 23 Mumbai 525 Design 34 Seattle 20 Bangalore 489 Search 29 Herzliya 20 Boston 391 Technology 26 Mumbai 19 New Delhi 348 Analytics 25 Chicago 18 Hong Kong 212 Networking 23 Tokyo 16 Mountain View 172 Apps 23 Hong Kong 15 Waltham 166 Media 23 Bangalore 14 Los Angeles 163 Security 21 Singapore 14 Seattle 158 News 21 Austin 13 Philadelphia 153 Video 21 Toronto 13 Tokyo 141 Services 21 106 Table 1A-4. Educational Background of VC Teams This table presents the educational background for the people associated with the VCs in the sample. The total number of people with valid educational background is 11,026. They are either ever-employed or currently-employed by the VCs in the sample, including the founders. Panel A counts the number of people who have completed a degree in the top 20 schools. A school is included if it is in the top 20 on the U.S. news rankings or it frequently appears in the VCs’ personnel’s educationalbackground. PanelBcountsthenumberofpeoplewhohavecompleted a degree in the three broad fields of engineering, business & economics, and law & politics. Panel C counts the number of people by their highest degrees. A. Top School List School Freq. School Freq. Harvard 1,246 INSEAD 126 Stanford 1,211 U Illinois 120 Upenn 669 Carnegie Mellon U 106 NYU 532 UT Austin 96 MIT 458 U Tel Aviv 96 Columbia 398 U Washington 94 UC Berkeley 366 Caltech 93 U Chicago 322 Boston U 85 Yale 223 Santa Clara 84 Princeton 219 LSE 82 Cornell 207 Boston College 76 UCLA 198 U Notre Dame 74 Dartmouth 182 San Jose State U 68 Duke 176 U Wisconsin Madison 65 Oxford 157 U Waterloo 63 Cambridge 148 Washington U 62 Northwestern 140 Galtech 61 Brown U 132 U Johns Hopkins 57 USC 126 B. Field C. Degree Field Freq. Degree Freq. Engineering 4,730 M.S. & M.A. 2,043 Business & Economics 4,272 M.B.A 3,515 Law & Politics 615 Ph.D. 652 107 Table 2-1. Summary of Co-Activism Events This table provides descriptive statistics for co-activism events. The data are from January 1994 through December 2013. Panel A breaks down the number of events by year. Panel B shows the number of events by activist types. We do not double count the type of activists. An event is classified as a hedge fund if it involves at least one hedge fund activist. An event is classified as an activist fund (mutual funds, etc.) if it involves at least one activist fund. An event is classified as an individual if all activists are individuals. Panel C outlines the number of events, the percentage among all events, and the percentage of success and partial success rates by objective categories. Following Brav et al. (2008), an event is classified as successful if its stated goal is achieved and a partial success if its stated goal is partially met through some settlement with the management. Panel D shows the percentage of tactic categories. A. Year of All Events Year Number of Events 1994 2 1995 5 1996 3 1997 11 1998 8 1999 8 2000 13 2001 8 2002 7 2003 9 2004 8 2005 10 2006 6 2007 10 2008 9 2009 8 2010 6 2011 4 2012 3 2013 7 108 B. Number of Events by Activist Types Fund Type Number of Events Percent of Events Individual 70 35.9 Investment Management 13 6.67 Mutual Fund 5 2.56 Hedge Fund 75 38.46 Private Equity/Venture Capital 28 14.36 Pension Funds 4 2.05 C. Objective Categories Objectives Number Percentage Success Partial Success (%) Max shareholder value 65 33.33 61.54 20 Capital Structure: - Excess cash, under-leverage, and dividends/repurchases 1 0.51 0 100 - Equity issuance, restructure debt, and recapitalization 12 6.15 50 16.67 Business Strategy: - Operational efficiency 5 2.56 80 20 - Lack of focus, business restructuring, and spinning off 1 0.51 0 0 - M&A: as target (against the deal/for better terms) 7 3.59 57.14 0 - M&A: as acquirer (against the deal/for better terms) 3 1.54 33.33 33.33 - Pursue growth strategies 5 2.56 80 0 Sale of Target Company: - Sell company or main assets to a third party 5 2.56 80 0 - Take control/buyout company and/or take it private 20 10.26 60 20 Governance: - Rescind takeover defenses 1 0.51 0 0 - Oust CEO, chairman 2 1.03 100 0 - Board independence and fair representation 68 34.87 75 10.29 109 D. Tactic Categories Tactics Percent of Events - Intend to communicate with the board/management on a regular basis with the goal of enhancing shareholder value 55.08 - Seek board representation without a proxy contest or confrontation with the existing management/board 13.9 - Make formal shareholder proposals, or publicly criticizes the company and demands change 15.51 - Threaten to wage a proxy fight in order to gain board representation, or to sue the company for breach of fiduciary duty, etc. 2.67 - Launch a proxy contest in order to replace the board 6.42 - Sue the company 1.07 - Intend to take to take control of the company, for example, with a takeover bid 5.35 110 Table 2-2. Capital Commitment and Exit Strategy of Co-Activism This table presents the capital commitment (in million) and exit strategy of co- activism. Panel A reports the initial and maximum capital investments. The initial capital is the stakes in the initial 13D filings. The maximum capital is the maximum stakes from the subsequent 13D/A filings. The capital is adjusted for inflation, dividends, and stock splits. Panels B breaks down the exit strategy of co-activism, following Brav et al. (2008). A. Capital Commitment Initial Capital Maximum Capital Percentiles Share (%) Mean Median Share (%) Mean Median Bottom 1% 0.5 27.7 27.25 4.9 82.02 88.51 Bottom 10% 2.3 76.99 78.85 5.3 171.98 167.57 Bottom 25% 6 354.46 362.35 7.7 543.6 540.63 Median 10.7 1445.29 1499.81 13.4 1849.51 1872.59 Top 25% 21.1 6499.68 6606.57 25 8004.5 8287.03 Top 10% 45.2 23097.78 23295.13 48 28959.65 28627.5 Top 1% 56.4 55732.96 56041.16 60.7 58038.98 57894.96 Mean 17.4 10872.24 10898.56 20.2 15325.03 15328.51 B. Exit Strategy Categories Percent of Events Sold shares on the open market 5.32 Target company sold 11.17 Target company merged into another 13.3 Liquidated 3.72 Shares sold back to target company 1.6 Still holding/no Information 64.89 111 Table 2-3. Characteristics of Target Firms This table presents the characteristics of firms targeted by co-activism. The accounting variables are all available as of the December prior to the filings. The 1st (2nd) column presents the mean (median) value of firm characteristic variables for target firms. Std Dev is the standard deviation of the value of firm charac- teristic variables. The “Avg Diff” column shows the difference in means between target firms and industry/size/book-to-market matched firms. We report both the t-statistics and the Wilcoxon signed rank statistics on the Avg Diff in the last two columns. Significance at 10%, 5%, and 1% levels are denoted by *, **, and ***. Variable definitions are detailed in Section 2.7. Target Firms Diff with Matched Firms Mean Median Std Dev Avg Diff T-Stat Wilcoxon ZSCORE 0.589 1.878 8.012 -1.426 -1.71 * -0.85 DIV/SHR 0.171 0 0.421 -0.115 -3.72 *** -4.91 *** TA 3164.852 174.091 18539.51 -691.18 -0.4 -3.27 *** REV 915.662 93.934 5061.183 -509.261 -1.12 -3.59 *** ME 729.164 90.105 2532.385 -940.407 -2.6 ** -2.69 *** BM 4.752 0.598 43.832 2.63 0.7 -1.44 Q 1.526 1.211 1.175 -0.133 -1.87 * -3.06 *** GROWTH 1.025 1.012 0.324 -0.365 -2.65 *** -4.44 *** LEV 0.554 0.597 0.253 0.016 0.87 1.22 CASH 382.589 20.767 2237.859 122.049 0.66 -3.41 *** DIVYLD 0.015 0 0.055 -0.037 -3.33 *** -5.27 *** PAYOUT 0.24 0 0.911 -0.056 -0.55 -2.81 *** HHI 0.177 0.097 0.2 0.001 0.03 -2.06 ** ROA -0.042 0.052 0.463 -0.069 -1.71 * -1.8 * CF/ASSETS -0.088 0.01 0.464 -0.073 -1.96 * -1.99 ** CFO/ASSETS 0.009 0.04 0.197 -0.015 -0.92 -0.48 CAPX/ASSETS 0.034 0.021 0.116 -0.016 -1.51 -1.86 * RND/ASSETS 0.094 0.043 0.121 -0.002 -0.18 -1.16 CASH/ASSETS 0.18 0.07 0.219 0.008 0.48 -0.88 CA/ASSETS 0.755 0.718 0.441 0.033 0.91 0.91 DEBT/ASSETS 0.545 0.581 0.249 0.019 1.06 1.31 CDEBT/ASSETS 0.066 0.016 0.098 0.005 0.52 -1.59 LTDEBT/ASSETS 0.194 0.104 0.221 0.026 1.4 0.55 ANALYST 0 0 0 0 0.26 -0.02 INST 0.004 0.004 0.003 0 1.29 0.4 AMIHUD 1.71 1.004 1.698 -0.004 -0.03 -1.55 STKRET -0.095 -0.176 0.455 -0.113 -2.34 ** -2.69 *** ABRET 0.083 0.046 0.283 0.069 2.59 ** 2.73 *** 112 Table 2-4. Probit Analysis of Target Firms This table reports the results of the probit analysis on firms targeted by co- activism. The accounting variables are all available as of the December prior to the filings. The probit analysis uses the industry/size/Book-to-Market matched firms in addition to firms targeted by co-activism. Significance at 10%, 5%, and 1% levels are denoted by *, **, and ***. Variable definitions are detailed in Section 2.7. Target Firms w/ Matched Firms T-Stat ABRET 0.548 ** 2.09 ROA -3.199 *** -4.31 ZSCORE 0.085 ** 2.09 CAPX/TOTAL ASSETS 2.545 1.36 DIV/SHR 0.572 1.34 DEBT/TOTAL ASSETS 19.234 * 1.77 CA/TOTAL ASSETS 0.428 1.49 ME 0 ** -2.36 Q -0.42 *** -4.02 GROWTH -0.216 -0.82 LEV -17.962 * -1.65 DIVYLD -23.737 ** -2.33 RND -3.376 ** -2.18 ANALYST 70267.28 *** 2.85 INST 51.549 * 1.85 HHI 0.703 1.61 INTERCEPT -1.541 ** -2.49 Pseudo R2 25.42% 113 Table 2-5. Abnormal Performance of Target Firms around the Filings Dates This table shows the average buy-and-hold abnormal returns around the filing date. The market-adjusted return is the difference between the target’s buy- and-hold return and the value-weighted NYSE/Amex/Nasdaq index from CRSP. The industry-adjusted return is the difference between the target’s buy-and-hold return and the return for all firms in the target’s Fama-French (1997) 48-industry code. Size-adjusted return is the difference between its buy-and-hold return over a selected time period and the buy-and-hold return for the same time period on the Fama-French size-matched portfolio of firms. The abnormal return over the (-T1, +T2) event window is for the T1 trading days prior to the filing date (day 0) through the T2 trading days afterward. Significance at 10%, 5%, and 1% levels are denoted by *, **, and ***. A. Abnormal Return over Event Window [-20 Day, 5 Day] Percentiles Market Adjusted Industry Adjusted Size Adjusted 10% -0.149 -0.169 -0.145 25% -0.068 -0.042 -0.053 Median 0.023 *** 0.032 *** 0.023 *** 75% 0.134 0.144 0.165 90% 0.359 0.376 0.43 Mean 0.06 *** 0.068 *** 0.069 *** B. Abnormal Return over Event Window [-20 Day, 20 Day] Percentiles Market Adjusted Industry Adjusted Size Adjusted 10% -0.247 -0.164 -0.159 25% -0.07 -0.059 -0.061 Median 0.028 ** 0.039 *** 0.035 *** 75% 0.153 0.167 0.165 90% 0.437 0.387 0.447 Mean 0.05 ** 0.066 *** 0.073 *** 114 C. Abnormal Return over Event Window [-20 Day, 1 Year] Percentiles Market Adjusted Industry Adjusted Size Adjusted 10% -0.739 -0.788 -0.776 25% -0.407 -0.418 -0.511 Median -0.046 -0.116 -0.139 75% 0.332 0.285 0.462 90% 0.872 0.707 0.839 Mean 0.07 0.025 0.026 D. Abnormal Return over Event Window [-20 Day, 5 Year] Percentiles Market Adjusted Industry Adjusted Size Adjusted 10% -1.315 -1.618 -1.438 25% -0.894 -1.139 -1.07 Median -0.354 *** -0.37 *** -0.453 *** 75% 0.257 0.208 0.235 90% 1.408 1.128 1.019 Mean -0.088 -0.307 *** -0.35 *** 115 Table 2-6. Performance of Target Firms by Objectives and the Degree of Hostility This table presents the average buy-and-hold abnormal returns over the [-20 Day, 20 Day] event window by co-activism objectives and the degree of hostility. The market-adjusted return is the difference between the target’s buy-and-hold return and the value-weighted NYSE/Amex/Nasdaq index from CRSP. The industry-adjusted return is the difference between the target’s buy-and-hold return and the return for all firms in the target’s Fama-French (1997) 48-industry code. Size-adjusted return is the difference between its buy-and-hold return over a selected time period and the buy- and-hold return for the same time period on the Fama-French size-matched portfolio of firms. The abnormal return during the (-T1, +T2) event window is for the T1 trading days prior to the filing date (day 0) through the T2 trading days afterward. Panel A shows the breakdown by objectives. Panel B shows the breakdown by hostile and non-hostile. Hostility involves tactics in threatened or actual proxy fight, lawsuit, and takeover (Panel D, Table 2-1). Significance at 10%, 5%, and 1% levels are denoted by *, **, and ***. A. Objectives Market Adjusted Industry Adjusted Size Adjusted Avg Diff Median Diff Avg Diff Median Diff Avg Diff Median Diff Max shareholder value 0.042 0.048 0.045 0.073 0.045 0.051 Capital Structure 0.132 0.039 0.184 0.084 0.231 0.103 Business Strategy 0.182 *** 0.106 *** 0.265 ** 0.197 ** 0.292 ** 0.207 ** Sale of Target Company 0.067 -0.002 0.082 * 0.047 * 0.088 * 0.025 * Governance 0.013 -0.011 0.017 0.004 0.021 0.01 B. Hostile / Non-Hostile Market Adjusted Industry Adjusted Size Adjusted Avg Diff Median Diff Avg Diff Median Diff Avg Diff Median Diff Hostile 0.077 *** 0.042 *** 0.09 *** 0.051 *** 0.101 *** 0.053 *** Non-Hostile -0.036 -0.011 -0.03 0.001 -0.036 -0.014 116 Table2-7. ChangesinFirmCharacteristicsPriorandAfterCo-Activism This table presents the operating performance T1 years before and T2 years after the filing date, where T1=0, 1 and T2=1, 5. The calendar year of the filing date is defined as Year 0. The accounting variables are all available as of the December in year T. Significance at 10%, 5%, and 1% levels are denoted by *, **, and ***. A. [0 Year to 1 Year] Avg Diff T-Stat Median Diff Wilcoxon ZSCORE 0.645 0.75 0.039 0.67 DIV/SHR -0.017 -0.2 0 -1.26 TA -768.287 * -1.88 -24.103 *** -3.93 REV -751.165 -1.18 -8.917 *** -3.68 ME -124.743 -0.62 -16.434 ** -2.48 BM -11.152 -0.98 -0.021 -0.92 Q -1.467 -0.92 -0.019 -0.83 GROWTH 5.27 1.12 -0.005 0.54 LEV -1.164 -1.05 -0.002 -0.58 CASH -139.524 -1.55 -4.55 *** -3.34 DIVYLD -0.015 -0.93 0 -0.3 PAYOUT 0.079 0.21 -0.001 -0.5 HHI 0.004 0.87 0 0.38 ROA 0.011 0.33 -0.007 -0.15 CF/ASSETS 0.024 0.61 0.006 0.38 CFO/ASSETS -0.021 -0.61 -0.009 -0.78 CAPX/ASSETS -0.019 -0.97 0.002 0.43 RND/ASSETS -0.034 -1.4 0 -0.32 CASH/ASSETS 0.004 0.43 -0.012 -0.8 CA/ASSETS -0.006 -0.23 -0.007 -0.9 DEBT/ASSETS -0.013 -0.78 0 -0.35 CDEBT/ASSETS 0.004 0.33 0.003 0.25 LTDEBT/ASSETS 0.003 0.22 -0.009 -1.14 117 B. [-1 Year to 1 Year] Avg Diff T-Stat Median Diff Wilcoxon ZSCORE 0.395 1.02 0.004 0.96 DIV/SHR 0.008 0.23 0 -0.46 TA -1170.8 ** -2.42 -53.034 *** -5.25 REV -1012.2 -1.39 -21.897 *** -5.32 ME -300.86 -1.08 -27.542 *** -3.41 BM 13.48 1.02 -0.079 -1.4 Q -1.212 -0.98 -0.022 -0.98 GROWTH 5.433 1.16 0.019 0.45 LEV -1.048 -1.05 -0.009 -0.74 CASH -132.75 ** -1.99 -4.859 *** -3.55 DIVYLD -0.011 -0.61 0 -0.23 PAYOUT 0.205 0.86 0.002 0.88 HHI 0.007 0.95 0 0.37 ROA 0.041 1.4 0.006 1.33 CF/ASSETS 0.056 1.37 0.007 1.19 CFO/ASSETS 0.007 0.25 -0.01 -0.34 CAPX/ASSETS 0.019 1.31 0.002 0.91 RND/ASSETS -0.027 ** -2.68 0 * -1.8 CASH/ASSETS 0.002 0.13 0.003 0.46 CA/ASSETS -0.009 -0.27 -0.012 -0.4 DEBT/ASSETS 0.001 0.03 -0.008 -0.33 CDEBT/ASSETS -0.016 -1.15 -0.003 -1.17 LTDEBT/ASSETS 0.031 1.31 -0.002 0.36 118 C. [0 Year to 5 Year] Avg Diff T-Stat Median Diff Wilcoxon ZSCORE 0.631 0.31 -0.437 -0.39 DIV/SHR 0.035 0.16 0 -0.43 TA -3867.1 -1.66 -161.56 *** -4.1 REV -603.08 ** -2.55 -48.301 *** -3.56 ME -520.3 ** -2.04 -47.326 *** -2.58 BM 1.127 0.83 -0.056 0.26 Q -0.887 -1.09 -0.018 -0.38 GROWTH 0.154 0.73 0.069 -0.2 LEV 0.023 0.24 0.006 0.67 CASH -122.99 -1.48 -14.17 *** -2.78 DIVYLD -0.011 -0.47 -0.001 -0.74 PAYOUT 0.107 0.53 -0.017 -0.2 HHI 0.008 0.39 0.001 0.46 ROA -0.091 -1.42 -0.011 -0.78 CF/ASSETS -0.114 -1.21 -0.017 -1.51 CFO/ASSETS -0.143 -1.45 0.003 -0.49 CAPX/ASSETS -0.018 -0.3 0.007 * 1.89 RND/ASSETS -0.001 -0.04 -0.004 -1.26 CASH/ASSETS 0.017 0.74 0.047 0.7 CA/ASSETS 0.076 1.65 0.054 1.38 DEBT/ASSETS 0.066 0.72 0.006 1.12 CDEBT/ASSETS -0.041 -0.83 0.009 0.51 LTDEBT/ASSETS 0.045 1.09 0.047 1.1 119 Appendix B Figures 120 Figure 1-1. Sequence of Events This figure presents the sequence of events that happen at the end of t. Existing startups at the beginning oft isE t−1 , after one-period of growth, their growth and implicit exit value become r t and s t , respectively. The law of motions for r t and s t is given by Eq(1) and Eq(2). If the implicit exit value s t ≥δ, the startup goes public or gets acquired (IPO/MA t ); if the implicit exit values t <−δ, the startup goes bankruptcy (D t ); if the implicit value−δ≤ s t < δ, the startup remains in the economy and ready for another round of competition for VC’s funding. The remaining ones are called funding candidates (J t ). The funding is then determined endogenously as a matching between the group of VC syndicatesI and the funding candidates J t . The funding decision is described in Section 2.3. After that, the newborn startups N t come, and the existing startups at the end of t (E t ) consists both J t and N t . A. Sequence of Events 121 B. Growth C. Exit 122 D. Funding E. Separate Growth Trajectories 123 Figure 1-2. Distribution of Imputed Values This figure presents the distributions of the imputed values for the latent variables. PanelAgivesthedistributionsfortheperiod-to-periodgrowth (r j t −r j t−1 )separately for the funded and unfunded cases. Panel B gives the distributions for the implicit exit value s j t separately for the funded and unfunded cases. Panel C gives the pairwise subjective value added v ij t separately for the matched and unmatched pairs of startups and VC syndicates. A. Period-to-Period Growth (r t −r t−1 ) 124 B. Implicit Exit Value s t C. Subjective Value Added v t 125 Figure 1-3. Distribution of Type Mixture This figure plots the distribution of the type mixture in a two-dimensional prob- ability simplex for the extended model with the hierarchical structure in startup returns. Model details are given in Section 1.7. Each point in the simplex charac- teristics a startup-specific vector p j ≡ (p (1) j ,p (2) j ,p (3) j ) which satisfies the condition that p (1) j +p (2) j +p (3) j = 1. I call p j the startup-specific type mixture. The three elements inp j represent the probabilities of the startupj belonging to the winner, loser, andbreak-evenertype. Therefore, thetypemixtureisahiddenstartupfixed- effect. In the simplex, the top vertex represents the pure winner type which has p (1) = 1 or equivalentlyp = (1, 0, 0), the leftmost vertex represents the pure break- evener type which has p (3) = 1 or equivalently p = (0, 0, 1), the rightmost vertex represents the pure loser vertex which has p (2) = 1 or equivalently p = (0, 1, 0). On the center of the gravity, the pink point has p = (1/3, 1/3, 1/3). 126 Figure 2-1. Abnormal Return around the Filing Date The graphs plot the buy-and-hold market-adjusted (blue), size-adjusted (red), and industry-adjusted returns (green) across different event windows. The market- adjusted return is the difference between the target’s buy-and-hold return and the value-weighted NYSE/Amex/ Nasdaq index from CRSP. The industry-adjusted return is the difference between the target’s buy-and-hold return and the return for all firms in the target’s Fama-French (1997) 48-industry code. Size-adjusted return is the difference between its buy-and-hold return over a selected time period and the buy-and-hold return for the same time period on the Fama-French size- matched portfolio of firms. Industry and size-adjusted return is the difference between its buy-and-hold return over a selected time period and the buy-and-hold return for the same time period on the Fama-French 48-industry-, size-, and book- to-market-matched portfolio of firms. [-T1, +T2] denotes the event window from T1 periods prior the Schedule 13D or 13D/A file date to T2 periods afterward. A. [-20 Day, +20 Day] 127 B. [+20 Day, 1 Year] C. [-20 Day, 5 Year] 128 Figure 2-2. Abnormal Share Turnover around the Filing Date The bars plot the percentage increase in the share trading turnover from 20 days prior the Schedule 13D file date to 20 days afterward compared to the average turnover rate during the preceding [-100 Day, -40 Day] event window. 129 References Aizenman, Kendall, and Jake Kendall, 2008, The internationalization of venture capital and private equity, Journal of Economic Studies, 39(5), 488-511 Altman, Edward, 1968, Financial ratios, discriminant analysis, and the prediction of corporate bankruptcy, Journal of Finance, 23, 589-609 Amihud, Yakov, 2002, Illiquidity and stock returns: Cross-section and time- series effects, Journal of Financial Markets, 5, 31-56 Amit, Raphael, Werner Antweiler, and James A. Brander, 2002, Venture capital syndication: improved venture selection vs. the value-added hypothesis, Journal of Economics and Management Strategy, 11(3), 423-452 Baum, Joel, Brian Silverman, 2004, Picking winners or building them? Alliance, intellectual, and human capital as selection criteria in venture financing and performance of biotechnology startups, Journal of Business Venturing, 19(3), 411-436 Bebchuk, Lucian, 2005, The case for increasing shareholder power, Harvard Law Review, 118, 835-914 Bebchuk, Lucian, Alon Brav, and Wei Jiang, 2013, The long-term effects of hedge fund activism, Unpublished Working Paper, Columbia University 130 Bengtsson, Ola and David Hsu, 2010, How do venture capital partners match with startup founders? Unpublished Working Paper Bengtsson, Ola, 2013, Relational venture capital financing of serial founders, Journal of Financial Intermediation, 22(3), 308-334 Bernstein, Shai, Xavier Giroud, and Richard Townsend, 2015, The impact of venture capital monitoring, Journal of Finance, forthcoming Bethel, Jennifer, Julia Liebeskind, and Tim Opler, 1998, Block share purchases and corporate performance, Journal of Finance, 53, 605-634 Black, Bernard S., 1998, Shareholder activism and corporate governance in the United States, in Peter Newman, ed.: The New Palgrave Dictionary of Economics and the Law, (Palgrave Macmillan: New York, NY) Blei, David, Andrew Ng, and Michael Jordan, 2003, Latent dirichlet allocation, Journal of Machine Learning Research, 3(4-5), 993-1022 Bonatti, Alessandro, and Johannes HÃűrner, 2011, Collaborating, American Economic Review, 101, 632-663 Bottazzi, Laura, Macro Da Rin, and Thomas Hellmann, 2011, The importance of trust for investment: evidence from venture capital, Unpublished Working Paper Bottazzi, Laura, Marco Da Rin, and Thomas Hellmann, 2008, Who are the active investors? Evidence from venture capital, Journal of Financial Economics, 89(3), 488-512 Brander, James A, Raphael Amit and Werner Antweiler, 2002, Venture-capital syndication: improved venture selection vs. the value added hypothesis, Journal of Economics and Management Strategy, 11, 423-452 131 Brav Aon, Wei Jiang, Frank Partnoy, and Randall Thomas, 2008, Hedge fund activism, corporate governance, and firm performance, Journal of Finance, 63, 1729-1775 Carter, Chris K. and Robert J. Kohn, 1994, On Gibbs sampling for state space models, Biometrika, 81, 541-553 Chemmanur, Thomas J., Karthik Krishnan, and Debarshi K. Nandy, 2011, How does venture capital financing improve efficiency in private firms? A look beneath the surface, Review of Financial Studies, 24(12), 4037-4090 Chen, Henry, Paul Gompers, Anna Kovner, and Josh Lerner, 2010, Buy local? The geography of successful and unsuccessful capital expansion, Journal of Urban Economics, 67(1) Cochrane, John, 2005, The risk and return of venture capital, Journal of Financial Economics, 75(1), 3-52 Cumming, Douglas, Fleming, and Suchard, 2005, Venture capitalist value- added activities, fundraising and drawdowns, Journal of Banking & Finance, 29(2), 295-331 Da Rin, Marco, Thomas Hellmann, and Manju Puri, 2012, A survey of Venture Capital Research Daniel, Kent, Mark Grinblatt, Sheridan Titman, and Russ Wermers, 1997, Measuring Mutual Fund Performance with Characteristic-Based Benchmarks, Journal of Finance, 52, 1035âĂŞ58 Demsetz, Harold, and Kenneth Lehn, 1985. The structure of corporate ownership: causes and consequences, Journal of Political Economy, 93, 1155-1177 132 Ewens, Michael, 2009, A new model of venture capital risk and return, Unpublished Working Paper, SSRN Fama, Eugene and Kenneth French, 1995, Size and book-to-market factors in earnings and returns, Journal of Finance, 50, 131-155 Fama, Eugene, and Kenneth French, 1993, Common risk factors in the returns on bonds and stocks, Journal of Financial Economics, 33, 3-53 Fama, Eugene, and Kenneth French, 1997, Industry costs of equity, Journal of Financial Economics, 43, 153-193 Fruhwirth-Schnatter, Sylvia, 1994, Data augmentation and dynamic linear models, Journal of Time Series Analysis, 15, 183-202 Fulghieri, Paolo, and Merih Sevilir, 2009, Size and focus of a venture capitalists portfolio, The Review of Financial Studies, 22(11), 4643-4680 Gillan, Stuart, and Laura Starks, 2007, The evolution of shareholder activism in the United States, Journal of Applied Corporate Finance, 19, 55-73 Goerge Constantinides, Milton Harris, and Rene Stulz (editors), Handbook of the Economics of Finance, vol 2, Amsterdam, North Holland Gompers, Paul, 1994, The rise and fall of venture capital, Business and Economic History, 23, 1-26 Gompers, Paul, Anna Kovner, Josh Lerner, 2009, Specialization and success: evidence from venture capital, Journal of Economics and Management Strategy, 18(3), 817-844 Gompers, Paul, Anna Kovner, Josh Lerner, and David Scharfstein, 2010, Performance persistence in entrepreneurship, Journal of Financial Economics, 96, 133 18-32 Gompers, Paul, Josh Lerner, 1997, Risk and reward in private equity invest- ments: the challenge of performance assessment, Journal of Private Equity, 1, 5-12 Grossman, Sanford, and Oliver Hart, 1980, Takeover bids, the free-rider problem, and the theory of the corporation, Bell Journal of Economics, 11, 42-64 Guercio, Diane Del, and Jennifer Hawkins, 1999, The motivation and impact of pension fund activism, Journal of Financial Economics, 52, 293âĂŞ340 Hamilton, Barton, Jack Nickerson, and Hideo Owan, 2003, Team incentives and worker heterogeneity: An empirical analysis of the impact of teams on productivity and participation, Journal of Political Economy, 111, 465-497 Hellmann, Thomas, and Manju Puri, 2000, The interaction between product market and financing strategy: the role of venture capital, The Review of Financial Studies, 13(4),959-984 Hellmann, Thomas, and Manju Puri, 2002, On the fundamental role of venture capital, The Economic Review, published by the Atlanta Federal Reserve Bank, 87, No. 4 Hellmann, Thomas, and Manju Puri, 2002, Venture capital and the profes- sionalization of the startup firms: empirical evidence, Journal of Finance, 57(1), 169-197 Himmelberg, Charles, Glenn Hubbard, and Darius Palia, 1999. Understanding the determinants of managerial ownership and the link between ownership structure and performance, Journal of Financial Economics, 53, 353-384 Hochberg, Yael V., Alexander Ljungqvist, and Yang Lu, 2007, Whom you know matters: venture capital networks and investment performance, Journal of 134 Finance, 62(1), 251-301 Hochberg, Yael V., Alexander Ljungqvist, and Yang Lu, 2010, Networking as a barrier to entry and the competitive supply of venture capital, Journal of Finance, 65(3), 829859 Hochberg, Yael V., Laura Anne Lindsey, and Mark M. Westereld, 2015, Resource accumulation through economic ties: evidence from venture capital, Journal of Financial Economics, forthcoming Hochberg, Yael V., Michael J. Mazzeo, Ryan C. McDevitt, 2015, Specialization and competition in the venture capital industry, Review of Industrial Organization, 46(4), 323-347 Hsu, H. David, 2004, What do entrepreneurs pay for venture capital affiliation? Journal of Finance, 59, 1805-1844 Hsu, H. David, and Ola Bengtsson, 2010, How do venture capital partners match with startup founders? Unpublished Working Paper Jensen, Michael C., 1986, Agency costs of free cash f low, corporate finance, and takeovers, American Economic Review, 76, 323âĂŞ329 Kaplan, Steven, Mark Klebanov, and Morten Srensen, 2012, Which CEO characteristics and abilities matter? Journal of Finance, 67(3), 973-1007 Karpoff, Jonathan, 2001. The impact of shareholder activism on target companies: a survey of empirical findings, Unpublished working paper, University of Washington Karpoff, Jonathan, Paul H. Malatesta, and Ralph A. Walkling, 1996, Cor- porate governance and shareholder initiatives: Empirical evidence, Journal of Financial Economics, 42, 365-395 135 Klein Aprial, and Emanuel Zur, 2009, Entrepreneurial shareholder activism: hedge funds and other private investors, Journal of Finance, 64, 187-229 Korteweg, Arthur and Srensen Morten, 2010, Risk and Return Characteristics of Venture Capital-backed Entrepreneurial Companies, Review of Financial Studies, 23(10), 3738-3772 Korteweg, Arthur, Markov Chain Monte Carlo Methods in Corporate Finance. In: P. Damien, P. Dellaportas, N. Polson, and D. Stephens (Eds.), Bayesian Theory and Applications, Oxford University Press Lerner, Josh, 1994, The syndication of venture capital investments, Financial Management, 23(3), 16-27 Lerner, Josh, 1995, Venture capitalists and the oversight of private firms, Journal of Finance, 50, 301-318 McCahery Joseph, Zacharias Sautner, and Laura Starks, Behind the Scenes: The Corporate Governance Preferences of Institutional Investors, Unpublished Working Paper, Tilburg University Nanda, Ramana and Matthew Rhodes-Kropf, 2013, Investment cycles and startup innovation, Journal of Finance Economics, 110(2), 403-418 Phalippou, Ludovic, 2010, Venture capital funds: flow-performance relation- shipsandperformancepersistence, Journal of Banking and Finance, 34(3), 568-577 Roe, Mark, 1994, Strong Managers Weak Owners: The Political Roots of American Corporate Finance (Princeton University Press, Princeton, N.J.) Sahlman, William. 1990, The structure and governance of venture-capital organizations, Journal of Financial Economics, 27, 473-521 136 Short, Helen, 1994, Ownership, control, financial structure and the perfor- mance of firms, Journal of Economic Surveys 8, 203-249 Slovin, Myron, and Marie Sushka, 1993, Ownership concentration, corporate control activity, and firm value: evidence from the death of inside blockholders, Journal of Finance, 48, 1293-1322 Song, Moon, and Ralph Walking, 1993, The impact of managerial ownership on acquisition attempts and target shareholder wealth, Journal of Financial and Quantitative Analysis, 28, 439-457 Sorensen, Morten, 2005, An economic and econometric analysis of market sorting with an application to venture capital, dissertation (Stanford University) Sorensen, Morten, 2007, How smart is smart Money? A two-sided matching model of venture capital, Journal of Finance, 62(6), 2725-2762 Sorensen, Morten, 2008, Learning by investing: evidence from venture capital, Unpublished Working Paper Sorenson, Olav, and Toby Stuart, 2001, Syndication networks and the spatial distribution of venture capital investments, American Journal of Sociology, 106, 1546-1588 Tykvova, Tereza, 2007, Who chooses whom? Syndicate, skills, and reputation, Review of Financial Economics, 16(1), 5-28 Wahal, Sunil, 1996, Pension fund activism and firm performance, Journal of Financial and Quantitative Analysis, 31, 1-23 Xuan, Tian, 2010, The causes and consequences of venture capital stage financing, Journal of Financial Economics, 101(2), 132-159 137 Yermack, David, 2010, Shareholder voting and corporate governance, Annual Review of Financial Economics, 2, 103-125 138

Abstract (if available)

Abstract My thesis consists of two essays in financial intermediation. ❧ The first chapter is ""The Impact of Venture Capital on the Life Cycles of Startups"". It shows that venture capitalists (VCs) endogenously improve the quality of funded startups. In turn, this makes startups more likely to get subsequent investments. The resulting feedback effect amplifies VCs' impact over time. To identify startup quality as a cause and an effect of funding separately, I develop and estimate a dynamic model of funding that incorporates multiple funding rounds. Using CrunchBase, a novel database that includes unfunded startups, I estimate both the differential impact of being funded by one VC versus another, and the overall impact of being funded by a VC versus remaining unfunded. A simulation exercise shows that the feedback effect magnifies the overall impact of VCs by a factor of seven. ❧ The second chapter is ""Is Collaborative Activism Effective?"". We define collaborative activism (co-activism) as a set of independent activists that pursue the same objective and work together to influence corporate decisions. Using a hand-collected dataset from 1994 to 2013, we find that co-activism targets underperforming firms, and is mostly nonhostile despite its initial attempt to control around 10% of stocks. Co-activism succeeds in roughly 80% of the cases through discussions with the management. The market responds favorably to co-activism around the date of filing, but market expectation tapers off over the subsequent year. Our evidence suggests that co-activism aims for short-term gains at the expense of long-term shareholder value, and pressures the companies to take myopic actions that are harmful in the long run.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Evolution of returns to scale and investor flows during the life cycle of active asset management

PDF

Essays in tail risks

PDF

Essays in corporate finance

PDF

Essays in behavioral and entrepreneurial finance

PDF

Innovation: financial and economics considerations

PDF

Essays on delegated portfolio management under market imperfections

PDF

Three essays in derivatives, trading and liquidity

PDF

Two essays on financial econometrics

PDF

Essays on real options

PDF

Essays on information and financial economics

PDF

Essays in corporate finance

PDF

Essays on understanding consumer contribution behaviors in the context of crowdfunding

PDF

Essays in empirical asset pricing

PDF

Essays on the effect of cognitive constraints on financial decision-making

PDF

Essays on interest rate determination in open economies

PDF

Pricing OTC energy derivatives: credit, debit, funding value adjustment, and wrong way risk

PDF

Two essays on major macroeconomic shocks in the Japanese economy: demographic shocks and financial shocks

PDF

Essays on service systems with matching

PDF

Efficient policies and mechanisms for online platforms

PDF

Simultaneous center of mass estimation and foot placement selection in complex planar terrains for legged architectures

Asset Metadata

Creator Ling, Yun (author)

Core Title Essays in financial intermediation

School Marshall School of Business

Degree Doctor of Philosophy

Degree Program Business Administration

Publication Date 03/03/2017

Defense Date 10/11/2016

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag collaborative activism,financial intermediation,hedge fund,OAI-PMH Harvest,venture capital

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Korteweg, Arthur (committee chair), Ahern, Kenneth (committee member), Matsusaka, John (committee member), Phillips, Gordon (committee member), Zapatero, Fernando (committee member)

Creator Email yling@usc.edu,yunling0115@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-346169

Unique identifier UC11258175

Identifier etd-LingYun-5116.pdf (filename),usctheses-c40-346169 (legacy record id)

Legacy Identifier etd-LingYun-5116.pdf

Dmrecord 346169

Document Type Dissertation

Rights Ling, Yun

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA