Why Do Entrepreneurs Switch Venture Capitalists?

Transcription

1 Why Do Entrepreneurs Switch Venture Capitalists? Douglas Cumming Schulich School of Business York University 4700 Keele Street Toronto, Ontario M3J 1P3 Canada Na Dai 1 School of Business SUNY-Albany 1400 Washington Avenue Albany, NY Dai (Corresponding author): ndai@uamail.albany.edu; phone: ; fax: We owe thanks to the seminar participants at ESEADE, SUNY at Buffalo, Florida State University, College of William and Mary. This paper has been accepted for presentation at the 22 nd Australasian Finance and Banking Conference at Sydney, Australia. 1

2 Why Do Entrepreneurs Switch Venture Capitalists? Abstract We examine the dynamics of the positive sorting in the venture capital industry. We empirically show that higher-quality companies associated with less reputable VCs are more likely to switch to more reputable ones. Conversely, lower-quality companies are more likely to switch to less reputable VCs when all the existing VCs stop investing in them. Companies that switch to more reputable VCs obtain smaller investment size and lower pre-money valuation in follow-on rounds. It takes significantly more time for lower-quality switchers to obtain subsequent financing. Keywords: Venture Capital; Switching JEL Codes: G24, G32 2

3 A venture capital firm should deliver and focus on its core competency and move on. Just like startups change CEOs as they mature, shouldn't companies change VCs as they mature? If there is a good startup CEO, shouldn't there also be good startup VCs? Some people can take a company from startup idea to billion-dollar business, but most need to be replaced along the way-this is true for both management teams and board members. Source: Vator News, Lessons learned from entrepreneur, by Auren Hoffman, August 27, Short URL: There exists substantial uncertainty in venture capital investments, especially in the early stages. It is well established that there exists a positive sorting mechanism in the venture capital market in which more reputable venture capitalists (VCs) invest in better companies (Sørensen (2007); Knill (2009)). But venture capital-backed companies typically take two to seven years before they come to fruition in an exit, such as an IPO or an acquisition. As both VCs and entrepreneurs learn about the potential of a start-up venture over time, it is natural to expect the two-sided matching between VCs and entrepreneurs to dynamically adjust. In this paper, we empirically examine, for the first time, the factors that influence the continuation or the break-up of the entrepreneur-vc relationship. Switching VCs is more common than one might expect. We show that during the period from 1991 to 2002, 42% of the follow-on rounds of financing have lead VCs that are different from those of previous rounds. In 17% of the rounds, lead VCs from previous rounds did not participate in the subsequent rounds at all. Furthermore, in about 9% of the rounds, all previous VCs (lead or syndicated) stopped investing in subsequent rounds. In this paper, we explore these switching dynamics between entrepreneurs and their lead VCs. As well, we examine the impact 3

4 of VC reputation and financial terms on switching decisions, and investigate whether companies switch to more reputable VCs regardless of less favorable financial terms, such as valuations. The switching process can be initiated by VCs or entrepreneurs. We hypothesize and empirically examine three underlying reasons why a VC and/or an entrepreneur would want to terminate an existing relationship. These hypotheses are: the information hold-up hypothesis, the graduation hypothesis, and the better-deal term hypothesis. The information hold-up hypothesis suggests that outside VCs will interpret the failure to reinvest by existing VCs as a negative signal regardless of the true quality of the entrepreneurial firm. Switching only happens when existing VCs are not satisfied with the performance or the potential of the entrepreneurial firms. If entrepreneurs are able to obtain any subsequent funding from other VCs at all, then the new VCs are likely to be less reputable than the existing ones. The new VCs are also likely to charge a higher discount rate, which indicates a lower valuation. The graduation hypothesis argues that entrepreneurs will initiate switching if they view that the benefits of having a more reputable VC rather than the existing VC are greater than the cost of switching. This hypothesis predicts that ventures have a higher probability of success if entrepreneurs are successful in obtaining subsequent funding from more reputable outside VCs. To get the certification of more reputable VCs, entrepreneurs are likely to accept less favorable financial terms. The better deal term hypothesis proposes that entrepreneurs re-select VCs at each round based on the bids made by both existing and outside VCs. Whoever bids the best deal terms, such as a larger capital infusion and a higher pre-money valuation, becomes the lead VC. This type of switching is also initiated by the entrepreneurs. Based on this hypothesis, we can observe 4

5 that switchers obtain better-deal terms than non-switchers. The reputation of the VCs, on the other hand, is of second-order importance. Our empirical analysis starts with analyzing the determinants of switching. We correct for the potential survivorship bias by using the Heckman selection model. Then we apply the Heckman two-stage framework to control for the endogeneity of switching and examine the impact of switching on subsequent round financial terms and searching cost. Three major findings emerge from our analysis. First, we find that higher-quality (lowerquality) entrepreneurial firms associated with less (more) reputable VCs are more likely to switch. This trend suggests that the learning of new information regarding the potential of the company (either positive or negative) by VCs and entrepreneurs drives the adjustment of the matching to reach the positive sorting equilibrium. Second, we find that higher-quality entrepreneurial firms are more likely to have more reputable new lead VCs although existing ones continue to participate in subsequent rounds. But lower quality firms are more likely to have less reputable new lead VCs when none of the existing VCs continue to invest in the company. Third, we document that higher-quality switchers in general obtain a smaller investment but a higher valuation in follow-on rounds than non-switchers, but they accept a lower valuation if they switch to more reputable VCs. Conversely, lower-quality switchers obtain a larger investment but a lower valuation than their non-switching peers. Furthermore, it typically takes more time for companies that switch to new VCs to obtain the subsequent round of financing. This is particularly true for lower-quality switchers. Overall, these empirical findings support the graduation hypothesis that higher quality entrepreneurial firms are willing to accept less favorable financial terms to be associated with more reputable VCs. Our findings also provide support to the information hold-up hypothesis by showing that it takes more time for 5

6 lower-quality firms to obtain follow-on round financing. Furthermore, they switch to less reputable VCs and obtain a lower valuation because none of their existing VCs are willing to continue investing in the company. Our paper relates to the theoretical and empirical literature on how firms select/switch financial intermediaries in bank lending and equity offerings. The former includes but is not limited to Greenbaum, et al (1989), Sharpe (1990), Diamond (1991), Rajan (1992), Boot and Thaker (1994), Berger and Udell (1995), Padilla and Pagano (1997), Degryse and Cayseele (2000), Schenone (2004), and Hale and Santos (2009). Among the latter, Chemmanur and Fulghieri (1994) and Fernando, Gatchev, and Spindt (2005) predict that, at equilibrium, higherquality firms are associated with high reputation underwriters even though they disagree on whether more reputable underwriters charge higher fees. Krigman, Shaw, and Womack (2001) empirically show that in the mid-1990s, 30% of the firms that conducted an SEO within three years of their IPO switched lead underwriter. They argue that the main reasons for switching are that firms graduate to higher reputation underwriters, and they strategically buy additional and influential analyst coverage from the new lead underwriter. In contrast, Ljungqvist, Marston, and Wilhelm (2006) find that aggressive analyst behavior does not increase their bank s probability of winning an underwriting mandate. Instead, the main determinant of the lead-bank choice is the strength of prior underwriting and lending relationships. In the venture capital literature, Sørensen (2007) suggests that there is a positive sorting between VCs and entrepreneurs. Hsu (2004) documents that entrepreneurs favor more reputable VCs and are willing to turn down the best financial terms in order to be associated with reputable VCs. Our paper contributes to this body of literature by analyzing the dynamics of a positive sorting in the VC industry due to the two-sided selection between entrepreneurs and VCs as new information is learned by both 6

7 parties. We also empirically show the roles of VC reputation and firm quality in this dynamic matching process. The remainder of the paper is organized as follows. Section I describes the three hypotheses. Section II describes the data and reports the summary statistics. Section III examines the fundamental reasons why switching happens. Section IV explores financial terms in followon rounds of switchers in comparison to non-switchers, and controls for the self-selection of switching. The final section concludes the paper and discusses the implications of our findings. I. Hypotheses Developments A. Information Hold-Up Hypothesis Staged financing is a commonly-used technique in the venture capital industry. The literature typically regards the staged financing works as a control or monitoring mechanism (Metrick (2006)), because an entrepreneur has to come back to VCs for funding at several points. Thus, it allows VCs to monitor the firm and to shut it down (i.e. not fund it) if the probabilities of success are poor. On the other hand, VCs that invest in earlier rounds of financing acquire soft information regarding management skills and the quality of the entrepreneurial firm, which are typically not available to outside VCs. Thus the asymmetric evolution of information between inside VCs and outside VCs allows the existing VCs to hold up the entrepreneurs when the firm needs the next round of financing, because outside VCs will interpret the failure to reinvest as a negative signal regardless of the true quality of the entrepreneurial firm. As a consequence, entrepreneurs face high costs to switch venture capitalists in subsequent rounds of financing. The implication is that entrepreneurs lack incentives to voluntarily switch venture capitalists. 7

8 Switching happens only when VCs are not satisfied with the performance or the potential of the entrepreneurial firm. If entrepreneurs are able to obtain any subsequent funding at all, the new VCs are less reputable than the existing VCs. Furthermore, the new VCs charge a higher discount rate, indicating a lower valuation. The predictions of the information hold-up hypotheses are: H1-a: Lower-quality entrepreneurial firms are more likely to switch (forced switch, abandoned by existing VCs). H1-b: New VCs are less reputable than existing VCs. H1-c: Firms switching VCs receive less favorable contract terms (e.g., valuation) than firms that do not switch. B. Graduation Hypothesis The economic benefits to an entrepreneurial firm of associating itself with reputable VCs appear to be well established. Lerner (1994) reports that seasoned VCs appear to be particularly proficient at taking companies public near market peaks. Gompers (1996) finds that younger VC firms strive for quick formation of their reputation in the VC industry by taking their portfolio companies public sooner, and underpricing them. Hochberg, Ljungqvist, and Lu (2007) document that VCs with a broader network increase the likelihood of successful exits for their portfolio companies. Sørensen (2007) shows that companies funded by more experienced VCs are more likely to go public primarily due to their ability to source better investments. Nahata (2008) finds that companies backed by more reputable VCs are more likely to exit successfully, access public markets faster, and have higher asset productivity at IPOs. Ivanov et al. (2008) find that companies backed by more reputable VCs have better post-issuance performance and better- 8

9 quality corporate governance. Chemmanur, Krishnan, and Nandy (2008) show that firms backed by more reputable VCs are able to improve productivity more in subsequent financing than those backed by less reputable VCs. Studies have also shown that entrepreneurs are aware of the benefits associated with VC reputations. According to Hsu (2004), entrepreneurs select offers among competing VC investors not only based on the financial terms, but more often by considering the reputation of the VC investors. If entrepreneurs view the benefits from a more reputable VC rather than the existing VCs as greater than the cost of switching, then we expect the switching to happen. This type of switching has an implication for the quality of the entrepreneurial firms. That is, ventures have higher probabilities of success if entrepreneurs are successful in obtaining subsequent funding from more reputable outside VCs. In order to get the certification of more reputable VCs, entrepreneurs are likely to accept less favorable financial terms. Thus, predictions based on the graduation hypothesis are as follows: H2-a: Higher-quality entrepreneurial firms are more likely to switch VCs. H2-b: New VCs are more reputable than existing VCs. H2-c: Entrepreneurial firms that switch to more reputable VCs are likely to accept less favorable contract terms (e.g., valuation) than those that do not switch. C. Better Deal Terms Hypothesis Alternatively, entrepreneurs re-select VCs at each round based on the bids made by both existing and outside VCs. Whoever bids the better deal terms, such as larger capital infusion and higher pre-money valuation, becomes the lead VC. Presumably, higher-quality entrepreneurial firms are more likely to attract competing bids from outside VCs, therefore, the better deal terms 9

10 hypothesis, similar to the graduation hypothesis, suggests that higher-quality entrepreneurial firms are more likely to switch. The difference between the better deal hypothesis and the graduation hypothesis lies in the reason that entrepreneurial firms switch VCs. The former suggests that entrepreneurial firms switch VCs in order to get a larger investment and a higher valuation, and the reputation of the lead VCs are of secondary importance; the latter emphasizes that entrepreneurial firms switch to be associated with more reputable VCs even at less favorable deal terms. Thus, the better deal hypothesis predicts the following: H3-a: Higher-quality entrepreneurial firms are more likely to switch lead VCs. H3-b: New VCs are not necessarily more reputable than existing VCs. H3-c: New VCs offer a greater investment and a higher valuation. II. Data and Sample The data on venture investments, including the valuation data, come from the VentureXpert database provided by the Thomason Financial Corporation. We start with all VC investments made in the U.S. from 1991 to We exclude entrepreneurial firms whose firstround financing was before Then we exclude investments by VC investors other than firms investing their own capital. 3 To be included in the sample, the observation also has to have the amount invested in each round, the size (commitment) of the lead VC fund, and the zip codes of the lead VC investor and the new venture. 2 We give a minimum five-year window from the investment to observe whether the venture has exited as an IPO or an acquisition (by the end of 2007), which requires excluding the financing rounds after We do not include other types of VC investors, such as corporate VC, financial institutions affiliated with VC funds, pension funds, and university foundations, etc. The purpose is to avoid the potential impact of organizational forms of VC investors on the switching decision. 10

11 The lead VC investor is defined as having a 10% or greater investment in a specific round of financing. This requirement ensures that the lead VC has a significant impact on the negotiation of the contract and the monitoring of the entrepreneurs. Our results are robust to alternative definitions of this band-width, for which we have considered a range from 5-15%. Also, note that financings which show up on proximate dates within a few weeks are treated as part of the same round, so as to not doublecount switches. These data filters leave us 19,871 rounds of VC investments in 9,242 entrepreneurial firms with relevant data. Among the 9,242 entrepreneurial firms, 4,982 firms have received more than one round of financing during the sample period. We investigate how these 4,982 entrepreneurial firms make their decisions on whether to switch VCs in subsequent rounds of financing. However, we only observe switching behavior among firms that have successfully obtained multiple rounds of financing, which itself can be endogenous. Without appropriately controlling for this selection effect, our results can be impacted by the potential survivorship bias. Therefore, in the analysis that follows, we apply the Heckman two-stage framework to correct for this survivorship bias. Thus our final sample consists of 14,889 rounds consisting of 4,260 observations of firms that only raised one round and 10,629 follow-on rounds of firms that have obtained multiple rounds. We define switching at the following three levels. Switching 1 is defined as when the lead VC firm in the follow-on round of financing is different from the lead VC firm in previous rounds. Switching 2 is defined as when the lead VC in the previous round does not participate in the follow-on round of financing. Finally, we define Switching 3 as when not only the lead VC but also other existing VCs do not participate in the follow-on round of financing. During the period from 1991 to 2002, 42% of the follow-on rounds of financing have lead VCs different from those of previous rounds. In 17% of the cases, the existing lead VCs do 11

12 not invest in the follow-on rounds at all and in 9% of the cases, all previous VCs stop investing in the follow-on rounds. [Insert Table I here.] Panel A of Table II summarizes the characteristics of entrepreneurial firms that switched their VC in comparison to those who did not. Specifically, we consider the industry, stage, location, and final exit of the entrepreneurial firms. We show that technology firms, firms at their later stage, and firms located in California are more likely to switch. Firms at their expansion stage are more likely to have different lead VCs, although their previous VCs often continue investing in later rounds. The table also indicates that switchers (all three levels) are more likely to go public, but they are less likely to get acquired if all previous VCs stop financing the subsequent rounds. [Insert Table II here.] Panel B of Table II compares the characteristics between existing lead VCs of switchers and those of non-switchers. We consider four aspects of VC investors: VC firm age, VC fund size, the VC firm s aggregate investment market share, and the distance between the lead VC and the entrepreneurial firm. VC firm age is measured as the difference between the round date and the founding date of the firm. VC fund size is measured by the total capital commitment to the VC fund, in millions of 2006 dollars. Following the method proposed by Nahata (2008), we measure VC firm aggregate investment market share as each VC firm s cumulative dollar investment from 1987 until a given calendar year and normalize it by the overall aggregate investment in the VC industry up to that same calendar year. 4 To measure the geographic distance between the lead VC and the entrepreneurial firm, we obtain the latitude and longitude 4 Our results are invariant to different definitions of VC reputation. For example, we considered a measure of VC reputation based on a rolling 5-year window of market share, among other sensitivity checks. 12

13 data for the center of each zip code from the U.S. Census Bureau s Gazetteer and estimate the distance between the centers of the two zip codes by using the following equation: d ij [ sin( lat ) sin( lat ) + cos( lat ) cos( lat ) cos( long long )] = 3963 ar cos i j i j i j, (1) where latitude (lat) and longitude (long) are measured in radians. Panel B of Table II suggests that the existing lead VCs of switchers are significantly smaller than non-switchers and significantly farther away from the entrepreneurial firms. Existing lead VCs of switchers have smaller market share than those of non-switchers, but this is not statistically significant. In all of the three switching scenarios, the lead VC in the follow-on round is different from the previous one. Herein, we consider how the characteristics of lead VCs change with switching. Specifically, we define entrepreneurial firms that change to more senior VCs, VCs with larger investment market share, larger VCs, and closer VCs as moving up the ladder and the opposite as moving down. As shown in Panel A of Table III, slightly more than 50% of the ventures that have switched VCs become associated with less reputable (younger and with smaller investment market share) and more distant VCs. In Panel B of Table III, we further divide the moving-up firms and moving-down firms into quartiles based on the changes in lead VC characteristics, respectively. Firms that fall in the lowest quartile in either direction are assigned to a no change subgroup. Moving-up firms that fall in the highest three quartiles are assigned to a positive change subgroup and those movingdown firms that fall in the highest three quartiles are assigned to a negative change subgroup. We then compare the changes in lead VC characteristics between the positive change and the negative change subgroups. [Insert Table III here.] 13

14 As shown in Panel B of Table III, the changes in lead VC characteristics after switching in either direction are significant. For instance, the mean and median lead VC firm age of the positive change group increases by 14 and 11, respectively, in comparison to these firms previous lead VC. In contrast, the negative change group s lead VC firm age declined by almost the same magnitude. The lead VCs investment market share of the positive change group ( negative change group) increases (decreases) by 2.4% on average (1.3% at the median). The change in lead VC fund size is also large. The average fund size of the lead VC increases by $370 million for the positive change group and declines by $258 million for the negative change group. The median change for the former is $164 million and $130 million for the latter. Some entrepreneurial firms switch to much closer VCs (on average, more than 1000 miles closer), but some firms show the opposite pattern. III. Why Entrepreneurs Switch Lead VCs? A. Existing VC Reputation, Entrepreneurial Firm Quality, and the Decision to Switch In this section, we examine how the decision to switch is impacted by the various characteristics of the existing lead VCs and the entrepreneurial firms. As we mentioned earlier, we only observe switching behavior when firms successfully obtain more than one round of financing, which is endogenous itself. To make sure this selection effect does not introduce systematic bias (survivorship bias) in our results, we apply the Heckman sample selection approach to address this concern. Specifically, we run a first stage probit regression in which the dependent variable is a dummy that is equal to one if the firm has 14

15 successfully obtained more than one round of financing during our sample period, and zero otherwise. The independent variables for this analysis include the stage of the firm (seed, early, or later stage), the industry of the firm (computer-related, communication, semiconductor, biotechnology, medical, or non-technology), the location of the firm (CA, MA, or other areas), the natural logarithm of the geographic distance between the firm and the VC, a syndicate dummy which is equal to one if more than one VC invests in a round, and round size measured as the natural logarithm of round investment. We also include the year dummies in the specification. Rho (the correlation coefficient between error terms of the two equations) is estimated. A significant Rho indicates that the selection effect is important and that the Heckman technique should be used to correct for the potential bias. In our second stage regressions, we analyze the determinants of switching with the correction of selection bias. Our dependent variables are the probabilities of Switching 1, Switching 2, and Switching 3, respectively. Our independent variables can be categorized into two groups: characteristics of the entrepreneurial firms and those of their existing lead VCs. The former group consists of a measure of firm quality, dummies indicating firm stage, firm location, and whether the firm is a technology venture. To measure the entrepreneurial firm's quality at the start of a new round, we run a probit regression in which the dependent variable is equal to one if the firm exited as an IPO or a M&A by the end of 2007, and zero otherwise. 5 The independent variables include various characteristics of entrepreneurial firms (stage, industry, location, round number), and the lead VC from the previous round (fund size, experience, distance, number of investors). The predicted probability of a successful exit then is estimated from the probit regression and used as a proxy for the firm s quality. We estimate the probability of a successful 5 Our findings are robust to just considering IPO as a successful exit. Also, we checked for M&A exits that possibly disguised write-offs (liquidation sales) based on the exit valuations. 15

16 exit for each firm at the start of each round, assuming new information is available in the new round and thus changes the probability of exit. Our findings are robust to a wide range of controls variables in this regression to proxy entrepreneurial firm quality; alternative specifications are available on request. To examine how existing lead VC characteristics impact the switching decision, we include the following four measures in regard to an existing lead VCs: age, fund size, investment market share, and distance. The switching process can be initiated by VCs or entrepreneurs. For instance, reputable VCs (more senior VCs and VCs with larger market share) might refuse to keep investing in an entrepreneurial firm if they are not satisfied with the exit potential of the firm, forcing the entrepreneur to seek future financing from other VCs. Alternatively, entrepreneurs might want to try out the opportunity to be associated with more reputable VCs as they become more confident with the potential of the firm. These scenarios suggest that higherquality ventures associated with less reputable VCs and lower-quality ventures associated with more reputable ones are more likely to rematch themselves with appropriate counterparties. Thus, it is important that we control for the interaction effect between VC characteristics and firm quality. Specifically, we interact firm quality with four dummy variables, which are equal to one if the VC is in the top 25% of the four measures (age, market share, size, and distance (shortest)) among active VCs in a specific investment year, and zero otherwise. [Insert Table IV here.] The first three specifications in Table IV do not include interaction terms and the second three specifications do. In five out of the six specifications, Rho is statistically significant, indicating a selection bias. 16

17 As shown in Table IV, in all six specifications, firm quality is significantly and positively associated with the likelihood of Switching 1 and Switching 2, but is significantly and negatively associated with the likelihood of Switching 3. The impact of firm quality on switching is also economically significant. For instance, the probability of Switching 1 (Switching 2) is at least 11.0% (7.0%) higher for the top quartile quality firms than for the median quality firms; on the other hand, the probability of Switching 3 is at least 2.0% higher for the bottom quartile quality firms than for median quality firms. Furthermore, this finding suggests that higher quality companies are more likely to have a different lead VC in follow-on rounds, and that the existing lead VC or other VCs will continue investing in the company instead of exiting immediately. In contrast, all existing VCs stop participating in future rounds of the lower-quality switchers. Based on the above finding, we conjecture that both higher-quality companies and lowerquality companies are likely to switch VCs, but due to very different reasons. The coefficients of the measures of VC characteristics provide further insights. We find that companies previously financed by less reputable VCs (more junior VCs and/or VCs with smaller investment market share) have a higher likelihood of Switching 1 and Switching 2, but companies previously financed by more reputable VCs have a higher likelihood of Switching 3. Specifically, one standard deviation decline in VC firm age increases the probability of Switching 1 by 4.1% and the probability of Switching 2 by 3.3%, and decreases the probability of Switching 3 by 0.6%. One standard deviation increase in VC firm investment market share, on the other hand, decreases the probability of Switching 1 by 1.4% and increases the probability of Switching 3 by 0.6%. Figure 1 describes the probability of all three levels of switching associated with entrepreneurial firm quality as well as reputation of lead VCs (top quartile versus bottom quartile). 17

18 [Insert Figure 1 about here.] Furthermore, we show that companies associated with smaller VC funds and more distant VCs, in general, are more likely to switch VCs. Specifically, one standard deviation decline in fund size from the mean increases the probability of switching by %; and one standard deviation increase in distance from the mean increases the probability of switching by %. These findings suggest the capital needs of entrepreneurial firms and the benefits/costs associated with geographic distance are also important determinants in the decision of the rematching between VCs and entrepreneurs. As shown in the last three specifications in Table IV, the interaction terms between the dummies (top 25%) of VC characteristics and firm quality also have significant influence on the likelihood of switching. For instance, the coefficients of the interaction terms between VC reputation (age and market share) and firm quality are significant and negative, suggesting that the relationship between reputable VCs and lower-quality companies, as well as the relationship between less reputable VCs and higher-quality companies, are more likely to be broken. Presumably, the discontinuation of the former relationship is more likely to be initiated by VCs if they acquire negative new information about the true potential of the company, while the discontinuation of the latter is more likely to be initiated by entrepreneurs who will pursue affiliation with more reputable VCs when positive new information is disclosed. The coefficients on the interaction between distance and firm quality are positive and significant, suggesting that more reputable VCs and higher-quality ventures are more willing to be geographically close to each other. The coefficients on the interaction between VC fund size and firm quality show different signs conditional on the definition of switching. It is positively correlated with 18

19 Switching 1, negatively associated with Switching 2, and positively but not significantly correlated with Switching 3. In addition to firm quality and VC reputation, we find that venture stage, industry, and location also significantly impact the likelihood of switching. Specifically, technology ventures located in hot states such as California (CA) and Massachusetts (MA), have a lower likelihood of Switching 1 and Switching 2, but a higher probability of Switching 3. Specifically, the probability of Switching 1(Switching 2) is 2.7% (12.1%) lower for technology ventures, but the probability of Switching 3 is 4.4% higher for technology ventures. Ventures located in CA or MA are % less likely for Switching 1 and Switching 2, and % more likely for Switching 3. In addition, we show that expansion and later stage ventures have a higher probability of Switching 1 and Switching 3. For instance, expansion and later stage ventures are 11.9% and 12.0% more likely for Switching 1, respectively, and 3.3% 7.4% more likely for Switching 3, in comparison to early stage ventures. Note that we considered other variables in the regressions, but their inclusion/exclusion was immaterial to the results of interest that are presented in the tables. For example, we included variables for the VC s stage preference as reported to the data vendor. We also ran the regressions for subsets of the data for different VCs based on stage preference, location, size, reputation, portfolio composition, among other things (see, e.g., Knill (2009)). Overall, inferences drawn from results not explicitly reported were not materially different. These and other specifications are available on request. In summary, we show that lower-quality companies associated with more reputable VCs and higher-quality companies associated with less reputable VCs are more likely to break their existing relationship. This is consistent with all three hypotheses. It also suggests that there are 19

20 different motivations for switching for different groups of entrepreneurial firms and/or VCs. In the next section below we present some extra details pertaining to the direction of switching that complement the results and robustness checks as discussed and presented in Table IV. B. Existing VC Reputation, Entrepreneurial Firm Quality, and the Direction of Switching As shown in Table III, there exist a fairly large number of observations in which entrepreneurial firms switch from less reputable VCs to more reputable VCs and the other way around. In the analysis that follows, we explore the determinants of the direction of switching by running probit regressions using a sample of 4,609 cases in which the lead VC of the follow-on round is different from the previous round. We focus on the change in VC reputation (VC age and VC investment market share) after switching. If the new lead VC is more senior or with greater market share, the dependent variables are set to equal one, and zero otherwise. The independent variables are similar to the ones we use in Table IV. In addition, we include Switching 2 and Switching 3 as independent variables to control for whether existing VCs continue investing in subsequent rounds even though they are not the lead VC any more. The results are reported in Table V. [Insert Table V here.] We find companies are 7.7% more likely to switch to less reputable VCs if existing VCs do not participate in the follow-on round, which is consistent with the finding in Table IV that lower-quality companies are associated with a higher probability of Switching 3. Furthermore, companies associated with less reputable VCs (more junior VCs and VCs with smaller investment market share) are more likely to switch to better ones. That is, one standard deviation 20

21 increase in VC age from the mean decreases the probability of switching to more senior VCs by %; one standard deviation increase in VC investment market share from the mean decreases the probability of switching by %. Interestingly, we show that companies that are already associated with relatively better VCs (top 25%), are actually more likely to switch to even more reputable ones. Companies with average quality that are associated with the top 25% are % more likely to switch to even more reputable ones. Our findings suggest that the reputation of the existing lead VC and whether it participates in the subsequent rounds of financing have both statistically and economically significant impacts on the direction of switching. IV. Switching and Financing Deal Terms A. Investment Size and Valuation Companies care about the identity of the investor, and when faced with multiple offers, companies are often in favor of more reputable investors and turn down less reputable ones even when they offer the best financial terms (Hsu (2004)). In this section, we analyze whether entrepreneurs turn down better deal terms in order to switch to more reputable ones and whether they switch to less reputable ones to obtain better financial terms. Specifically, we compare the size of investment (millions of 2006 dollars), and the pre-money valuation (millions of 2006 dollars) between switchers and non-switchers. Our analysis in Section III suggests that switching is endogenous. In other words, the decision to switch is impacted by VC and company characteristics, which presumably also determine the size of investment and valuation. To 21

22 control for this self-selection issue, we use the classic Heckman two-stage regression framework. In the first stage, we run the probit regressions analyzing the decision to switch (as shown in the last three specifications in Table IV). Then Inverse Mills Ratios (IMRs) are estimated off the probit regressions and included in the second stage regressions as independent variables. The coefficients of the IMRs imply the impact of self-selection (switching) on subsequent round investment size and valuation. The second stage regression results are reported in Table VI. For each switching type (1-3), we analyze the investment size and pre-money valuation of the follow-on round. Both dependent variables are in natural logarithm format. The independent variables, which include IMR, that adjust for the self-selection of switching, are: size of the previous round; pre-money valuation of the previous round; company characteristics such as stage, industry, and location; characteristics of the new round lead VC including VC age, size, investment market share, distance; dummies indicating whether it is a follow-on fund; and whether the lead VC forms a syndicate. Finally, we include Ln(Flow) as a control for the hotness of the VC industry. [Insert Table VI here.] We find that switchers raise smaller rounds than non-switchers, controlling for the endogeneity of switching, if the existing VCs continue investing in the company even though they are not the lead VC any more (Switching 1 and Switching 2), but the coefficient is significant only for the case of Switching 2. Specifically, switchers (Switching 2) follow-on round is $0.7 million smaller than non-switchers, assuming other aspects are similar. On the other hand, we show that switchers raise significantly larger follow-on rounds than nonswitchers if none of the existing VCs continue investing in the company (Switching 3). Or, switchers are able to raise a follow-on round of $43.0 million more than non-switchers. 22

23 In Table IV, we show that Switching 1 is more likely to be related to higher-quality firms, and lower-quality ones are more likely to switch VCs when no existing VCs are interested in the subsequent round of investment. The impact of switching on subsequent round valuation is consistent with this empirical finding. As shown in Table VI, Switching 1 is significantly and positively associated with subsequent round valuation, but Switching 2 does not have a significant impact on valuation. On the other hand, Switching 3 is significantly and negatively associated with lower valuation. To put the differences in dollars, Switching 1 increases subsequent round valuation by $1.7 million, but Switching 3 is associated with a decline of $0.6 million in valuation. To further examine whether entrepreneurs are in favor of more reputable VCs and turn down better financial terms from less reputable ones (Hsu (2004)), we next explore the impact of switching direction on round size and valuation. [Insert Table VII here.] As shown in Table VII, we show that when the new lead VC is more senior than the previous lead VC, switching does not have a significant impact on either round size or valuation. On the contrary, when the new lead VC is of greater market share than the previous one, we show that companies raise significantly smaller follow-on rounds. Specifically, firms that switch to more reputable VCs have a follow-on round $0.2 million smaller than that of those who switch to less reputable ones. Switching to more reputable VCs also has a negative impact on valuation, although only a marginally significant one. For instance, firms that switched to VCs with larger market share have a valuation $0.1 million lower than their counterparties. These results provide some evidence that there is often a trade-off between VC reputation and the financial terms VCs offer. When entrepreneurs switch to more reputable VCs (for instance, VCs 23

24 with greater market share), they accept less favorable financial terms (e.g., smaller capital infusion and lower valuation). This finding is consistent with Hsu (2004). Lastly, we note that there are additional terms in VC contracts that might be affected by switching, just as there are terms which might affect the ability to switch. These specific contractual rights include, but are not limited to, antidilution, drag-along and redemption rights. Publicly available datasets such as that which is used in this paper do not include such contractual terms. Further research could investigate the interplay between contractual terms and switching with more detailed hand-collected data. B. Duration between Rounds Another cost associated with switching is the searching cost. Switchers have to shop around to find a new lead VC and the new lead VC needs to conduct its own due diligence. It potentially takes more time for the company to obtain the follow-on financing in comparison to directly getting the funding from its existing VC. In this section, we examine the significance of the searching cost, which is defined as the duration between two rounds, when switching takes place. Furthermore, we examine whether switching to more reputable VCs incurs higher searching costs. Similar to our analysis of investment size and valuation, we control for the endogeneity of switching and the direction of switching using the Heckman two-stage framework. The results of the second stage regressions are reported in Table VIII. [Insert Table VIII.] We find that switching significantly increases the time needed to finish the follow-on financing in all three scenarios. Specifically, Switching 1 on average increases the duration by 24

25 1.2 months, Switching 2 on average increases the duration by 1.8 months, and Switching 3 on average increases the duration by 3.0 months. On the other hand, when companies switch to more reputable VCs (senior ones and VCs with larger market share), it generally takes the company less time to finish the follow-on round. Specifically, firms get their follow-on round financing months faster than non-switchers. It might be that more reputable VCs are more efficient at due diligence, which helps speed up the financing process. Alternatively, because only higher-quality companies are able to switch to more reputable VCs, they might have more than one competitive offer on the table, which might also speed up the closing of the transaction. Our analysis in this section shows that depending on the nature of switching, it has a significant impact on follow-on round investment size, valuation, and the duration between rounds. Overall, Switching 1 is associated with smaller investment size, higher valuation, and longer duration. Switching 2 is associated with smaller investment size and longer duration, but does not have a significant impact on valuation. Switching 3 is associated with larger investment size, lower valuation, and longer duration. Furthermore, in this section, we also show there is a tradeoff between VC reputation and financial terms. When firms switch to more reputable VCs, they get a smaller capital infusion and a lower valuation. Nevertheless, switching to more reputable VCs actually speeds up the financing process. 25

26 V. Conclusion and Discussion We study the dynamic of the relationship between entrepreneurs and their lead VCs from their very first round of financing until the entrepreneurial firm exits. We show that during the period from 1991 to 2002, 42% of the follow-on rounds of financing have lead VCs different from those of previous rounds. This finding raises several questions. For instance, why do entrepreneurial firms want to switch VCs; to be associated with more reputable VCs, or to get more favorable financial terms, or simply because existing VCs decided to abandon the investment? As far as we know, this is the first study addressing the above questions. Our empirical findings show that higher-quality entrepreneurial firms are more likely to switch lead VCs and typically at least one of the existing VCs continues participating in the follow-on rounds. On the other hand, lower-quality companies are more likely to have a different lead VC while none of the existing VCs invest in the follow-on rounds. Furthermore, we show that higher-quality (lower-quality) entrepreneurial firms that are associated with less (more) reputable VCs are less likely to maintain their current relationship. In addition, we document that higher-quality switchers in general obtain a smaller investment but a higher valuation in followon rounds than non-switchers, but they accept a lower valuation if they switch to more reputable VCs. In contrast, lower-quality switchers obtain a larger investment but a lower valuation than their non-switching peers. Furthermore, it typically takes more time for switchers to obtain the subsequent round of financing. This is particularly true for lower-quality switchers. Our findings provide some new insights on the relationship between entrepreneurial firms and VCs. Sørensen (2007) suggests that there exists a positive sorting in the venture capital market, in that higher-quality companies are associated with more reputable VCs. Assuming both VCs and entrepreneurs learn about the quality of the firm as new information is disclosed, this 26

27 learning process for both sides determines that the matching between VCs and entrepreneurs has to be dynamic. For instance, if reputable VCs acquire negative new information regarding the potential of the company, they will stop investing in the company. Alternatively, while entrepreneurs learn positive new information regarding the quality of the company, they might pursue more reputable VCs. We find supporting evidence for both motivations. Our findings also have implications for the relative importance of VC reputation and financial terms during the matching between VCs and entrepreneurial firms. Hsu (2004) suggests that entrepreneurs value the reputation of VCs and are often willing to accept less favorable financial terms in order to be associated with the more reputable VCs. We confirm that VC reputation is an important factor that determines the dynamic matching between entrepreneurs and VCs by showing that higher-quality entrepreneurial firms try to switch to more reputable VCs if they were affiliated with less reputable VCs in early rounds: switching entrepreneurial firms obtain a smaller capital infusion and often accept a lower pre-money valuation. 27

28 References Berger, A.N., and G.F. Udell, 1995, Relationship lending and lines of credit in small firms finance. Journal of Business 68, Boot, A. W. A., and A.V. Thakor, 1994, Moral hazard and secured lending in an infinitely repeated credit market game, International Economic Review 35, Chemmanur, T. J., K. Krishnan, and D. Nandy, 2008, How does venture capital financing improve efficiency in private firms? A look beneath the surface, Working Paper, Boston College. Chemmanur, T., and P. Fulghieri, 1994, A theory of the going public decision, Review of Financial Studies 12, Degryse, H., and P. V. Cayseele, 2000, Relationship lending within a bank-based system: Evidence from European small business data, Journal of Financial Intermediation 9, Diamond, D., 1991, Monitoring and reputation, the choice between bank loans and directly placed debt, Journal of Political Economy 99, Fernando, C.S., V.A. Gatchev, and P.A. Spindt, 2005, Wanna dance? How firms and underwriters choose each other, Journal of Finance 60, Gompers, P.A., 1996, Grandstanding in the venture capital industry, Journal of Financial Economics 43,

29 Greenbaum, S., G. Kanatas, and I. Venezia, 1989, Equilibrium loan pricing under the bank-client relationship, Journal of Banking and Finance 13, Hale, G., and J. A. C. Santos, 2009, Do banks price their information monopoly? Journal of Financial Economics, forthcoming. Hochberg, Y., Ljungqvist, A., Lu, Y., 2007, Venture capital networks and investment performance. Journal of Finance 62, Hsu, D. H., 2004, What do entrepreneurs pay for venture capital affiliation?, Journal of Finance 59, Ivanov, V., C. N. V. Krishnan, R. W. Masulis, and A. K. Singh, 2008, Does venture capital reputation matter? Evidence from successful IPOs, Working Paper, University of Kansas. Krigman, L., W. H. Shaw, and K. L. Womack, 2001, Why do firms switch underwriters?, Journal of Financial Economics 60, Ljungqvist, A., F. Marston, and W. J. Wilhelm, 2006, Competing for securities underwriting mandates: banking relationships and analyst recommendations, Journal of Finance, Lerner, J., 1994, Venture capitalists and the decision to go public, Journal of Financial Economics 35, Metrick, A., 2006, Venture capital and the finance of innovation. Wiley. 29