Do Entrepreneurs Make Good VCs? * Ye Cai Leavey School of Business Santa Clara University [email protected](408) 554-5157 Merih Sevilir Kelley School of Business Indiana University [email protected](812) 855-2698 Xuan Tian Kelley School of Business Indiana University [email protected](812) 855-3420 Current draft: May 2013 Abstract Using hand-collected data on the backgrounds of venture capitalists (VCs), we show that in a typical venture capital firm in our sample, 14% of VCs have been entrepreneurs before becoming a VC, referred to as entrepreneur VCs (EVCs). We use the local supply of EVCs as an instrument to establish causality and find that EVCs have a positive impact on venture capital firm performance. In addition, the positive relation between EVCs and firm performance is stronger for smaller, younger venture capital firms, and for firms specializing in early-stage investments and in high-tech industries. Contrary to VCs’ prior experience in entrepreneurship, neither prior experience in Wall Street nor in Main Street is significantly related to firm performance. We also find that EVCs have better individual performance. Overall, our results are consistent with the idea that EVCs have a better understanding of the business of starting and developing a new firm due to their first-hand entrepreneurship experience, which helps enhance firm performance. * We thank Shai Bernstein, Utpal Bhattacharya, Joan Farre-Mensa, Richard Townsend, and participants at the 2013 AFA meetings, the 2012 Kauffman-SFS Entrepreneurial Finance and Innovation conference. We thank Michael Flores, Yifei Mao, Aseem Puri, and Zhong Zhang for their excellent research assistance. We remain responsible for all errors and omissions.
45
Embed
Do Entrepreneurs Make Good VCs? - Queen's Universityweb.business.queensu.ca/faculty/jdebettignies/docs/... · an advanced understanding of the importance of VCs in promoting new firm
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
“Anyone who is not a former entrepreneur should not be in VC (industry).”
Eric Benhamou, a co-founder of Bridge Communications,
ex-CEO of 3Com, and founder of Benhamou Global Ventures
1. Introduction
Venture capitalists (VCs) play an important role in the creation and growth of new firms.
They provide not only monetary capital to meet the financing needs of entrepreneurs but also
human capital in the forms of advising, monitoring, and business networking. Although we have
an advanced understanding of the importance of VCs in promoting new firm growth and success,
our understanding of what makes a good VC is still limited. In this paper, we study VCs who
have been entrepreneurs before becoming a VC, referred to as entrepreneur VCs (EVCs).
It is intuitive to expect that EVCs might have a better understanding of the process of
starting and managing a new business due to their first-hand experience of founding a new firm.
Since they were entrepreneurs before, they might have a greater ability in working and
communicating with entrepreneurs more efficiently, relative to VCs without any prior experience
in entrepreneurship. Consistent with these views, anecdotal evidence suggests that EVCs have
become more common in recent years. For example, based on a recent article from the New York
Times, insiders in the venture capital industry believe that “venture capital is once again
attracting the right mix of former founders and operators who are truly passionate about
nurturing companies and who have hard work-won insights that can help founders succeed.”1
Using hand-collected data on the professional backgrounds of VCs, we provide evidence
that in a typical top performing venture capital firm in our sample, 14% of VCs have been former
1 Source: “Do former entrepreneurs make better venture capitalists” – NYTimes.com, November 16, 2011.
1
entrepreneurs in that they founded at least one start-up company before becoming a VC.2 We
also show a positive relation between the fraction of EVCs in a venture capital firm and the
venture capital firm’s performance. Specifically, we find that the fraction of EVCs in a venture
capital firm is positively related to the venture capital firm’s market share in the IPO market.
Similarly, we document a positive relation between the fraction of EVCs in a venture capital firm
and the fraction of portfolio firms the venture capital firm is able to take public, an important
performance metric in the venture capital industry.
One challenge in interpreting our baseline findings is that the positive relation between
EVCs and venture capital firm performance could be driven by certain unobservable venture
capital firm or individual VC characteristics, and hence, may not necessarily reflect a causal
effect of EVCs on firm performance. We address this issue by following Bottazzi, Da Rin, and
Hellmann (2008) and constructing an instrument based on the local supply of EVCs.
Specifically, by employing a two-stage least square (2SLS) analysis, we find that local
availability of EVCs is a strong predictor of a venture capital firm’s fraction of EVCs in the first
stage, and in the second stage, we continue to observe a positive and significant effect of EVCs
on firm performance. Our 2SLS analysis helps alleviate the endogeneity concerns and suggests a
positive effect of EVCs on the performance of venture capital firms.
One potential limitation of our instrument is that EVCs and successful venture capital
firms are likely to cluster within the same geographical areas because there are benefits from
doing so as established by existing work in agglomeration economics. To address this concern,
we examine the relation between EVCs and venture capital firm performance based on whether a
venture capital firm is located within an area with high entrepreneurship activity. If the positive
2 Our study focuses on the top 300 venture capital firms recorded in the Thomson Venture Economics database
using the 2010 venture capital firm reputation scores developed in Nahata (2008).
2
relation between the supply of EVCs and firm performance obtains as a result of clustering of
EVCs and high quality venture capital firms within the same geographical areas with high
entrepreneurship activity, we would expect the positive relation between EVCs and firm
performance to be stronger for venture capital firms located in areas with higher
entrepreneurship activity. We, however, show that the positive relation between EVCs and firm
performance is more pronounced for venture capital firms located in areas with lower
entrepreneurial activity. This result helps strengthen the validity of our instrument.
To help establish causality further, we explore the cross-sectional differences in the effect
of EVCs on firm performance. We find that the positive relation between the fraction of EVCs
and venture capital firm performance is stronger for smaller and younger venture capital firms.
To the extent that such firms have less experience in providing a more active and hands-on
investment guidance relative to larger and older firms, they are likely to benefit more from
EVCs’ prior experience in entrepreneurship. In addition, we find that the fraction of EVCs is
more strongly related to firm performance for venture capital firms specializing in early-stage
investments and in high-tech industries. This result is consistent with the idea that EVCs’
advantage in understanding and nurturing entrepreneurs might be greater for high-tech and early-
stage start-ups where VCs’ ability to communicate with entrepreneurs is more critical for
success.
The above cross-sectional tests lend credence to our inferences of the positive effect of
EVCs on venture capital firm performance. As discussed above, a major concern of our tests is
that omitted variables driving an endogenous matching between EVCs and venture capital firms
with good performance bias the results. However, it is difficult to conceive an omitted variable
that biases our results equally in venture capital firms that are large or small, that are young or
3
old, that specialize in early-stage or late-stage ventures, and that specialize in high-tech or low-
tech ventures. Our evidence on differential effects of EVCs on firm performance along these
dimensions alleviates the endogeneity concern to some extent and suggests that our results are
not entirely driven by an endogenous matching between EVCs and venture capital firms with
good performance. It appears that a treatment effect is at least partially in effect.
While exploring the relation between EVCs and venture capital firm performance, we
also study other important background information regarding individual VCs. Specifically, we
look at whether a given VC has an operational background in terms of having worked in a non-
finance industry (i.e., Main Street) before becoming a VC. Similarly, we check whether a given
VC has worked in the finance industry (i.e., Wall Street) before becoming a VC. We find that
previous experience in neither Wall Street nor Main Street is related to venture capital firm
performance while an entrepreneurship background remains positively and significantly related
to performance.
Finally, we find that former entrepreneurship experience of a VC is also positively related
to her own individual success, measured through a VC ranking survey conducted by the Forbes
magazine, suggesting that former entrepreneurs make successful VCs. In addition, a previous
background in neither Wall Street nor Main Street is related to a better individual VC
performance measured by the Forbes performance ranking. These findings then raise an
interesting question on why we observe Wall Street and Main Street VCs in venture capital
firms. To answer this question, we analyze firms employing all three types of VCs and find that
firms with a more diverse profile of VC human capital exhibit better performance. Hence, it
appears that backgrounds in entrepreneurship, Wall Street, and Main Street are complementary
4
to each other, and bringing together VCs with a diverse set of experience and skills within the
same venture capital firm is value-enhancing.
Our paper makes a new contribution to the literature on the importance of VC human
capital by documenting the existence of EVCs and by establishing a positive relation between
EVCs and venture capital firm performance. While it is well-known that the majority of angel
investors are retired entrepreneurs, to our best knowledge, our paper is the first studying VCs
with a previous entrepreneurship background. Our findings are complementary to the evidence
reported in Bottazzi, Da Rin, and Hellmann (2008) that VCs’ prior business experience is
positively related to their interactions with their portfolio firms, which, in turn, contributes to the
success of portfolio firms, based on a sample that includes the survey for 17 European countries.3
Our findings are also related to Zarutskie (2010) that studies the relation between professional
and educational backgrounds of VCs and venture capital firm performance.4 Our result on the
importance of VCs with an entrepreneurship background is particularly relevant within the
context of the findings in Ewens and Rhodes-Kropf (2013) that VC human capital at the partner
level is much more important than venture capital firm organizational capital in explaining firm
performance. Finally, our results are consistent with the view in Kaplan and Schoar (2005) that
that heterogeneity in VC skills is a potential explanation for performance persistence in the
3 We extend this line of inquiry one step further by using a more refined classification of prior experience of VCs
before becoming a VC. More specifically, we classify VCs into one of three categories, VCs with prior
entrepreneurship experience of starting a new business, VCs with prior experience in Wall Street, and VCs with
prior experience in Main Street, and show that it is the VC’s prior entrepreneurship experience which contributes to
firm performance. 4 Note that while Botazzi, Da Rin, and Hellmann (2008) include former experience in entrepreneurship in their
definition of prior business experience, they do not differentiate between business and entrepreneurship experience
explicitly. Similarly, although Zarutskie (2010) studies the effect of a VC’s previous background in an
entrepreneurial start-up company, she does not differentiate between VCs who worked as executives in a start-up
and VCs who started a new business themselves. In our analysis, a VC is defined as an EVC only if he has founded
a new firm before becoming a VC.
5
venture capital industry. Our paper proposes that prior experience in entrepreneurship represents
an important source of skill and ability in the venture capital industry.
Our paper also contributes to an extensive body of work in venture capital literature that
establishes VCs’ value-enhancing role in the growth and success of the start-up companies. VCs
provide help with the professionalization and enhancement of the management team (Hellmann
and Puri, 2002; Chemmanur et al., 2012), exercise intensive monitoring and corporate
governance (Barry et al., 1990; Lerner, 1995; Gompers, 1995; Kaplan and Stromberg, 2003;
Tian, 2011), motivate innovation (Hellmann and Puri, 2000; Kortum and Lerner, 2000; Tian and
Wang, 2012), and ease their portfolio firms’ access to public capital markets (Megginson and
Weiss, 1991; Lerner, 1994; Lee and Wahal, 2004; Nahata, 2008).
The rest of the paper is organized as follows. Section 2 describes our data. Section 3
analyzes the relation between EVCs and the performance of the venture capital firm where they
work. Section 4 explores whether an entrepreneurship background is related to a metric
measuring a VC’s individual success. Section 5 concludes.
2. Sample Selection and Summary Statistics
We construct our VC sample based on the following procedures. We start with the top
300 venture capital firms recorded in the Thomson Venture Economics database using Nahata
(2008) venture capital firm reputation scores as of year 2010. This venture firm reputation score
is based on the dollar market value of all companies a venture firm took public since 1980,
normalized by the aggregate market value of all venture capital-backed companies that went
public since 1980. Because the Venture Economics database provides an incomplete list of VCs
working at the venture capital firms, we search the websites of these 300 venture capital firms
and hand-collect the names and biographies of individual VCs if available. We exclude 110
6
venture capital firms from our sample since they do not provide any information about their VCs
on their websites.
We then match our sample of VCs with the BoardEx database of Management
Diagnostics Ltd. BoardEx contains extensive biographical information on corporate directors and
top executives including their educational background, employment history, and other
professional and social activities. We employ a three-round matching procedure based on VC
names as well as the names of the venture capital firms where they work. Specifically, the first
round starts with merging our VC sample with the BoardEx database using both the VCs’ first
and last names and the names of their venture capital firms. We are able to identify 306 VCs
from the BoardEx database in this round. In the second round, we look for unique matches of the
VCs’ first and last names in BoardEx. Since these names are uniquely identified in the BoardEx
database, mismatching concerns are minimized. However, we still perform a double check using
VC age to screen out mismatches. This second round of matching yields an additional 1,133
VCs. In the last round, we focus on the matches based on VC names that are not unique in the
BoardEx database (for example, there are four VCs whose name is David Cowan covered by
BoardEx, and one of them is a VC at Bessemer Venture Partners). We are able to match an
additional 125 VCs in the third round by screening out mismatches using both VC age and
educational background as additional criteria. Our final sample contains 1,564 individual VCs
working for 154 venture capital firms as of July 2010 with complete coverage of educational and
professional backgrounds of VCs.
To identify EVCs, we read each VC’s biography carefully and classify a VC as an EVC if
she has founded at least one start-up firm before becoming a VC. We also collect the industry
information and SIC codes of the firms where VCs worked before becoming a VC from
7
BoardEx. For privately held firms, we manually collect industry data from the Dun & Bradstreet
database. We then construct two other variables, WVC and MVC, which capture other
professional backgrounds for a VC. If a VC has worked at a financial firm (SIC code 6000-6999)
before joining a venture capital firm, we classify her as a Wall Street VC, denoted by WVC. If a
VC has worked at a non-financial firm before becoming a VC, we classify her as a Main Street
VC, denoted by MVC. A VC’s prior entrepreneurship experience, Wall Street experience, and
Main Street experience are not necessarily mutually exclusive.
We collect each VC’s personal data from BoardEx such as age, gender, educational
degree (MBA, JD, MD, or Ph.D.), and possession of professional certifications (CPA or CFA).
We construct each VC’s tenure in the current venture capital firm using the starting year
information provided in her biography. We further obtain venture capital firm characteristics
from the Venture Economics database. This database includes the number of start-up companies
the venture capital firm has invested in, the number of financing rounds it has participated in, and
the investment outcomes of its portfolio firms.
Table 1 presents the summary statistics for the venture capital firm characteristics in our
sample. In a typical venture capital firm, 13.9% of VCs are classified as EVCs. The fractions of
VCs with prior Wall Street and Main Street experience are 36.8% and 72.5%, respectively. The
average VC in our sample is 54.5 years old, and has worked in the current venture capital firm
for 12.5 years. Regarding VCs’ educational background, on average, 46% of VCs are MBA
degree holders, 16.3% have JD/MD/Ph.D. degrees, 38.5% graduated from elite schools, and 6%
of them have a professional certification such as a CPA or a CFA.5
5 Elite schools include Harvard University, Yale University, MIT, Stanford University, University of Pennsylvania,
Princeton University, and Dartmouth College. These are the most commonly attended schools by VCs in our
sample.
8
An average venture capital firm in our sample is 21 years old, has invested in 179 start-up
companies, and has participated in 475 financing rounds since 1980. 20.6% of their portfolio
companies eventually went public. Recall that these 154 firms are among the most successful
and reputable 300 venture capital firms. Therefore, it is not surprising that the venture capital
firms in our sample are more experienced and more successful than an average venture capital
firm recorded in the Venture Economics database.
3. EVCs and Venture Capital Firm Performance
In this section, we examine the relation between the fraction of EVCs in a venture capital
firm and the performance of the venture capital firm. In Section 3.1, we report our baseline
results on the relation between EVCs and venture capital firm performance. Section 3.2
addresses the identification concerns using an instrumental variable approach. In Section 3.3, we
examine the robustness of our baseline results using a recent sample of venture capital deals.
Section 3.4 further investigates the effect of EVCs on venture capital firm performance across
various types of venture capital firms. In Section 3.5, besides the entrepreneurship background,
we include VCs’ Wall Street and Main Street backgrounds and examine the relative importance
of different backgrounds on venture capital firm performance.
3.1. EVCs and venture capital firm performance
To study the relation between EVCs and the performance of venture capital firm where
they work, we estimate the following empirical model using ordinary least square (OLS)
regression:
Performance = α + β × Fraction of EVCs + δ'Z + Industry + u (1)
9
where the observational unit is a venture capital firm. Our first measure of Performance is the
natural logarithm of a venture capital firm’s IPO market share, Ln(IPO market share), a widely
used performance and reputation measure for venture capital firms introduced by Nahata (2008).
We also use the proportion of a venture capital firm’s portfolio companies that eventually go
public, IPO exit, as an alternative measure of venture capital firm performance.6 The main
variable of interest in this analysis is Fraction of EVCs, which equals the fraction of VCs with
prior entrepreneurship experience within a venture capital firm.7 Z is a vector of venture capital
firm and individual VC characteristics that could contribute to the performance of a venture
capital firm. It includes the venture capital firm’s investment experience (Ln(no. of firms invested
in the past), Ln(no. of rounds invested in the past), or Ln(VC firm age)), a measure of industry
concentration based on the industry composition of the venture capital firm’s investments
(Industry concentration) and the venture capital firm’s focus on early stage investments (%
investment in early ventures). In terms of individual VC characteristics, we include the average
age of VCs in a firm (Ln(average VC age)), the average tenure in the firm (Ln(average VC
tenure)), educational background (Fraction of MBA VCs, Fraction of JD/MD/PHD VCs,
Fraction of VCs from elite schools), and possession of professional certification (Fraction of VCs
with CFA/CPA). Industry, defined as the industry in which the venture capital firm is most
active, accounts for unobservable variations within the 18-industry classifications of Venture
6 Although both IPOs and acquisitions have been considered as successful exit by previous studies (e.g., Gompers
and Lerner, 2000; Brander, Amit, and Antweiler, 2002; Sørensen, 2007; Bottazzi, Da Rin, and Hellmann, 2008;
Chemmanur, Krishnan, and Nandy, 2011), existing literature suggests that going public is a more desirable exit
pathway than acquisitions for both entrepreneurs and venture capital investors. For example, Brau, Francis, and
Kohers (2003) show that IPO firms enjoy a 22% “valuation premium” relative to firms that are acquired, and
Sahlman (1990) argues that almost all of the returns for venture capital investors are earned on their eventually
going public portfolio firms. Bayar and Chemmanur (2011) suggest that only the best-quality VC-backed firms can
access the public capital markets through an IPO. 7 For robustness, we replace Fraction of EVCs with an indicator that equal one if the venture capital firm has at least
one EVC and zero otherwise. The results are quantitatively and qualitatively similar.
10
Economics that may influence the venture capital firm’s performance.8 Standard errors are
heteroskedasticity-robust.
Note that we measure the fraction of EVCs in a venture capital firm as of 2010, and
examine its effects on venture capital firm’s performance based on its portfolio firms in the past.
This empirical design is mainly due to the nature of our data collection process, as we start with
top 300 venture capital firms as of year 2010, hand collect each VC’s biography, and code them
as EVCs based on the information provided in their biographies. Ideally, for each venture capital
firm, we would like to have a complete list of their VCs as well as each VC’s biography as of
1980 to construct our EVC variable. However, it is impossible for us to obtain such information
starting in 1980 as this information did not exist as of 2010. The fact that the average tenure of a
VC in his current firm is 12.5 years in our sample partially mitigates this concern.9 In addition, in
Section 3.3, we conduct a robustness analysis in which we examine the relation between the
fraction of EVCs in a venture capital firm in 2010 and the performance of the firm in 2011 and
2012.
Table 2 reports the regression results estimating equation (1). The dependent variable in
columns (1) – (3) is Ln(IPO market share). Because our three measures of venture capital firm
investment experience (the number of firms the firm has invested in, the number of rounds it has
participated in, and firm age) are highly correlated with each other, we include them one by one
in the first three models in Table 2. The coefficient estimates of Fraction of EVCs are positive
and significant in all three specifications, suggesting that the fraction of EVCs is positively
related to the venture capital firm’s IPO market share. Since IPO market share is in the logarithm
form, the coefficient estimate of Fraction of EVCs gives us the semi-elasticity of a firm’s IPO
8 If a venture capital firm invests in multiple industries, we choose the industry in which the firm invests the largest
amount of capital since 1980 for the industry fixed effect. 9 We find that the average tenure of an EVC is 11.8 years.
11
market share with respect to the fraction of EVCs in the firm. The magnitude of the Fraction of
EVCs coefficient estimate in column (1) suggests that increasing the fraction of EVCs from the
25th
percentile (0) to the 75th
percentile (0.2) of its distribution is associated with a 10.5% (=
0.523 * 0.2) increase in the venture capital firm’s IPO market share. This translates to a 0.15
percentage point increase in the firm’s IPO market share, given that the mean IPO market share
is 1.4 percentage points in our sample.
In columns (4) - (6), we replace the dependent variable with IPO exit. The coefficient
estimates of Fraction of EVCs are positive and significant at the 5% level in all columns,
suggesting that a larger fraction of EVCs in a venture capital firm is positively related to the IPO
exit rate of portfolio firms backed by the venture capital firm. For example, the coefficient
estimate of Fraction of EVCs in column (4) suggests that an increase in the fraction of EVCs
from the 25th
percentile to the 75th
percentile of its distribution is associated with a 1.3
percentage points (= 0.063 * 0.2) increase in the venture capital firm’s IPO exit rate.
It is worth pointing out that due to the small sample size of our analysis with 154
observations, the power of the empirical tests is typically very low. However, we are still able to
find both an economically meaningful and statistically significant positive relation between the
fraction of EVCs and the performance of the venture capital firm.
We also notice that the economical magnitudes and significance levels of the coefficient
estimates of Fraction of EVCs are similar across columns (1) - (3) as well as across columns (4) -
(6), suggesting that using any one of the three measures of venture capital firm investment
experience gives us consistent results. In later analyses, to save space, we present results using
the number of start-ups the venture capital firm has invested in as the main proxy for its
12
investment experience. Our results are both quantitatively and qualitatively similar when we use
the other two measures of firm investment experience.
3.2. Identification
In the previous section, we show that the fraction of EVCs in a venture capital firm is
positively and significantly related to the Firm performance. One concern with these baseline
results is that our findings may be driven by an endogenous matching between EVCs and venture
capital firms with better performance. More specifically, certain unobservable venture capital
firm or VC characteristics omitted from the baseline regression may be positively related to the
fraction of EVCs in a venture capital firm as well as the firm’s performance. For example, better
venture capital firms may be more likely to hire former entrepreneurs who have already
established that they are successful individuals as well as entrepreneurs. Therefore, our earlier
results may not necessarily reflect a causality flowing from EVCs to firm performance. To help
establish causality, we construct an instrumental variable (IV) and use a two-stage least square
(2SLS) approach.
Our instrument is based on the rationale that the fraction of EVCs in a given venture
capital firm should be high when the local supply of EVCs is high (see Bottazzi, Da Rin, and
Hellmann (2008) for a similar argument). Thus, we expect that, other things equal, a given
venture capital firm is more likely to hire EVCs and therefore have a higher fraction of such VCs
when local EVCs are in greater supply. Following this intuition, we construct the instrument,
Local Fraction of EVCs, as the number of EVCs divided by the total number of VCs in the same
MSA area as the venture capital firm. The IV reflects the local supply of EVCs and thus should
affect the fraction of EVCs in a venture capital firm located in the same area.
13
Panel A of Table 3 presents the results from the first-stage regressions with Fraction of
EVCs as the dependent variable. The main independent variable of interest is the constructed
instrument. All other control variables are the same as those in the baseline OLS regressions. The
coefficient estimates of the instrument are positive and significant at the 1% level. The result
suggests that venture capital firms indeed tend to hire a greater fraction of EVCs when the local
supply of such VCs is high. The t-statistic of the instrument is very high with a magnitude of 5.2.
Therefore, based on the rule-of-thumb diagnostics suggested by Staiger and Stock (1997), the
instrument is highly correlated with the endogenous right-hand-side variable in the second stage
and does not appear to suffer from the weak instrument problem. It is also interesting to note
that, among the other control variables, the percentage of early investments is positively related
to the fraction of EVCs, suggesting that venture capital firms specializing in early stage
investments are more likely to hire EVCs. This result is intuitive as the benefit of hiring a VC
with a background in entrepreneurship is probably greater for firms focusing on early stage
investments.
Table 3 Panel B reports the second-stage regression results, with venture capital firm
performance as the dependent variable and the fitted value of the fraction of EVCs as the main
independent variable. The coefficient estimates of the instrumented Fraction of EVCs variable
remain positive and significant. The economic magnitudes are even larger than those in the OLS
analysis: an increase in the fraction of EVCs from the 25th
percentile to the 75th
percentile of its
distribution is associated with a 0.31 percentage points increase in the venture capital firm’s IPO
market share, and a 3.4 percentage points increase in the venture capital firm’s IPO exit rate.
There is one important concern with our IV. Based on the agglomeration economies
literature, it is likely that venture capital firms, EVCs, and entrepreneurial start-ups cluster within
14
locations of greater entrepreneurial activity. If so, regions of high entrepreneurship activity
would have a greater proportion of venture capital firms as well as a greater number of EVCs. To
the extent that venture capital firms in such regions have a greater success potential, there would
be a positive relation between the local supply of EVCs and venture capital firm performance
due to the unobserved factor of agglomeration economies. To address this concern, we analyze
the relation between EVCs and firm performance based on whether a venture capital firm is
located within a location with high entrepreneurship activity. If the above argument was true, we
would expect the positive relation between EVCs and firm performance to be stronger for
venture capital firms located in areas with higher entrepreneurship activity.
Chen, Gompers, Kovner, and Lerner (2010) show geographic concentration by both
venture capital firms and venture capital financed entrepreneurial firms in three metropolitan
areas – San Francisco, Boston, and New York. The Venture Economics database provides
headquarters MSA data for venture capital firms. For missing ones, we manually search for their
headquarters MSA information. Consistent with Chen et al. (2010), about two-thirds (100 out of
154) of our sample venture capital firms are headquartered in one of these three metropolitan
areas.
We define a venture capital firm to be within-cluster if it is headquartered in one of the
three metropolitan areas mentioned above. For the rest of the sample, we define them as outside-
cluster venture capital firms. We separately estimate equation (1) for outside-cluster venture
capital firms and within-cluster venture capital firms, and report the regression results in Table 4.
Panel A reports the results from the OLS regressions. In columns (1) and (2) the dependent
variable is Ln(IPO market share), while in columns (3) and (4) the dependent variable is IPO
exit. We find a positive and significant effect of Fraction of EVCs on both performance measures
15
in the subsample of outside-cluster venture capital firms, while the coefficient estimates of
Fraction of EVCs in the subsample of within-cluster venture capital firms are not significantly
different from zero. We report the results from the 2SLS regressions (second stage) using the
local supply of EVCs as the IV in Panel B. The structure of Panel B is parallel to that of Panel A,
and we find similar results.
Our findings from Table 4 suggest that EVCs play a more important role in enhancing
firm performance for venture capital firms located in lower entrepreneurship activity areas.
These findings also help alleviate our concern of endogenous matching between EVCs and
venture capital firms. If the matching channel is in effect, we would expect to observe a stronger
relation between EVCs and firm performance for within-cluster venture capital firms, which is
the opposite of what we find.
Taken together, our 2SLS analysis suggests that, while the IV approach may not be able
to perfectly identify the direction of causality due to its own limitations, the observed positive
relation between EVCs and venture capital firm performance is unlikely to be entirely driven by
an endogenous matching between EVCs and venture capital firms with better performance.
Instead, there appears, at least partially, a positive treatment effect of EVCs on the performance
of venture capital firms.
3.3. Robustness Tests Using a Recent Sample of Venture Deals
As mentioned earlier, one concern for our analyses is that we measure the fraction of
EVCs in a venture capital firm as of 2010, and examine its effects on the venture capital firm’s
performance based on its portfolio firms in the past. To address this concern, in this section, we
examine the relation between the fraction of EVCs in a venture capital firm as of 2010 and the
firm’s performance in 2011 and 2012.
16
Table 5 presents the results that repeat our tests reported in Sections 3.1 and 3.2 with the
dependent variable replaced with the newly constructed venture capital firm performance
variables based on firms’ performance in 2011 and 2012. Panel A reports the OLS regression
results estimating equation (1). The coefficient estimates of Fraction of EVCs are positive and
significant for both dependent variables, consistent with our baseline results reported in Table 2.
In Panel B, we report the results from the 2SLS approach with the instrument, Local Fraction of
EVCs, as discussed in Section 3.2. Since the first-stage regressions are identical to those reported
in Table 3 Panel A, we only report the second-stage regressions with the fitted values of the
fraction of EVCs as the main independent variable in Table 5 Panel B. The coefficient estimates
of Fraction of EVCs are again positive and significant at the 5% level for both dependent
variables, consistent with the results reported in Table 3 Panel B.
In summary, we continue to observe the positive effect of EVCs on firm performance
using a sample of portfolio firms exited in 2011 and 2012. This finding together with the fact that
an average VC (EVC) has been with her current firm for the last 12.5 (11.8) years alleviates the
concern arising from using current information about VCs in explaining past firm performance.
3.4. EVCs across Various Venture Capital Firms
So far we provide evidence that the fraction of EVCs in a venture capital firm is
positively related to firm performance. One may expect that the magnitude of this effect could
differ across different types of venture capital firms. In this subsection, we examine the
performance implications of EVCs based on the experience of the venture capital firm, whether
it specializes in early stage investments, and whether the firm is in a high-tech industry. These
tests should shed further light on the possible causal effect of EVCs on venture capital firm
performance.
17
3.4.1. Experienced vs. Inexperienced Venture Capital Firms
Venture capital firms with less investment experience may have a poorer understanding
of the business of start-up investment, and they could benefit to a greater extent from the first-
hand entrepreneurship experience of EVCs. Therefore, we expect the positive relation between
EVCs and venture capital firm performance to be stronger for venture capital firms with less
investment experience.
We obtain the number of portfolio firms that the venture capital firm has invested in since
1980 from the Venture Economics database. We separate the sample into large venture capital
firms and small venture capital firms based on whether the number of portfolio firms that the
venture capital firm has invested in since 1980 is above or below the sample median. We
separately estimate equation (1) for small venture capital firms and large venture capital firms,
and report the regression results in Table 6. In columns (1) and (2), the dependent variable is
Ln(IPO market share). While the coefficient estimate of Fraction of EVCs is statistically
significant at the 5% level in the subsample of small venture capital firms, the coefficient
estimate of Fraction of EVCs in the subsample of large firms is positive but not statistically
significant. We find similar results in columns (3) and (4) in which IPO exit is the dependent
variable: the coefficient estimate of Fraction of EVCs is positive and significant in the subsample
of small venture capital firms, while insignificant in the subsample of large venture capital firms.
As an alternative measure of venture capital firm’s investment experience, we obtain the
age of venture capital firms from the Venture Economics database. A venture capital firm is
defined as a young (old) venture capital firm if its age is below (above) the sample median. Table
7 reports the estimated coefficients of equation (1) for young and old venture capital firms,
respectively. The structure of Table 7 is parallel to that of Table 6. We find a positive and
18
significant coefficient estimate of Fraction of EVCs in the subsample of young firms, while
positive but not statistically significant coefficient estimate of Fraction of EVCs in the subsample
of old firms. These findings in Table 6 and 7 are consistent with our conjecture that less
experienced (smaller or younger) firms are more likely to benefit from EVCs.
3.4.2. Early-stage vs. Late-stage Venture Capital Firms
EVCs could be more important for the survival of early-stage start-ups in terms of
reducing the information gap between VCs and entrepreneurs given that providing efficient
support and advice to a start-up firm matters most during the early stages of the start-up.
Therefore, we expect the positive effect of EVCs on venture capital firm performance to be
stronger for the venture capital firms investing in early-stage start-ups.
The Venture Economics database provides information about the development stage of a
start-up company when it receives the first-round venture capital financing. We define a start-up
company as an early-stage start-up if it is in either the “start-up/seed” or “early” stages when it
receives the first-round venture capital investment, and as a late-stage start-up if it is in
“expansion”, “later stage”, or “buyout/acquisition” stages when it gets the first-round venture
capital financing. We then calculate the proportion of early-stage start-ups in a venture capital
firm’s portfolio since 1980, and partition the sample into two subsamples, venture capital firms
investing in early-stage ventures and venture capital firms investing in late-stage ventures, based
on whether the proportion of early-stage start-ups a venture capital firm has invested in is higher
or lower than the sample median.
We report the results based on the development stages of start-ups in Table 8. The
coefficient estimate of the Fraction of EVCs is positive and significant at the 5% level in column
(1) but is insignificant in column (2), where the dependent variable is Ln(IPO market share).
19
Similarly, we observe a positive and significant effect of Fraction of EVCs on IPO exit in the
subsample of venture capital firms specializing in early-stage ventures, and a positive but
insignificant coefficient estimate in the subsample of venture capital firms specializing in late-
stage ventures. These results suggest that EVCs matter most for success for venture capital firms
specializing in early-stage start-ups.
3.4.3. High-tech vs. Non-high-tech Venture Capital Firms
If EVCs help improve performance, we may expect they have a greater ability to do so in
venture capital firms that invest more in high-tech industries. For high-tech start-ups, the
information gap between entrepreneurs and VCs is likely to be bigger, and EVCs could play a
more important role in reducing this gap and therefore enhance firm performance.
The Venture Economics database provides information about the industry classifications
of venture capital firms. Given the small-sample nature of our study, we use the most general 3-
industry classifications provided by the database: Medical/Health/Life Science, Information
Technology, and Non-high Technology. We group the first two industries and label them as
“High-tech industries”, and label the third industry group as “Non-high-tech industries”. If a
venture capital firm invests in multiple industries, we choose the industry in which the venture
capital firm invests the largest amount of capital and assign that industry to the venture capital
firm. We separate our sample of venture capital firms into firms specializing in high-tech
industries and firms specializing in non-high-tech industries, and estimate equation (1) for each
subsample separately.
Table 9 presents the results on the relation between EVCs and venture capital firm
performance for high-tech and non-high tech venture capital firms. The dependent variables are
Ln(IPO market share) in columns (1) and (2), and IPO exit in columns (3) and (4). We find that
20
the Fraction of EVCs has positive and significant coefficient estimates on both measures of firm
performance in the subsample of venture capital firms investing in high-tech industries. The
coefficient estimates of Fraction of EVCs in the subsample of venture capital firms investing in
non-high-tech industries are, however, not statistically significant.
The cross-sectional tests reported in this section provide further support to our inferences
of the positive effect of EVCs on venture capital firm performance as it is hard to come up with
an omitted variable that biases our results equally in all cross-sectional dimensions discussed
above. Overall, it appears that a treatment effect is at least partially in effect.
3.5. Entrepreneurship, Wall Street, and Main Street Experience
Our evidence so far has shown a positive relation between the fraction of EVCs and the
performance of the venture capital firm. In this section, we explore the relative importance of a
VC’s background in entrepreneurship, Main Street, and Wall Street (EVCs, MVCs, and WVCs,
respectively) on the performance of a venture capital firm. Specifically, we estimate the
following model:
Performance = α + β1 × Fraction of EVCs + β2 × Fraction of WVCs
+ β3 × Fraction of MVCs + δ'Z + Industry + u
(2)
where the observational unit is a venture capital firm. The dependent variable, Performance,
could be one of two variables used before, Ln(IPO market share) and IPO exit. The main
variables of interest are Fraction of EVCs, Fraction of WVCs, and Fraction of MVCs. Z is a
vector of controls defined in equation (1). We control for venture capital industry fixed effects.
Standard errors are heteroskedasticity-robust.
Table 10 reports the results estimating equation (2). The coefficient estimates of Fraction
of EVCs remain positive and significant in both columns. The magnitudes of Fraction of EVCs
21
are slightly larger than those in the baseline regressions reported in Table 2 but remain very
much comparable. We also notice that the coefficient estimates of Fraction of WVCs are positive
but statistically insignificant, suggesting that having VCs with a previous experience in Wall
Street is not significantly related to a venture capital firm’s performance. The coefficient
estimates of Fraction of MVCs are statistically insignificant, suggesting that VCs’ previous
background in Main Street also does not matter for firm performance. In summary, these results
show that neither prior Wall Street experience nor prior Main Street experience of VCs is
significantly related to venture capital firm performance while the fraction of EVCs in a venture
capital firm remains positively related to its performance.
If Wall Street and Main Street VCs are not positively related to performance, why do we
observe them in venture capital firms? Are there complementarities among VCs with different
backgrounds? To answer these questions, we examine whether venture capital firms with all
three types of VCs are associated with better performance. We define an indicator variable
Having each type of VCs as one if a venture capital firm has at least one EVC, one WVC, and
one MVC, and zero otherwise. Table 11 reports the OLS regression. Consistent with the
complementarity hypothesis, firms having all three types of VCs significantly outperform their
counterparts. Compared to firms without all three types of VCs, firms that have all three types of
VC human capital are associated with a 25.8% increase in the venture capital firm’s IPO market
share. This translates into a 0.36 percentage point increase in the firm’s IPO market share, given
that the mean IPO market share is 1.4 percentage points in our sample. Having all three types of
VC human capital also increases the venture capital firm’s IPO exit rate by 2.7 percentage
points, which is economically significant given that the mean IPO exit rate in our sample is
20.6%.
22
These results are interesting within the context of findings in Ewens and Rhodes-Kropf
(2013) that individual VC human capital is much more important than venture capital firm
organizational capital in explaining performance. This result raises an interesting question: what
makes VCs come together and work within the same organization, if organization capital has
little impact on performance? Our paper provides a potential explanation for this question that
bringing VCs with diverse human capital together within the same organization creates
complementarities among them, with a positive effect on firm performance.
4. EVCs and Individual VC Performance
In this section, we examine whether a prior background in entrepreneurship manifests
itself in a VC’s own personal performance and success. As argued before, VCs who are former
entrepreneurs are likely to have hands-on experience in founding and developing start-ups with a
better understanding about the nature of entrepreneurship. After they become VCs, the skills they
obtained from their prior entrepreneurship experience may help them better screen projects and
create value for entrepreneurial start-ups. In other words, their prior experience in founding a
new firm could give them a comparative advantage in becoming a more successful and
recognized individual VC.
As a measure of the performance of an individual VC, we use the Forbes Midas List
2011’s Top Tech Investors which lists the 100 most successful VCs in 2011.10
Specifically, we
estimate the following probit model to examine whether EVCs are more likely to be listed on the
Forbes performance ranking:
Top VC = α + β × EVC + δ'Z + Industry + u
(3)
10
Forbes Midas List 2011’s Top Tech Investors is available at http://www.forbes.com/lists/midas/2011/midas-list-
complete-list.html. See http://www.forbes.com/2011/04/05/midas-list-methodology.html for a detailed methodology