Venture Capital Investment Cycles: The Impact of Public Markets

Paul Gompers, Anna Kovner, Josh Lerner, and David Scharfstein*

December 1, 2005

It is well documented that the venture capital industry is highly volatile and that much of this volatility is associated with shifting valuations and activity in public equity markets. This paper examines how changes in public market signals affected venture capital investing between 1975 and 1998. We find that venture capitalists with the most industry experience increase their investments the most when public market signals become more favorable. Their reaction to an increase is greater than the reaction of venture capital organizations with relatively little industry experience and those with considerable experience but in other industries. The increase in investment rates does not affect the success of these transactions adversely to a significant extent. These findings are consistent with the view that venture capitalists rationally respond to attractive investment opportunities signaled by public market shifts.

*Harvard University. Gompers, Lerner, and Scharfstein are also affiliates of the National Bureau of Economic Research. We thank for their research assistance Vanessa Broussard, Daniel Goodman, Leif Holtzman, Alex Lee, Miriam Tawil, and Chenling Zhang. We thank Raffi Amit, Bob Gibbons, Ed Rock, and seminar participants at Harvard University, the National Bureau of Economic Research, the Stockholm Institute for Financial Research, the University of Chicago, and the University of Pennsylvania for helpful comments. Harvard Business School’s Division of Research and the National Science Foundation provided financial assistance. All errors and omissions are our own.

Venture Capital Investment Cycles: The Impact of Public Markets

December 1, 2005

It is well documented that the venture capital industry is highly volatile and that much of this volatility is associated with shifting valuations and activity in public equity markets. This paper examines how changes in public market signals affected venture capital investing between 1975 and 1998. We find that venture capitalists with the most industry experience increase their investments the most when public market signals become more favorable. Their reaction to an increase is greater than the reaction of venture capital organizations with relatively little industry experience and those with considerable experience but in other industries. The increase in investment rates does not affect the success of these transactions adversely to a significant extent. These findings are consistent with the view that venture capitalists rationally respond to attractive investment opportunities signaled by public market shifts.

1. Introduction

The high volatility of the venture capital industry is well documented. This

volatility manifests itself in a number of ways: the funds flowing to venture capital firms,

the investments firms make in portfolio companies, and the financial performance of

portfolio companies and venture capital firms (Gompers and Lerner, 2004). Much of this

volatility appears to be tied to valuations in public equity markets. An increase in IPO

valuations leads venture capital firms to raise more funds (Gompers and Lerner, 1998b;

Jeng and Wells, 2000), an effect that is particularly strong among younger venture capital

firms (Kaplan and Schoar, 2005). Moreover, returns of venture capital funds appear to be

highly correlated with the returns on the market as a whole (Cochrane, 2005; Kaplan and

Schoar, 2005; Ljundqvist and Richardson, 2003).

Many industry observers (see, for instance, Gupta, 2000) argue that the volatility

of the venture capital industry is a symptom of overreaction by venture capitalists and

entrepreneurs to perceived investment opportunities. These swings result in periods in

which too many competing companies are funded, followed by ones in which not enough

companies have access to capital. The boom of 1998-2000 provides an extreme

illustration of these problems. Funding during these years grew dramatically—in real

terms, the financing level in 2000 was more than 30 times the level in 1991—and was

concentrated in two areas: Internet and telecommunication investments, which accounted

for 39% and 17% of all venture disbursements in 1999. Considerable sums were devoted

to supporting very similar firms: e.g., nine dueling Internet pet food suppliers and close to

two dozen companies that undertook the extremely capital-intensive process of building

second cable networks in residential communities were funded. Meanwhile, many

2

apparently promising areas such as advanced materials, energy technologies, and micro

manufacturing languished unfunded as venture capitalists raced to focus on the most

visible and popular investment areas.

This alleged overreaction may have its roots in the behavioral biases of venture

capitalists who irrationally associate past investment successes with future investment

opportunities. Or it may stem from venture capitalists who feel compelled to follow the

herd out of concern for the reputation consequences of being contrarians (Scharfstein and

Stein, 1990). Indeed, in 1999, even private equity firms with investment mandates to

invest in leveraged buyouts felt compelled to back Internet startups.

A contrasting view is that the volatility of the venture capital industry stems not

from overreaction, but from the inherent volatility of fundamentals. In this view,

fluctuation in venture capital investment activity is simply a response to changes in

investment opportunities. There may be shocks to the investment opportunities of

existing entrepreneurial firms or entry by new entrepreneurs, both of which increase the

demand for capital.

This paper takes a step towards distinguishing between the “overreaction view”

and the “fundamentals view” by examining the responses of different classes of venture

investors. We start with the observation (and empirically document) that the most

experienced venture capital firms generally have the best performance (Sorensen, 2004).

We then examine how these most successful investors respond to public market signals

of investment opportunities. Are they more prone to increase their investments when the

market heats ups? And, how well do they do on these investments relative to less

experienced venture capitalists? If we find that the most experienced investors are more

3

prone to increase their investment levels when the market heats up, this suggests that

shifts in fundamentals are likely an important component of venture capital investing.

This interpretation is further supported if there is also little degradation in their

performance. On the other hand, if we observe that the least experienced venture

capitalists are most likely to increase their investment activity during hot markets, this

lends more credibility to the view that overreaction is a more important cause of volatility

in the venture capital industry.

Our empirical results indicate that investment by the most experienced venture

capital firms—notably those with the most industry experience—are most responsive to

public market signals of investment opportunities. We start by showing that venture

capital investment activity at the industry level is very sensitive to public market signals

of industry attractiveness such as Tobin’s Q and IPO activity; a shift from the bottom to

the top quartile in these measures increases the number of investments by more than

20%. This effect is driven largely by venture capital firms with the most experience

doing deals in the industry. Overall experience (across all industries) has no effect on

investment sensitivity to industry Q and IPO activity once we control for industry

experience. Moreover, although the success rate for deals done in a hot market is lower

than it is for deals done in a cold market, the difference is small. This difference between

hot and cold market performance is even smaller for experienced venture capital firms

than it is for less experienced venture capital firms. These findings suggest that an

important component of volatility in venture capital investment activity is driven by

volatility of fundamentals.

4

Of independent interest is our finding of the importance of industry-specific rather

than overall experience. It points to the importance of industry-specific human capital

and suggests that a critical part of venture capital investing is the network of industry

contacts to identify good investment opportunities as well as the know-how to manage

and add value to these investments. These contacts and know-how come only from long-

standing experience doing deals in an industry.

This broad question examined in this paper—the extent to which cycles in venture

capital are driven by overreaction to public market signals or changes in the industry’s

fundamentals—is related to a substantial stream of research in financial economics.

While the hypothesis remains controversial (e.g., Fama, 1998), a growing body of

evidence (e.g., DeBondt and Thaler, 1985, 1990) suggests that the stock market

overreacts to news, particularly at horizons greater than one year. (Hong and Stein (1999)

provides a theoretical framework for understanding these patterns.) More recently,

corporate investment has also been shown to be affected by the non-fundamental portion

of stock prices (Baker, Stein, and Wurgler, 2003). By way of contrast to much of this

literature, this analysis suggests that changing public market signals reflects changing

fundamentals.

This rest of the paper is organized as follows. The next section describes the

construction of the data and provides some basic summary statistics. Section 3 examines

the impact of shifts in valuations and IPO activity and the determinants of venture capital

organization investment activity. In that section, we also look at how investment success

depends upon both the investment cycle and the characteristics of the venture capital

organizations. Section 4 concludes the paper.

5

2. The Data

A. Constructing the Sample

Our data on venture investments come from Thomson Venture Economics

(Venture Economics). This database provides information about both venture capital

investors and the portfolio companies in which they invest. We consider an observation

to be the first record of a venture capital organization and portfolio company pair, i.e., the

first time a venture capitalist invests in a particular company. This approach results in a

dataset with multiple observations for most portfolio companies since several venture

capital firms typically invest in a company. We exclude follow-on investments by a

venture capital organization in the same portfolio company since these investments may

result from different considerations than do initial investments.

Our analysis focuses on data covering investments from 1975 to 1998, dropping

information prior to 1975 due to data quality concerns.1 In keeping with industry

estimates of a maturation period of three to five years for venture companies, we drop

information after 1998 so that the outcome data can be meaningfully interpreted. As a

result, we are not studying investments made at the height of the Internet boom (1999 and

2000) and the crash that followed. In addition, we limit our sample to venture firms that

have invested in more than five portfolio companies to insure that we are capturing true

investment firms and only include firms that invest in more than one year of the sample.

Our rationale is that we will find it difficult to characterize these firms’ investment

activities given their very limited track record.

1Gompers and Lerner (2004) discuss the coverage and selection issues in Venture Economics data prior to 1975.

6

From 1975 to 1998, Venture Economics provides information on 1,008 venture

capital firms investing in 12,582 companies that met our qualifications. This results in a

sample of 32,686 observations of unique venture capital firm – portfolio company pairs.

B. Critical Measures and Summary Statistics

Before we turn to an analysis of investment cycles, there are three data

construction issues we need to address.

The first issue is how to classify venture capital industries. Our approach is to

assign all investments into nine broad industry classes based on Venture Economics'

classification of the industry. The original sample of investments was classified into 69

separate industry segments. However, these 69 industries are too narrowly defined for

our purposes, as they do not correspond to lines of specialization within or across venture

capital firms. These 69 industries were thus combined to arrive at nine broader industries.

The industries we construct from the narrower definitions are: Internet and Computers,

Communications and Electronics, Business and Industrial, Consumer, Energy, Biotech

and Healthcare, Financial Services, Business Services, and all other. While any industry

classification is somewhat arbitrary, we believe that our classification scheme captures

businesses that have similarities in technology and management expertise that would

make specialization in such industries meaningful. In addition, this scheme minimizes

the subjectivity associated with classifying firms into narrower industry groupings.

Panel A of Table 1 shows the distribution across the nine broad industries. The

first column is the number of companies in each industry. It is no surprise that Internet

and Computers is the largest industry with 3,660 companies. Biotech and Healthcare,

7

Communications and Electronics, and Consumer are the next largest industries with

between 1,785 and 2,169 companies. The other industries are considerably smaller. The

table also reports the number of observations for each industry in our sample; there are

more observations than companies because there are multiple venture capital investors in

most of the companies in our sample. On average, there are 2.6 venture capital investors

in each company. The overall industry distribution provides some comfort that our

industry classification is meaningful. While there is variation in the number of

observations across industries, there are enough observations in each industry to make

our analysis feasible.

The second challenge has to do with the measurement of perceived investment

opportunities. We use two measures of perceived investment opportunities in our

analysis, industry Q and IPO activity. Because we do not know whether these measures

overstate or understate true investment opportunities, we will refer to industry Q and IPO

activity simply as “public market signals.”

The measurement of Q follows the standard approach in the investment literature.

We calculate Q as the ratio of market value of the firm to the book value of assets, where

the market value of the firm is measured as the book value of assets plus the market value

of equity less the book value of equity. Since we cannot observe the Q of private firms

that constitute the pool of potential venture capital investments, we use an estimate of Q

for public companies as a proxy. However, in order to do so, we need to link the SIC

codes of public companies to Venture Economics industries on which our data is based.

Our procedure is to identify the SIC codes of all Venture Economics firms that went

public. Because there are multiple SIC codes associated with each of our nine industries,

8

we construct Q as a weighted average of the industry Q of the public companies in those

SIC codes, where the weights are the relative fractions of firms that went public within

our nine industries. Within the SIC code, Q is calculated by equally weighting all public

companies.

Our second, less standard measure is the level of venture capital-backed IPO

activity in an industry. We use this measure for both theoretical and practical reasons.

The theoretical rationale is based on the observation that IPOs are by far the most

important (and profitable) means for venture capitalists to exit an investment (Gompers

and Lerner, 2004). Thus, an increase in the number of IPOs in a particular sector may

make investing in that sector more attractive. In addition, an increase in IPO activity may

also attract more potential entrepreneurs into a sector, thereby increasing the pool of

potential investments and the likelihood that a venture capitalist will find an attractive

one. The practical rationale for using IPO activity is that our Q measure may not

accurately reflect the shifts in public investors’ appetite for venture capital-backed firms

both because it uses data on mature public companies and relies on an inexact match

between SIC codes and Venture Economics codes. Given the strong link between IPO

activity and market valuations (Pagano, Panetta, and Zingales, 1998 and Ritter and

Welch, 2002), the IPO measure may actually be a better proxy for the public market’s

perception of the types of investments in our sample.

Figure 1 shows the relationship between industry venture capital investment

activity and the two measures of public market signals for four of the industries in the

sample. In Internet and Computers, the correlation between IPOs and investment activity

appears to be very high throughout the period. This high correlation can also be seen in

9

Q in Figure 2. In other industries, the relationship is less pronounced. For instance, in

both Biotechnology and Healthcare and Energy, the number of investments declined

during the last half of the 1990s, even as the number of IPOs in the industry climbed.

The final challenge is to measure the experience of the venture capital groups in

the sample. The second panel of Table 1 presents data on three characteristics of venture

capital firms that we use throughout the paper. The first such characteristic, “Overall

Experience,” is the total number of investments made by a venture capital firm prior to

the time of the investment in question. The second characteristic, “Industry Experience,”

is constructed similarly, but includes only investments in the same industry as the

investment in question. The third characteristic, “Specialization” is the fraction of all

previous investments that the venture capital organization made in a particular industry,

i.e., this specialization measure is the ratio of industry to overall experience.2 Throughout

the paper, we use all prior investments by the venture capital firm to compute these

measures, whether the investment is made by the current or a previous fund. (Venture

organizations typically raise new funds every three to five years.)

Panel B of Table 1 presents the distribution of overall experience, industry

experience, and specialization measures. We use as observations the annual activity of

each active venture organization in each industry where they could have potentially

invested. (We define a group as active in all years starting immediately following the

year of their first investment in the sample until the year of their last investment.) Since

many of these observations include cases where the venture capital firm did not invest in

an industry in a particular year, we report the sub-sample that includes only firms that

2We also exclude investments in firms classified in the category “All Other,” since these seem to include many firms where classification information was simply missing, and it is not necessarily meaningful to interpret public market signals in this category.

10

were investors in the industry in a given year. In addition, we provide summary data for

1985, 1990, and 1995.

Overall, venture capital firms made an average of 36.3 previous investments, of

which 4.1 were in the same industry. The numbers are higher if one conditions the

observation on the venture capital firm making an investment in the industry during the

year. The medians of these experience measures are considerably lower, reflecting the

skewness of the distribution. Not surprisingly, there is an increase in experience over

time. On average, investments are made by venture capital firms with 19.75% of their

investments in the industry of the company in which they are investing. This suggests

that most venture capital firms spread out their investments across industries.

One natural concern with the experience measure is that it will grow mechanically

over time: because the venture industry in 1998 is much more mature than it was in 1975,

there will be many more organizations with extensive experience. We also present in the

table several measures of experience that adjust for the changing maturity of the industry.

The first of these, adjusted experience, is the number of investments made by a given

venture capital organization prior to that investment in year t less the average experience

of all other venture firms active in year t prior to year t. The second, adjusted log

experience, is the difference between the logarithm of the number of investments made

by a given venture capital organization prior to year t and logarithm of the average the

number of investments made by all other venture firms active in the year prior to year t.

(Adjusted industry and non-industry experience is in each case computed similarly, but

only using the average experience in that or other industries.) We present the adjusted

experience measure because it is easier to interpret; the measure of adjusted log

11

experience is the one that is used in the actual regressions. Once we make this

adjustment, there is no secular trend in the experience level.3

Table 2 breaks out venture capital firm characteristics by quartile, and examines

the relationships among them. Industry experience and specialization quartiles were

calculated separately for each industry and year, so that industries with fewer investments

would not be disproportionately sampled in lower quartiles and the highest experience

quartiles would not disproportionately reflect later investments. The first quartile

represents the least experienced or specialized firms, while the fourth quartile measures

the highest. (We use a similar approach throughout the paper.) Not surprisingly, venture

capital firms in the higher quartiles of industry experience have made more investments

overall than firms in lower quartiles of industry experience. In the interests of space, we

do not present similar tabulations for the adjusted measures, but they display a similar

pattern.

In Panel B of Table 2, we undertake a correlation analysis between the various

measures. We see high correlation between industry experience and overall experience,

whether we use the unadjusted or adjusted log measure. Specialization, on the other

hand, is not highly correlated with the experience measures; in fact, it is negatively

3When we adjust experience in a year, we are adjusting based on the mean experience of all firms in that year at the beginning of the year with one observation per firm. When we summarize firm experience using the first observation per year of that firm, the average adjusted experience is accordingly zero. However, when we summarize by each firm-industry-year observation, we are no longer necessarily pulling the first observation of the firm in that year. For example, a firm that invested in 5 deals prior to 1995 and in 1995 invested in one internet company followed by one biotech concern has an experience level of 5 for the internet deal and 6 for the biotech deal. The firm was denoted as having an experience level of 5 at the time of their first investment in the year, which we used to compute experience at the firm-year level. If the typical venture firm had done five deals prior to 1995, its adjusted experience in that year would be zero. But when the summary statistics are computed at the firm-industry-year level, which uses the first investment in each industry in each year, the adjusted experience will not average to zero. Rather, the average will be 0.5, since the adjusted experience level for that firm is different for the biotech observation (6-5=1) than the internet observation (5-5=0).

12

correlated with overall experience (unadjusted or adjusted). This negative correlation is

driven by the firms in the highest specialization quartile who make fewer investments

than those firms who specialize less. The pattern is probably due to the fact that extreme

specialization limits the pool of investments from which a venture capital firm can

choose.

Table 3 presents a summary of how three indicators of venture capital activity

vary with the measures of investment opportunities. For each industry, we divide yearly

observations into four quartiles based on the volume of IPOs and Tobin’s Q in the

previous year. (Hence, six observations of “Internet and Computers” will be assigned to

each of the four quartiles.4) There is no clear linear pattern between the number of

investments and the indicators of investment opportunity. As the proxies for investment

opportunities increase (i.e., as we move from the first to the fourth IPO quartile), the

round number of the investment increases: in markets with greater perceived

opportunities, later-stage investments become better represented. We also see that as

investment opportunities increase, we see greater representation of more experienced

venture organizations. We will defer discussing the interpretation of this result until the

regression analysis in Table 6.

4The number of observations is not the same in each quartile, because the number of venture groups changes over time: the years with many IPOs and high Q are concentrated at the end of the sample, when many more venture groups were active. The number of observations is smaller when we present the round number of the investment and experience because these observations represent a VC-company investment pair, rather than a VC-industry-year pair (which may be one in which the VC did not make any investment).

13

3. Analysis

A. The Determinants of Investments

We first focus on understanding how public market signals affect the investment

decisions of venture capitalists. In Section 3.B, we turn to understanding the

determinants of investment success.

Table 4 presents a regression-based analysis of the relationship between the

number of investments and our public market signals. The first column shows the results

of regressing the logarithm of the annual number of investments in an industry on the

lagged logarithm of the number of IPOs in the industry, including industry and year fixed

effects. The coefficient estimate implies that an increase in IPO activity from the bottom

to the top quartile increases the number of investments by 22%. Likewise, the second

column indicates that there is a strong positive relationship between industry investment

activity and Q. An increase from the bottom Q quartile to the top Q quartile increases

industry investment by 22%. The third and fourth columns of Table 4 report the results

of using detrended variables in the regression. For each industry, we detrend both

industry investments and the public market measures. We then use the residuals in the

regression. Again the magnitude of the effects is large and similar across regressions,

although the explanatory power of the IPO measure appears to be significantly greater

than that of Q. In the final two columns, we add an industry-specific AR(1) term to the

specifications. This approach is motivated by the concern that both the dependent and

independent variables may be serially correlated. Both coefficients continue to take on

the expected positive sign. While the coefficient for lagged IPOs is no longer statistically

14

significant, that for lagged Q is strongly significant.5 These regressions would appear to

validate the use of Q and IPO activity as measures of public market signals that affect

venture capital investments.

Table 5 begins to look at the relationship between venture capital firm

characteristics and investment behavior. As before, we use as observations each venture

capitalist-industry pair in each year the venture organization is active (i.e., all years

following the first observation of an organization and ceasing in the year in which the

organization's final investment is observed). We first present results using IPO activity

and then check for robustness using the Q measure. The results are essentially the same

using either measure.

The first column of Table 5 repeats the industry level regression at the venture

organization-industry level. We include both industry and year fixed effects. Not

surprisingly, the regression indicates that venture capital firms tend to increase their

investments in years and industries in which IPO activity increases. The coefficient of

0.039, which is statistically significant, implies that an increase in IPO activity from the

25th percentile to the 75th percentile boosts the venture organization’s investment activity

in the industry by 4.9%.

As the second column of Table 5 indicates, there is also a strong positive

relationship between overall experience and investment activity. (As noted above, in all

the regressions, we employ the adjusted log experience measures defined on page 10.

5We also repeat a variety of the regressions below using a similar specification in unreported analyses. In general, the results are quite similar. The key results from Table 5 (the positive effect of the public market proxies and the greater impact of industry than non-industry experience, though in this specification non-industry experience is also positive) and Table 6 (the interaction of industry experience and IPOs is positive and the interaction of non-industry experience and IPOs is negative) continue to hold.

15

This is also true for the measures of industry and non-industry experience.) The third

column decomposes experience into industry experience and non-industry experience.

The regression indicates that what drives the relationship is industry experience; prior

investment activity outside the industry has no appreciable relationship to investment

activity within the industry. The average venture capital firm in the highest quartile of

industry experience invests 24% more in the industry than a firm in the lowest quartile of

industry experience.

Columns 4 and 5 of Table 5 add industry specialization to the regressions. In both

regressions, it is clear that prior focus on a particular industry increases future investment

in the industry. The results in column 5 indicate that an organization in the top industry

specialization quartile makes 8% more investments in that the industry than one in the

bottom quartile. Finally, the last two columns of Table 5 replicate the results in columns

3 and 5 using Q rather than IPO activity as the measure of the public market. The basic

patterns continue to hold in these regressions, and the magnitude of the effects is similar.

The next two tables present our main results on how venture capital firms with

different characteristics respond to changes in public valuations and activity. In Table 6,

we add to the specifications in Table 5 variables that interact our public market measures

with our measures of firm characteristics, i.e., overall experience, industry experience,

and industry specialization. Throughout our discussion of the results, when we refer to

periods with high IPO activity we are referring to those in the top quartile of IPO activity;

low IPO activity refers to those periods in the bottom quartile. Likewise, high overall

experience, industry experience, and specialization refers to venture capital firms in the

16

top quartile, while those with low overall experience, industry experience, and

specialization refers to those in the bottom quartile.

The first column of Table 6 indicates that the industry investment activity of more

experienced venture capital firms is more sensitive to IPO activity than it is for less

experienced venture capital firms. This effect is statistically significant. It is also much

larger in magnitude than the effect for the average firm in the sample. At the mean of the

other variables, experienced venture capital organizations (again, where experience is

defined as a firm at the 75th percentile in adjusted log experience) invest 9.7% more when

IPO activity is high (at the 75th percentile) than when it is low (at the 25th percentile). By

contrast, relatively inexperienced venture capital firms (those at the 25th percentile of in

adjusted log overall experience) actually invest very slightly less at times when IPO

activity is high rather than low. The results also indicate that industry experience

increases the level of investment, not just the sensitivity of investment to IPO activity.

In columns 2 through 4, we repeat the analysis decomposing overall experience

into industry experience and experience in other industries instead of the overall

experience measure. More industry-experienced venture capital firms invest 8.7% more

when IPO activity is high (column 2). Column 3 reports that for firms with little industry

experience, the differential is only 1.2%: non-industry experience also displays the same

pattern, but the effect is weaker. When both experience measures are included in the

same regression (column 4), only industry experience retains its positive effect. In fact,

the non-industry experience interaction with industry IPO activity is negative in this

regression. When IPO activity is high, industry-experienced venture capital firms with

relatively low experience out of the industry invest 6% more than when it is low, while

17

venture capital firms with high experience out of the industry, but low experience within

the industry invest 6% less when IPO activity is high.

The fifth and sixth columns of Table 5 look at the effect of industry specialization

on investment behavior. Consistent with our findings on industry experience, we find

that more specialized venture capital firms tend to increase their industry investments by

more than less specialized firms when IPO activity increases. The effect, however, is

small in column 5, implying an increase in investment by 5.8% for specialized firms and

4.0% for less specialized firms. When overall experience and specialization are used in

column 6, both interactions terms continue to be economically and statistically

significant.

Finally, the last two columns in Table 5 report the results using Q as an alternative

public market measure. Those columns replicate the basic findings in columns 4 and 6 of

the table. The magnitude of the effects is similar to those estimated using IPO activity.

A natural question with the results in this table is why less experienced venture

capitalists invest less in a sector when investment opportunities appear to be more

attractive. We believe that this pattern reflects a crowding out effect: the less experienced

venture capitalists may wish to invest in the sector as well, but cannot get a “seat at the

table” in the transactions being completed. While ultimately the supply of transactions in

a given sector may adjust of accommodate demand, in the short-run there may be intense

competition for transactions.6

In Table 7, we check whether our results are driven by venture capital firms that

choose not to invest in a given industry. Thus, we eliminate from the regressions all

6This hypothesis is consistent with evidence in the venture industry of the phenomenon of “money chasing deals,” that is, inflows of capital into venture funds driving valuations upwards (Gompers and Lerner, 2000).

18

observations in which the venture capital firm made no investments in the industry in a

given year. All of the findings in Table 6 continue to hold although the magnitude of the

effects is somewhat smaller. Thus, in part this phenomenon is driven by less experienced

firms “opting out” of markets with substantial investment opportunities, but even those

inexperienced groups that are still active reduce their level of activity.

Collectively, these results suggest that industry-specific human capital is an

important channel through which experience influences the reactions of venture capital

firms to shifts in public market signals. Contrary to popular wisdom, it does not appear

that the booms and busts are being driven by the investment behavior of inexperienced or

new venture capital firms. In fact, these results suggest that the cyclicality seen in the

venture capital industry is driven mostly by the more successful venture firms, that is,

those with the most experience. Section 3.B considers the question of whether the

sensitivity of more experienced firms to public market signals is a rational reaction to

fundamentals or an overreaction.

B. The Determinants of Investment Success

In this section we explore whether the greater responsiveness of more experienced

venture capital firms to public market signals is efficient. If these experienced firms are

able to ramp up the number of investments they make in response to public market

signals, but suffer a significant degradation of performance, the investment response to

public market signals may, in fact, be an overreaction. In addition to the practitioner

accounts alluded to above, there are at least two reasons to believe this might be the case.

First, Baker, Wurgler, and Stein (2003) show that industrial firms whose investment is

19

most sensitive to Q have the lowest subsequent stock returns following periods of heavy

investment. A similar effect might be observed among experienced venture capital firms

whose investment is most sensitive to Q and IPO activity. Second, at the same time that

venture capital firms are buying equity in portfolio companies, these companies are, of

course, issuing equity. We know from numerous studies, including Loughran and Ritter

(1995), that when firms issue equity, their subsequent stock returns are abnormally low.

To assess this question, we examine the performance of the companies in which

the venture capital firms invest. Ideally, one would have data on the actual returns on the

firm’s investment. Unfortunately, the best we can do is to determine whether the

investment resulted in what would appear to be a profitable exit for the venture capital

firm. This is most likely the case if the company went public, registered for an IPO (as of

the date we collected the data from Venture Economics), or was acquired or merged.

Venture Economics does not collect valuation information for all of the companies that

were merged or acquired and it is possible that these outcomes are not as lucrative as

those where the company exited with a public offering. However, investments in the

category we characterize as successes are likely to have generated higher returns that the

investments those that have not yet exited or have been characterized as bankrupt or

defunct. (We also repeat the analysis below eliminating acquisitions in order to avoid

these ambiguities.)

The final column of Table 2 provides some initial indications of the patterns of

success by venture capital firm characteristics. The tabulations suggest that investments

made by venture capital firms with more general—and especially more industry-

specific—experience are more successful. The patterns with specialization are non-

20

linear, but the least specialized organizations appear to be the poorest performers. One

consideration in the definition of specialization is that specialization may be related to

experience. Thus, our interpretation of these tabulations must be cautious, because of the

lack of controls for industry, time period, and (in the specialization analysis) experience.

Table 8 and 9 examine the determinants of success in a regression framework.

The dependent variable is a dummy variable, which takes on the value one if the

company was successful before the end of 2003. In addition to the industry and year

controls used earlier, we also control for the stage of the company and the financing

round at the time of the investment, since these are likely to be associated with the

outcome. As in our previous regressions, we exclude observations occurring after 1998

in order for the outcomes of the investments to be meaningful.

In the first column of Table 8, since no venture organization-specific independent

variables are used, each portfolio company is used as an observation. (In this case, the

round control refers to the first financing round where there was professional venture

financing.) In all other regressions in Tables 8 and 9, each initial investment by a venture

capital firm in a portfolio company is used as an observation. In these instances, standard

errors are clustered by portfolio company.

The first two columns of the table suggest there is no statistically significant

relationship between IPO activity and success in the sample as a whole. The third

column of Table 8 indicates that more experienced venture capital firms are more likely

to make successful investments. However, the fourth column shows that the effect of

experience is limited to venture capital firms with industry experience. Investments

made by venture capitalists with the most industry experience are 4% more likely to

21

succeed than those made by the least experienced venture capitalists. Given a baseline

success rate of 54%, this amounts to a significant increase in the probability of success.

The regressions with industry specialization in columns 6 and 7 support this basic finding

on the role of industry specialization. Columns 8 and 9 replicate the results using Q as

our measure of the public market signal and report results of similar statistical

significance and economic magnitude.

Table 8 makes it clear that venture capital firms with industry experience do not

perform worse on average, as a result of being more sensitive to shifts in public market

activities. Table 9 digs deeper by investigating whether experienced venture capital firms

perform worse on the investments they make when IPO activity and Q are high. The

results indicate that just the opposite is true. Overall, venture capital firms do somewhat

worse on the investments they make when there is a lot of IPO activity and Q is high,

although the estimated effect is statistically insignificant. However, the more

experienced venture capitalists exhibit less degradation in their performance than do the

less experienced venture capitalists. For instance, in column 3, for an industry

experienced venture capital firm (again defined as one at the 75th percentile) at the

median of non-industry experience, as we move from a market where there are few IPOs

to one where there are many offerings, the probability of success declines by only 0.1%.

For a venture firm inexperienced in the industry and at the median of non-industry

experience, the decline is 2.4%. Based on the results in Table 8 and Table 9, it would be

hard to argue that the greater responsiveness of experienced venture capital firms to IPO

activity and Q comes at the expense of performance.

22

In the final column of Table 9, we only look at a subset of the exits. In particular,

we drop observations where the investment was exited by merger or acquisition. As noted

above, acquisitions in some cases can generate very attractive returns to the venture

investors, while in other cases can only yield pennies for each dollar of invested capital.

When we eliminate this ambiguous class of exits, there continues to be no indication that

the greater responsiveness of experienced venture capitalists hurts performance.

C. Robustness Analyses

This section summarizes further analyses we undertook to determine whether our

basic findings are robust. We first discuss two tables with additional regressions, and then

turn to a variety of unreported regressions.

Using quartiles rather than continuous variables. Table 10 repeats the three

analyses, now measuring experience using quartiles rather than as a continuous variable.

In each case, we replicate an analysis reported earlier using industry and non-industry

experience: column 4 in Tables 6 and 7 and column 3 in Table 8. We find that the same

basic patterns hold: more experienced investors are more likely to invest when IPO

markets are strong. Firms in the fourth quartile of industry experience are more than four

times as responsive to IPO markets as are firms in the first quartile of industry experience

(Table 10, Column 1). There is a similar pattern when looking at the sample of only

investors. Similarly, there is no clear pattern between success and investment rates,

except for greater success for investments by the most experienced venture organizations

in the markets with the greatest investment opportunities.

23

Restricting the sample size. Table 11 looks at whether the results are robust when

the sample is limited. In particular, in the above analyses, we use as observations each

active venture group and industry. In some cases, a venture group may be very unlikely

to invest in an industry because they have no capabilities to assess such transactions. We

thus eliminate as observations any cases where the venture capital firm had not invested

in an industry previous to the year of the observation (reported in column 1) or

alternatively the venture capital firm never invested in the industry during the entire

sample period (reported in column 2). We estimate regressions equivalent to those in

column 4 in Table 6, but only report the interaction terms. The results are quite similar

and the magnitudes of the coefficients are slightly larger.

Addressing the possibility of capital constraints. Throughout the paper, we have

used the number of previous IPOs in an industry as a proxy for perceived investment

opportunities. This variable, however, might also be related to the availability of capital.

Typically, venture capitalists will distribute shares to their investors between one and two

years following an IPO (Gompers and Lerner, 1998a). If the investors seek to maintain a

constant allocation to venture capital, they may rapidly reinvest those funds. This pattern

may imply in periods following many venture capital-backed IPOs, venture capitalists

would have considerably more funds to invest. Thus, a relationship between lagged IPOs

and investments may appear even if lagged IPOs do not capture investment

opportunities.7

7Even if this story holds, it should be noted that it is by no means clear that the limited partners will invest in the same venture capital organizations. Nor is it certain that the venture funds will reinvest in the same sectors as where they recently took firms public.

24

To address this capital constraints story, we first examine the subset of venture

funds whose capital under management is above the median in the year of the

observation. Kaplan and Schoar (2005) have shown that there is a concave relationship

between investment success and future fundraising: more successful groups appear to

limit the amount of capital they raise, even though they could raise many times the

amount. So in these firms, capital constraints are unlikely to be binding. We also examine

those funds with an above-median success rate. Successful venture firms find it easier to

raise capital (Gompers and Lerner, 1998b; Kaplan and Schoar, 2005). Again, these

firms—even if they have relatively limited capital under management—have the potential

to raise large sums and should not be constrained. In columns 3 and 4 of Table 11, we

repeat the analysis in Table 6, limiting the sample to these two subsets of funds where

capital constraints are unlikely to be an issue. We find that results continue to hold as

before.

Alternative Proxies for Public Market Signals. Our analysis used Q and the IPO

activity of venture capital-backed firms as proxies for public market signals. We

expanded our IPO activity measure to include all IPOs, not just those that were venture

capital-backed. The two measures are highly correlated (0.81) since both measures

include venture capital-backed IPOs. Not surprisingly, the results were not appreciably

altered. We also considered several other market based measures, including the

earnings-to-price ratio, market-to-book ratio, and historical industry returns. When we

use all of these measures in unreported regressions, we obtain similar results to those

presented above.

25

Column 5 of Table 11 presents the results using one such alternative measure, the

success of IPOs in the past year in that industry. When IPOs are being well received, it

may be an indication of investor perception of opportunity in that sector. (See Hanley

(1993) for a discussion of why underwriters do not fully adjust the offering price for an

issue which encounters unexpectedly strong demand.) To measure the price change, we

compute the average of the percentage change between the offer price and the share price

in the first trading day (IPO "pop") for all venture-backed IPOs in that sector in the past

year from information in Security Data Company’s Corporate New Issues database and

the Center for Research in Security Prices pricing data.8 The results show that the

interaction terms are positive, and the coefficient on industry experience is larger than

that on non-industry experience. However, the results were less robust, becoming

insignificant when we used thirty-day returns as our measure of IPO market performance.

The IPO pop is perhaps less likely to serve as a signal of fundamental opportunities in a

sector, since may be measuring something fundamentally different, such as market

sentiment or underwriter market power.

Alternative Success Measures. Our primary outcome measure codes all mergers

and acquisitions as successes. To validate this choice, we further researched the 3,650

outcomes that Venture Economics recorded as mergers or acquisitions using the Factiva

database and the SDC mergers and acquisitions database, finding values for 1,263

companies. Of the 508 merged or acquired companies for which Venture Economics had

information on the total amount invested in the company and for which we found

valuation information, 431 companies (94%) had merger or acquisition values greater

8Based on discussions with practitioners, we expect that it is absolute rather than relative performance that matters the most in evaluating the success of IPOs. We ran similar analyses adjusting the returns for the market performance, measuring the market both with NASDAQ and the S&P 500, with similar results.

26

than the total amount invested in the company, with a median sale price of seven times

the amount of money invested. This supports our thesis that merged or acquired

companies are likely to have been high-return investments for venture capital firms.

However, one must be cautious in this interpretation since we were unable to find

information on the majority of the mergers and acquisitions, either because they were

purchased by other private entities or purchased by public companies in deals that were

not accompanied by a press release (perhaps because of their small size). Making the

highly conservative assumption that all companies whose value we could not determine

were not successful, we then redefined a successful investment as one in which the

company went public, or was in registration for a public offering, or was in a merger or

acquisition for which we were able to find a value. In unreported regressions, the results

were similar to those presented.

One Observation per Company. Since the dataset includes multiple observations

on the same portfolio companies, each outcome reflects not only a given venture capital

firm’s characteristics, but those of the other venture capitalists invested in the company.

As an additional robustness check to the relationship between experience, industry

experience, specialization, and success, we used a sample with one observation for each

portfolio company and the average levels of each variable of the venture capitalists

investing in the company. In these specifications, both industry and non-industry

experience are positively associated with success, as is specialization, although the

coefficient on specialization is not significantly different than zero. In the absence of

more information about the specific roles that each venture capital organization plays in

the selection and development of the company, it is difficult to draw any conclusions

27

from the interaction of the different venture capitalists which invested in the company.

This is a rich topic for future research.

4. Conclusions

The venture capital industry is a highly volatile one, as dramatic fluctuations in

fundraising and investment activity over the past few years demonstrate. These

fluctuations seem to be related to changes in the public market valuations and activity.

Practitioner accounts and the academic literature suggest that it would be valuable to

understand the impact of this volatility on the success of venture capital investments: do

public market shifts lead venture capitalists to make poor investment choices, or rather do

they provide valuable information to investors? We address this question by examining

the determinants and success of investments by the venture industry as a whole, as well

as by subclasses of firms with different levels of experience and specialization.

We analyze over forty thousand venture capital investment decisions over the past

two decades. We find that the greatest response to shifts in the public markets is not by

new or inexperienced groups, but rather by specialized organizations with considerable

industry experience. Not only do the investments of these organizations tend to be more

successful in general, but there is no appreciable degradation in their performance with

the changing conditions.

Our results suggest that shifts in public markets provide information, whether

directly to the venture investors or else to individuals who then seek venture financing.

Not all venture groups, however, are able to take advantage of this information: the

28

critical factor appears to be human capital.9 As noted in the introduction, the fundamental

pattern seen here—that the changing public market signals seems to reflect shifting

fundamentals—runs counter to much of the recent work in financial economics on

investor overreaction.

The greater investment sensitivity is associated with industry, but not non-

industry, investment experience. Whether that effect is from greater knowledge of the

industry or better networks that allow for recruitment of senior management, customers,

and strategic partners needs further exploration.

A variety of open issues remain for future research. First, as we acknowledge

above, the precise mechanisms behind the relative performance of more industry

experienced and specialized organizations remain unclear. For instance, is it possible to

disentangle the relative importance of superior investment selection and ability to add

value from the ability to persuade entrepreneurs to accept ones’ capital? (While Kaplan

and Stromberg (2004) present an intriguing initial look at the venture capital decision-

making process, many open questions remain. Sorensen (2004) represents another

important step in untangling these issues.) Second, because we sought to examine

investment outcomes, our analysis only extends through 1998: we do not analyze the

9One might have thought that overall experience would also have been an important explanation for two reasons. First, the most experienced venture capital firms tend to have the greatest access to financial capital. They may already have raised large funds or they may have established reputations and networks that enable them to raise easily additional capital. Second, firms with the most overall experience may have access to a large pool of human capital that they can redeploy across sectors. That is, one might think of venture capital firms as having an internal labor market to complement an internal capital market. However, our finding that industry experience is the key driver of investment activity suggests that it is not easy to redeploy venture capitalists across sectors. This would be the case if human capital in other sectors—in the case of venture capitalists within an organization that specialize in a given industry, say biotechnology—were unable or unwilling to shift focus to a different industry, e.g., the Internet. This prediction is in line with the view that diversified firms have a difficult time redeploying capital into sectors with more investment opportunities: see Scharfstein and Stein (2000), Scharfstein (1998), and Rajan, Servaes, and Zingales (2000). Fulghieri and Sevilir (2004) model some of these issues in a venture capital context.

29

events of 1999 and 2000. While the venture capital market has seen many cycles in the

past, the magnitude of the boom and bust during this period was second to none.

Understanding whether the patterns delineated above continued to hold during that most

dramatic of cycles is an important question for future researchers to examine.

30

References

Baker, Malcolm, Jeremy C. Stein, and Jeffrey Wurgler, When Does the Market Matter? Stock Prices and the Investment of Equity-Dependent Firms, The Quarterly Journal of Economics 118. 2003. 969-1005. Cochrane, John H., The Risk and Return of Venture Capital, Journal of Financial Economics 75. 2005. 3-52. DeBondt, Werner F. M., and Richard H. Thaler, Does the Stock Market Overreact? Journal of Finance 40. 1985. 793-805. DeBondt, Werner F. M., and Richard H. Thaler, Do Security Analysts Overreact? American Economic Review Papers and Proceedings 80. 1990. 52-57. Fama, Eugene F., Market Efficiency, Long-Term Returns, and Behavioral Finance, Journal of Financial Economics 49. 1998. 283-306. Fulghieri, Paolo, and Merih Sevilir, Size and Focus of Venture Capital Portfolios, Unpublished University of North Carolina Working Paper. 2004. Gompers, Paul A., and Josh Lerner, Venture Capital Distributions: Short-Run and Long-Run Reactions, Journal of Finance 53. 1998a. 2161-2183. Gompers, Paul A. and Josh Lerner, What Drives Venture Capital Fundraising?, Brookings Proceedings on Microeconomic Activity. 1998b. 149-204. Gompers, Paul A. and Josh Lerner, Money Chasing Deals?: The Impact of Fund Inflows on the Valuation of Private Equity Investments, Journal of Financial Economics 55. 2000. 281-325. Gompers, Paul A., and Josh Lerner, The Venture Capital Cycle. Cambridge and London: MIT Press. 2004. Gupta, Udayan, Done Deals: Venture Capitalists Tell Their Stories. Boston: Harvard Business School Press. 2000. Hanley, Kathleen W., The Underpricing of Initial Public Offerings and the Partial Adjustment Phenomenon, Journal of Financial Economics 34. 1993. 231-250. Hong, Harrison, and Jeremy C. Stein, 1999, A Unified Theory of Underreaction, Momentum Trading, and Overreaction in Asset Markets, Journal of Finance 54. 1999. 2143-2184.

31

Jeng, Leslie A., and Philippe C. Wells, The Determinants of Venture Capital Funding: Evidence across Countries, Journal of Corporate Finance 6. 2000. 241-289. Kaplan, Steven N., and Antoinette Schoar, Private Equity Performance: Returns, Persistence and Capital, Journal of Finance 60. 2005. 1791-1823. Kaplan, Steven N., and Per Stromberg, Characteristics, Contracts, and Actions: Evidence from Venture Capitalist Analyses, Journal of Finance 59. 2004. 2177-2210. Ljungqvist, Alexander, and Matthew P. Richardson, The Cash Flow, Return and Risk Characteristics of Private Equity, Unpublished New York University Working Paper. 2003. Loughran, Tim, and Jay Ritter, The New Issues Puzzle, Journal of Finance 50. 1995. 23-51. Pagano, Marco, Fabio Panetta, and Luigi Zingales. Why Do Companies Go Public? An Empirical Analysis, Journal of Finance 53. 1998. 27-64. Rajan, Raghuram G., Henri Servaes, and Luigi Zingales, The Cost of Diversity: Diversification Discount and Inefficient Investment, Journal of Finance 55. 2000. 35-80. Ritter, Jay R., and Ivo Welch, A Review of IPO Activity, Pricing, and Allocations, Journal of Finance 57. 2002. 1795-1828. Scharfstein, David S., The Dark Side of Internal Capital Markets II: Evidence from Diversified Conglomerates, National Bureau of Economic Research Working Paper 6352. 1998. Scharfstein, David S. and Jeremy C. Stein, The Dark Side of Internal Capital Markets: Divisional Rent-Seeking and Inefficient Investment, Journal of Finance 55. 2000. 2537-64. Sorensen, Morten, How Smart is Smart Money? An Empirical Two-Sided Matching Model of Venture Capital, Unpublished University of Chicago Working Paper. 2004.

32

Figure 1: IPOs and Number of Investments for Selected Industries The graphs show years on the x-axis, the number of venture investments in the industry as a line calibrated on the left y-axis and the number of IPOs as bars calibrated on the right y-axis.

IPOs and Number of Investments -- Internet and Computers

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1975 1979 1983 1987 1991 1995 1999 2003

Inve

stm

ents

020406080100120140160180200

IPO

s

Investments IPOs

IPOs and Number of Investments -- Biotech and Healthcare

0

200

400

600

800

1000

1200

1400

1975 1979 1983 1987 1991 1995 1999 2003

Inve

stm

ents

0

10

20

30

40

50

60

70

80

IPO

s

Investments IPOs

IPOs and Number of Investments -- Communications

0

500

1000

1500

2000

2500

3000

1975 1979 1983 1987 1991 1995 1999 2003

Inve

stm

ents

0

10

20

30

40

50

60

70

IPO

s

Investments IPOs

IPOs and Number of Investments -- Energy

0

20

40

60

80

100

120

1975 1979 1983 1987 1991 1995 1999 2003In

vest

men

ts

0

5

10

15

20

25

30

IPO

s

Investments IPOs

33

Figure 2: Q and Number of Investments for Selected Industries The graphs show years on the x-axis, the number of venture investments in the industry as a line calibrated on the left y-axis and Q as bars calibrated on the right y-axis.

Q and Number of Investments -- Internet and Computers

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

1975 1979 1983 1987 1991 1995 1999 2003

Inve

stm

ents

0

1

2

3

4

5

6

7

Q

Investments Q

Q and Number of Investments -- Biotech and Healthcare

0

200

400

600

800

1000

1200

1400

1975 1979 1983 1987 1991 1995 1999 2003

Inve

stm

ents

0

0.5

1

1.5

2

2.5

3

3.5

4

Q

Investments Q

Q and Number of Investments -- Communications

0

500

1000

1500

2000

2500

3000

1975 1979 1983 1987 1991 1995 1999 2003

Inve

stm

ents

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

Q

Investments Q

Q and Number of Investments -- Energy

0

20

40

60

80

100

120

1975 1979 1983 1987 1991 1995 1999 2003

Inve

stm

ents

0

0.5

1

1.5

2

2.5

3

Q

Investments Q

Table 1: Sample Characteristics Panel A: Sample by Industry Industry Companies Obs. Internet and Computers 3,660 11,404 Communications and Electronics 1,785 6,009 Business / Industrial 1,054 1,699 Consumer 1,798 3,229 Energy 608 1,331 Biotech and Healthcare 2,169 6,827 Financial Services 499 745 Business Services 396 600 All other 613 842 Total 12,582 32,686

Panel B: Sample Characteristics 0.25 0.50 0.75 Mean S.D. N Sample Overall Experience 9 18 42 34.08 44.01 76,296 Industry Experience 0 1 4 3.74 8.41 76,296 Specialization 0.00% 4.17% 16.67% 10.81% 15.68% 75,944 Adjusted Overall Experience -24.84 -13.85 7.76 -0.52 43.33 76,296 Adjusted Industry Experience -2.27 -0.73 0.44 0.00 7.62 76,296 Adjusted Non-Industry Experience -21.60 -11.83 7.45 0.00 39.02 76,296 Adjusted Log Overall Experience -0.7374 -0.1462 0.6650 -0.0216 0.9449 76,296 Adjusted Log Industry Experience -0.5329 -0.1564 0.4898 0.0000 0.8212 76,296 Adjusted Log Non-Industry Experience -0.6976 -0.1065 0.6937 0.0000 0.9628 76,296 Investors Only Overall Experience 11 27 61 48.21 59.69 15,985 Industry Experience 1 4 11 8.81 14.23 15,985 Specialization 4.00% 15.15% 29.03% 19.92% 20.13% 15,922 Adjusted Overall Experience -18.38 -3.65 28.21 15.42 58.18 15,985 Adjusted Industry Experience -2.33 -0.11 4.73 3.45 13.11 15,985 Adjusted Non-Industry Experience -15.63 -3.86 22.89 12.46 49.53 15,985 Adjusted Log Overall Experience -0.4670 0.3156 1.0687 0.2994 1.0465 15,985 Adjusted Log Industry Experience -0.3298 0.3899 1.1027 0.4047 0.9787 15,985 Adjusted Log Non-Industry Experience -0.5188 0.2913 1.0812 0.2624 1.1082 15,985

Table 1: Sample Characteristics (cont’d)

Sample in Selected Years 1985 Overall Experience 9 15 35 27.38 31.96 420 Adjusted Overall Experience -18.38 -12.38 7.62 0.00 31.96 420 Industry Experience 0 1 3 2.87 5.98 3,360 Adjusted Industry Experience -2.04 -0.49 0.51 0.00 5.31 3,360 Specialization 0.00% 3.86% 16.67% 10.79% 15.47% 3,336 1990 Overall Experience 11 21 46 36.97 43.23 541 Adjusted Overall Experience -25.97 -15.97 9.03 0.00 43.23 541 Industry Experience 0 1 4 4.20 8.37 4,328 Adjusted Industry Experience -2.62 -0.78 0.55 0.00 7.59 4,328 Specialization 0.00% 5.45% 17.19% 11.57% 15.59% 4,328 1995 Overall Experience 11 22 55 42.49 53.22 574 Adjusted Overall Experience -31.49 -20.49 12.51 0.00 53.22 574 Industry Experience 0 1 5 4.83 10.19 4,592 Adjusted Industry Experience -3.58 -0.89 0.42 0.00 9.34 4,592 Specialization 0.00% 4.76% 17.28% 11.53% 16.41% 4,575

Panel A shows the distribution of the sample by industry which includes 12,582 unique companies compiled by Venture Economics, and 32,686 unique VC-company pairs. Panel B summarizes characteristics of venture capital firms in the sample including organization–years only for years after which the organization has been observed making an investment, and ceasing in the year after which the final investment is made. It excludes observations for years before VCs has made 5 investments and excludes VCs who invest in only one year of the sample. The table then presents the same statistics, further restricted to those industry-years where the VCs actually made a new investment (“Investors Only”). It also shows these characteristics in three selected years. Statistics include investments from 1975 to 1998, inclusive, and exclude the industry category all other. Overall Experience is the number of investments made by the venture capital firm previous to the date of its first investment in the portfolio company. Industry Experience is the number of investments made by the venture capital firm previous to the date of its first investment in the portfolio company in that industry. Specialization is Industry Experience divided by Overall Experience.

Table 2: Venture Capital Firm Characteristics

Panel A: Characteristics by Quartile Number of Number of Investments Industry Investments Specialization Success N Mean S.D Mean S.D Mean S.D Mean Overall Experience Quartile

1 3,242 7.51 1.36 2.54 2.33 33.51% 0.2959 52.7% 2 4,750 13.39 3.21 4.43 3.98 32.54% 3.2123 52.5% 3 7,836 28.56 9.54 8.47 7.43 29.12% 3.9809 53.8% 4 16,016 113.06 82.15 27.06 27.14 23.97% 0.2680 56.6%

Industry Experience Quartile 1 3,546 18.38 19.54 0.51 0.73 5.12% 0.0828 49.9% 2 3,650 18.70 16.93 2.78 1.44 23.46% 0.1828 52.8% 3 6,977 30.26 24.83 6.28 3.54 31.24% 0.2383 54.0% 4 17,671 100.61 84.63 26.78 25.67 31.33% 0.2012 57.0%

Specialization Quartile 1 3,298 30.84 37.31 1.51 4.41 2.92% 0.0481 49.9% 2 5,338 76.30 73.87 11.61 13.67 15.55% 0.0789 56.1% 3 10,406 91.14 89.75 21.64 27.16 23.12% 0.1204 56.7% 4 12,802 51.95 61.41 18.51 21.66 42.35% 0.2351 53.3%

Panel B: Correlations Adjusted Log

(N=31,844) Overall Industry Special- Overall Industry Non-

Industry Experience Experience ization Experience Experience Experience Overall Experience 1.0000 Industry Experience 0.7804 1.0000 Specialization -0.1057 0.2943 1.0000 Adj. Log Overall Experience 0.8419 0.6471 -0.1733 1.0000 Adj. Log Industry Experience 0.6998 0.7509 0.3230 0.7881 1.0000 Adj. Log Non-Industry Exp. 0.7874 0.5203 -0.4416 0.9421 0.6042 1.0000

Panel A shows the composition of the Overall Experience, Industry Experience, and Specialization quartiles and mean values for selected characteristics of the quartiles. Data are on a VC-company pair observation level. Quartiles were composed at the beginning of each calendar year based on the values at the end of the previous year for each venture capital organization with investments in that year. Industry experience and specialization quartiles were calculated separately for each industry, so that industries with fewer investments would not be disproportionately sampled in lower quartiles. The first quartile represents the least experienced or specialized, while the fourth is the highest. Panel B details the simple correlations between Overall Experience, Industry Experience, and Specialization.

Table 3: Summary Statistics by IPO and Q Quartile IPO Quartile Mean N

1 Firm Investments in Industry In Year 0.47 8,071 Round Number of Investment 1.79 3,559 Adjusted Overall Experience 19.64 3,559

2 Firm Investments in Industry In Year 0.36 18,944 Round Number of Investment 2.06 6,472 Adjusted Overall Experience 26.82 6,472

3 Firm Investments in Industry In Year 0.43 22,765 Round Number of Investment 2.18 9,431 Adjusted Overall Experience 36.40 9,431

4 Firm Investments in Industry In Year 0.50 26,516 Round Number of Investment 2.18 12,382

Adjusted Overall Experience 38.76 12,382Total Firm Investments in Industry In Year 0.44 76,296 Round Number of Investment 2.11 31,844 Adjusted Overall Experience 33.50 31,844 Q Quartile Mean N

1 Firm Investments in Industry in Year 0.37 4,562 Round Number of Investment 1.50 1,523 Adjusted Overall Experience 8.71 1,523

2 Firm Investments in Industry in Year 0.43 18,593 Round Number of Investment 2.06 7,628 Adjusted Overall Experience 26.16 7,628

3 Firm Investments in Industry in Year 0.48 23,088 Round Number of Investment 2.20 10,490 Adjusted Overall Experience 29.32 10,490

4 Firm Investments in Industry in Year 0.43 30,053 Round Number of Investment 2.15 12,203

Adjusted Overall Experience 44.77 12,203Total Firm Investments in Industry in Year 0.44 76,296 Round Number of Investment 2.11 31,844 Adjusted Overall Experience 33.50 31,844 Data are on a VC-industry-year observation level for “Firm Investments in Industry in Year,” and at a VC-company observation level for “Round Number of Investment” and “Adjusted Overall Experience.” Quartiles were computed separately for each industry so that industries with fewer investments would not be disproportionately sampled in lower quartiles. The first quartile represents the lowest number of IPOs or Q level, while the fourth is the highest.

Table 4: Impact of Public Market Signals (1) (2) (3) (4) (5) (6) Detrended Detrended Dependent Variable Industry Industry Industry Industry Industry Industry Investments Investments Investments Investments Investments Investments Model OLS OLS OLS OLS Panel Regression Corrected for AR(1) Lagged IPOs 0.2264 0.3508 0.0417 [4.25] *** [6.08] *** [0.93] Lagged Q 0.4797 0.3617 0.3303 [4.07] *** [2.25] ** [2.81] *** Industry Fixed Effects Yes Yes No No No No Year Fixed Effects Yes Yes No No No No Detrended No No Yes Yes No No Adj. R-squared 92.37% 92.30% 16.27% 2.59% N 192 192 192 192 192 192

The sample consists of yearly observations with one observation per industry year for 1975 to 1998, inclusive, excluding the industry all other. The dependent variable is the is the log of the number of investments made by all venture organizations in industry g in year t. Lagged IPOs is the log of the number of initial public offerings (IPOs) of venture-backed companies in industry g in year t-1. Lagged Q is the market to book ratio of companies in SIC codes mapping to industry g weighted by the number of public venture backed IPOs in that SIC code (equal weighted by companies within each SIC code) in year t-1. Detrended regressions are the pooled regressions of the residuals of the dependent and independent variables regressed against year, with a separate regression run for each industry. Controls in some regressions include industry and year fixed effects. ***, **, * indicate statistical significance at the 1%, 5% and 10% level, respectively.

Table 5: Investment Patterns (No Interactions)

(1) (2) (3) (4) (5) (6) (7) Dependent Variable

Firm Industry

Firm Industry

Firm Industry

Firm Industry

Firm Industry

Firm Industry

Firm Industry

Investments Investments Investments Investments Investments Investments Investments PM Measure IPOs IPOs IPOs IPOs IPOs Q Q PM Measure 0.0396 0.0393 0.0396 0.0392 0.0392 0.0484 0.0482 [13.66] *** [13.69] *** [14.05] *** [13.77] *** [13.81] *** [5.97] *** [6.02] *** Experience 0.0911 0.0964 0.0964 [12.48] *** [13.13] *** [13.13] *** Industry Exp. 0.1726 0.1726 [24.22] *** [24.17] *** Non-Ind. Exp. -0.0102 -0.0102 [2.12] ** [2.12] ** Specialization 0.7281 0.7333 0.7333 [20.94] *** [21.35] *** [21.31] *** Fixed Effects: Industry Industry Industry Industry Industry Industry Industry Year Year Year Year Year Year Year Adj. R-squared 15.13% 18.64% 24.01% 19.03% 22.82% 23.92% 22.73% N 76,296 76,296 76,296 75,944 75,944 76,296 75,944

The sample consists of aggregated investments by industry by year for 1,008 VCs in 8 industries from 1975 to 1998, inclusive, as compiled by Venture Economics. Observations includes VC organization–years only for years after which the organization has been observed making an investment, and cease in the year after which the final investment is made. It excludes observations for years before VCs has made 5 investments and excludes VCs who invest in only one year of the sample. The dependent variable is the log of the number of investments made by venture organization f in industry g in year t. The public market measure (PM Measure) is either Lagged IPOs, the log of the number of initial public offerings (IPOs) of venture-backed companies in industry g in year t-1 or Lagged Q, the market to book ratio of companies in SIC codes mapping to industry g weighted by the number of public venture backed IPOs in that SIC code (equal weighted by companies within each SIC code) in year t-1. Experiencet is the difference between the log of the number of investments made by venture capital organization f prior to year t and the average of the number of investments made by all organizations prior to year t. Industry Experiencet is the difference between the log of the number of investments made by venture capital organization f in industry g prior to year t and the average of the number of investments made by all organizations in industry g prior to year y. Non-Industry Experience is the difference between the log of the number of investments made by venture capital

organization f in industries other than g (~g) prior to year t and the average of the number of investments made by all organizations in all industries other than g (~g) prior to year t. Specializationt is the difference between the number of investments made by venture capital organization f in industry g divided by the number of investments made by the venture organization in total prior to year t and the average of the same figure for all organizations in year t. Controls include industry and year fixed effects. T-statistics in italics below coefficient estimates are based on robust errors allowing for data clustering by venture capital organization. ***, **, * indicate statistical significance at the 1%, 5% and 10% level, respectively.

Table 6: Investment Patterns (Includes Interactions of IPOs)

(1) (2) (3) (4) (5) (6) (7) (8)

Dependent Variable Firm

Industry Firm

Industry Firm

Industry Firm

Industry Firm

Industry Firm

Industry Firm

Industry Firm

Industry Invest. Invest. Invest. Invest. Invest. Invest. Invest. Invest. PM Measure IPOs IPOs IPOs IPOs IPOs IPOs Q Q PM Measure 0.0405 0.0396 0.0396 0.0396 0.0392 0.0396 0.0484 0.0484 [14.26] *** [14.23] *** [14.09] *** [13.94] *** [13.70] *** [14.24] *** [5.81] *** [5.89] *** Experience 0.0039 0.0005 0.0364 [0.49] [0.06] [2.55] ** Industry Experience 0.0367 0.0072 0.0581 [3.52] *** [0.72] [4.21] *** Non-Industry Exp. 0.0225 0.0473 0.0751 [2.82] *** [7.39] *** [6.20] *** Specialization 0.1713 0.1817 0.3397 [2.21] ** [2.46] ** [4.23] *** Exp. * PM Measure 0.0346 0.0378 0.0323 [7.28] *** [8.14] *** [3.63] *** Ind. Exp.* PM Measure 0.0441 0.0577 0.0594 [9.72] *** [13.79] *** [8.08] *** Non-Ind. Exp. * PM Measure 0.0180 -0.0225 -0.0462 [3.63] *** [6.51] *** [6.40] *** Specialization * PM Measure 0.1755 0.1731 *** 0.1858 *** [6.84] *** [7.14] [5.02] Controls: Industry Industry Industry Industry Industry Industry Industry Industry Year Year Year Year Year Year Year Year Adj. R-squared 19.25% 24.54% 17.37% 24.76% 19.21% 23.67% 24.32% 23.00% N 76,296 76,296 76,296 76,296 75,944 75,944 76,296 75,944

The sample consists of aggregated investments by industry by year for 1,008 VCs in 8 industries from 1975 to 1998, inclusive, as compiled by Venture Economics. Observations includes VC organization–years only for years after which the organization has been observed making an investment, and cease in the year after which the final investment is made. It excludes observations for years before VCs has made 5 investments and excludes VCs who invest in only one year of the sample. The dependent variable is the log of the number of investments made by venture organization f in industry g in year t. The public market measure (PM Measure) is either Lagged IPOs, the log of the number of initial public offerings (IPOs) of venture-backed companies in industry g in year t-1 or Lagged Q, the market to book ratio of companies in SIC codes mapping to industry g weighted by the number of public venture backed IPOs in that SIC code (equal weighted by companies within each SIC code) in year t-1. Experience is the difference between the log of the number of investments made by venture capital organization f prior to year t and the average of the number of investments made by all organizations prior to year t. Industry Experience is the difference between the log of the number of investments made by venture capital organization f in industry g prior to year t and the average of the number of investments made by all organizations in industry g prior to year t. Non-Industry Experience is the difference between the log of the number of investments made by venture capital organization f in industries other than g (~g) prior to year t and the average of the number of investments made by all organizations in all industries other than g (~g) prior to year t. Specialization is the difference between the number of investments made by venture capital organization f in industry g divided by the number of investments made by the venture organization in total prior to year t and the average of the same figure for all organizations in year t. Controls include industry and year fixed effects. T-statistics in italics below coefficient estimates are based on robust standard errors allowing for data clustering by venture capital organization. ***, **, * indicate statistical significance at the 1%, 5% and 10% level, respectively.

Table 7: Investment Patterns for Organizations That Made Investments in that Industry in that Year

(1) (2) (3) (4) (5) (6) (7) (8)

Dependent Variable Firm

Industry Firm

Industry Firm

Industry Firm

Industry Firm

Industry Firm

Industry Firm

Industry Firm

Industry Invest. Invest. Invest. Invest. Invest. Invest. Invest. Invest. (>0) (>0) (>0) (>0) (>0) (>0) (>0) (>0) PM Measure IPOs IPOs IPOs IPOs IPOs IPOs Q Q PM Measure 0.0844 0.0799 0.0888 0.0779 0.0806 0.0731 0.1502 0.1459 [11.68] *** [12.22] *** [12.30] *** [12.16] *** [11.56] *** [10.82] *** [11.91] *** [10.95] *** Experience 0.0215 0.0154 0.0079 [1.38] [0.99] [0.52] Industry Experience 0.0744 0.0599 0.0573 [4.03] *** [3.59] *** [3.44] *** Non-Ind. Exp. 0.0309 0.0201 0.0133 [2.00] ** [1.55] [1.06] Specialization 0.3814 0.3858 0.3735 [3.89] *** [4.01] *** [3.87] ***

0.0177 0.0256 0.0281 Experience * PM Measure [3.14] *** [4.60] *** [5.19] ***

0.0196 0.0358 0.0366 Industry Experience * PM Measure [3.25] *** [6.44] *** [6.70] ***

0.0053 -0.0224 -0.0202 Non-Industry Experience * PM Measure [0.95] [5.13] *** [4.71] ***

0.0451 0.0639 0.0674 Specialization * PM Measure [1.53] [2.20] ** [2.33] ** Fixed Effects: Industry Industry Industry Industry Industry Industry Industry Industry Year Year Year Year Year Year Year Year Adj. R-squared 17.23% 23.04% 15.35% 24.07% 17.94% 22.81% 24.26% 23.02% N 15,985 15,985 15,985 15,985 15,922 15,922 15,985 15,922

The sample consists of aggregated investments by industry by year for 1,008 VCs in 8 industries from 1975 to 1998, inclusive, as compiled by Venture Economics. Observations include VC organization–years only for years in which the organization has made an investment in that industry. It excludes observations for years before VCs has made 5 investments and excludes VCs who invest in only one year of the sample. The dependent variable is the log of the number of investments made by venture organization f in industry g in year t. The public market measure (PM Measure) is either Lagged IPOs, the log of the number of initial public offerings (IPOs) of venture-backed companies in industry g in year t-1 or Lagged Q, the market to book ratio of companies in SIC codes mapping to industry g weighted by the number of public venture backed IPOs in that SIC code (equal weighted by companies within each SIC code) in year t-1. Experience is the difference between the log of the number of investments made by venture capital organization f prior to year t and the average of the number of investments made by all organizations prior to year t. Industry Experience is the difference between the log of the number of investments made by venture capital organization f in industry g prior to year t and the average of the number of investments made by all organizations in industry g prior to year t. Non-Industry Experience is the difference between the log of the number of investments made by venture capital organization f in industries other than g (~g) prior to year t and the average of the number of investments made by all organizations in all industries other than g (~g) prior to year t. Specialization is the difference between the number of investments made by venture capital organization f in industry g divided by the number of investments made by the venture organization in total prior to year t and the average of the same figure for all organizations in year t. Controls include industry and year fixed effects. T-statistics in italics below coefficient estimates are based on robust standard errors allowing for data clustering by venture capital organization. ***, **, * indicate statistical significance at the 1%, 5% and 10% level, respectively.

Table 8: Success

(1) (2) (3) (4) (5) (6) (7) (8) (9) Dependent Variable Success Success Success Success Success Success Success Success Success PM Measure IPOs IPOs IPOs IPOs IPOs IPOs IPOs Q Q PM Measure 0.0010 -0.0103 -0.0097 -0.0121 -0.0121 -0.0116 -0.0118 -0.0276 -0.0288 [0.09] [0.84] [0.79] [0.98] [0.98] [0.94] [0.96] [1.33] [1.39] Experience 0.0223 0.0261 0.0261 [7.74] *** [8.65] *** [8.67] *** Industry Experience 0.0270 0.0270 0.0270 [8.66] *** [6.53] *** [6.50] ***

0.0000 0.0000 Non-Industry Experience [0.00] [0.01] Specialization 0.0496 0.0849 0.0851 [2.99] *** [4.89] *** [4.90] *** Fixed Effects: Industry Industry Industry Industry Industry Industry Industry Industry Industry Stage Stage Stage Stage Stage Stage Stage Stage Stage Round Round Round Round Round Round Round Round Round Year Year Year Year Year Year Year Year Year Adj. R-squared 7.17% 8.95% 9.14% 9.22% 9.22% 9.00% 9.23% 9.22% 9.23% N 11,969 31,844 31,844 31,844 31,844 31,844 31,844 31,844 31,844

The sample consists of outcome data through 2003 for investments made by 1,008 VCs in 11,969 companies from 1975 to 1998, inclusive, as compiled by Venture Economics. The first specification includes only one observation per company while the remaining specifications include one observation per company-VC pair. The remainder of the specifications includes one observation per unique VC-company pair. The dependent variable is Success a binary variable =1 if the portfolio company was acquired, merged, in registration for an IPO, or went public by the end of 2003, and =0 otherwise. The public market measure (PM Measure) is either Lagged IPOs, the log of the number of initial public offerings (IPOs) of venture-backed companies in industry g in year t-1 or Lagged Q, the market to book ratio of companies in SIC codes mapping to industry g weighted by the number of public venture backed IPOs in that SIC code (equal weighted by companies within each SIC code) in year t-1. Experience is the difference between the log of the number of investments made by venture capital organization f prior to year t and the average of the number of investments made by all organizations prior to year t. Industry Experience is the difference between the log of the number of investments made by venture capital organization f in industry g prior to year t and the average of the number of investments made by all

organizations in industry g prior to year t. Non-Industry Experience is the difference between the log of the number of investments made by venture capital organization f in industries other than g (~g) prior to year t and the average of the number of investments made by all organizations in all industries other than g (~g) prior to year t. Specialization is the difference between the number of investments made by venture capital organization f in industry g divided by the number of investments made by the venture organization in total prior to year t and the average of the same figure for all organizations in year t. Controls include industry, investment stage, round number, and year fixed effects. T-statistics in italics below coefficient estimates are based on robust standard errors allowing for data clustering by venture capital organization. ***, **, * indicate statistical significance at the 1%, 5% and 10% level, respectively.

Table 9: Success (Includes Interactions)

(1) (2) (3) (4) (5) (6) (7) (8) Dependent Variable Success Success Success Success Success Success Success IPO PM Measure IPOs IPOs IPOs IPOs IPOs Q Q IPOs PM Measure -0.0121 -0.0190 -0.0182 -0.0148 -0.0205 -0.0359 -0.0359 -0.0223 [0.96] [1.48] [1.42] [1.19] [1.60] [1.69] * [1.72] * [1.57] Experience 0.0119 0.0052 0.0111 [1.09] [0.46] [1.01] Industry Exp. 0.0001 -0.0159 -0.0005 -0.0213 [0.01] [1.04] [0.04] [1.38]

0.0258 0.0228 0.0212 Non-Industry Experience [2.08] ** [2.14] ** [1.54] Specialization -0.1179 -0.1183 -0.0107 [1.52] [1.49] [0.19]

0.0032 0.0065 0.0047 Experience * PM Measure [0.98] [1.93] * [1.44]

0.0082 0.0134 0.0130 0.0153 Industry Experience * PM Measure [2.24] ** [2.93] *** [2.21] ** [3.03] ***

-0.0081 -0.0105 -0.0073 Non-Industry Exp. * PM Measure [2.23] ** [2.29] ** [1.81] *

0.0487 0.0597 0.0424 Specialization * PM Measure [2.22] ** [2.64] *** [1.78] * Controls: Industry Industry Industry Industry Industry Industry Industry Industry Stage Stage Stage Stage Stage Stage Stage Stage Round Round Round Round Round Round Round Round Year Year Year Year Year Year Year Year Adj. R-squared 9.14% 9.23% 9.25% 9.00% 9.26% 9.25% 9.25% 10.35% N 31,844 31,844 31,844 31,844 31,844 31,844 31,844 21,333

The sample consists of outcome data through 2003 for investments made by 1,008 VCs in 11,969 companies from 1975 to 1998, inclusive, as compiled by Venture Economics. The specifications include one observation per unique VC-company pair. The dependent variable in the first seven regressions is Success, a binary variable =1 if the portfolio company was acquired, merged, in registration for an IPO, or went public by the end of 2003, and =0 otherwise. In the final regression, the dependent variable is coded as one only if the portfolio company went public or was in registration for an IPO by the end of 2003 (in this case, mergers and acquisitions are dropped from the sample). The public market measure (PM Measure) is either Lagged IPOs, the log of the number of initial public offerings (IPOs) of venture-backed companies in industry g in year t-1 or Lagged Q, the market to book ratio of companies in SIC codes mapping to industry g weighted by the number of public venture backed IPOs in that SIC code (equal weighted by companies within each SIC code) in year t-1. Experience is the difference between the log of the number of investments made by venture capital organization f prior to year t and the average of the number of investments made by all organizations prior to year t. Industry Experience is the difference between the log of the number of investments made by venture capital organization f in industry g prior to year t and the average of the number of investments made by all organizations in industry g prior to year t. Non-Industry Experience is the difference between the log of the number of investments made by venture capital organization f in industries other than g (~g) prior to year t and the average of the number of investments made by all organizations in all industries other than g (~g) prior to year t. Specialization is the difference between the number of investments made by venture capital organization f in industry g divided by the number of investments made by the venture organization in total prior to year t and the average of the same figure for all organizations in year t. Controls include industry, investment stage, round number, and year fixed effects. T-statistics in italics below coefficient estimates are based on robust standard errors allowing for data clustering by venture capital organization. ***, **, * indicate statistical significance at the 1%, 5% and 10% level, respectively.

Table 10: Results in Quartiles

Table 6 Table 7 Table 9 Dependent Variable Firm Industry Firm Industry Investments Investments (>0) Success PM Measure IPOs IPOs IPOs PM Measure 0.0311 0.0735 -0.0213 [6.38] *** [7.99] *** [1.35] Industry Experience Quartile Dummy:

2 0.0102 -0.0596 0.0546 [0.62] [2.05] ** [1.31]

3 -0.0198 -0.0097 0.0145 [1.86] * [0.47] [0.45]

4 -0.0579 0.0231 -0.0037 [3.34] *** [0.92] [0.12]

Interaction of Industry Experience Quartile Dummy and PM Measure: 2 0.0058 0.0315 -0.0062 [0.94] [3.19] *** [0.47]

3 0.0486 0.0368 0.0093 [8.55] *** [4.72] *** [0.89]

4 0.1384 0.0931 0.0247 [14.64] *** [9.44] *** [2.47] **

Non-Industry Experience Quartile Dummy: 2 0.0488 0.0092 0.0169 [4.96] *** [0.42] [0.49]

3 0.0701 0.0643 0.0093 [6.29] *** [2.68] *** [0.30]

4 0.0849 0.0417 0.0096 [5.51] *** [1.50] [0.31] Interaction of Non-Industry Experience Quartile Dummy and PM Measure:

2 -0.0413 -0.0376 -0.0069 [8.22] *** [4.67] *** [0.66]

3 -0.0397 -0.0536 -0.0107 [6.59] *** [5.97] *** [1.14]

4 -0.0297 -0.0369 -0.0041 [3.37] *** [3.39] *** [0.43] Controls: Industry Industry Industry Year Year Year Stage Round Adj. R-squared 23.57% 22.17% 9.25% N 76,296 15,985 31,844

The sample consists of aggregated investments by industry by year for 1,008 VCs in 8 industries and outcome data through 2003 for 11,969 separate companies from 1975 to 1998, inclusive, as compiled by Venture Economics. See the earlier tables for the definitions of the sample used in the regressions. The dependent variable is the log of the number of investments made by venture organization f in industry g in year t or Success a binary variable =1 if the portfolio company was acquired, merged, in registration for an IPO, or went public by the end of 2003, and =0 otherwise. The public market measure (PM Measure) is Lagged IPOs, the log of the number of initial public offerings (IPOs) of venture-backed companies in industry g in year t-1. Quartiles were composed at the beginning of each calendar year based on values at the end of the previous year for each venture capital organization with investments in that year. Industry experience and specialization quartiles were calculated separately for each industry, so that industries with fewer investments would not be disproportionately sampled in lower quartiles. The first quartile represents

the least experienced or specialized, while the fourth is the highest. Controls include industry and year (and in the third regression, investment stage and round number) fixed effects. T-statistics in italics below coefficient estimates are based on robust standard errors allowing for data clustering by venture capital organization. ***, **, * indicate statistical significance at the 1%, 5% and 10% level, respectively.

Table 11: Robustness Checks

(1) (2) (3) (4) (5) Sample Ever Invested Ever Invested Above Above Full Sample in Past in Sample Median Median Fund Size Success Rate Dependent Variable Firm Industry Firm Industry Firm Industry Firm Industry Firm Industry Investments Investments Investments Investments Investments PM Measure IPOs IPOs IPOs IPOs IPO Pop Industry Experience * PM Measure 0.1049 0.0628 0.0658 0.0647 0.0803 [16.32] *** [12.37] *** [11.61] *** [10.33] *** [4.86] *** Non-Industry Experience * PM Measure -0.0358 -0.0258 -0.0346 -0.0169 0.0036 [7.74] *** [6.42] *** [7.69] *** [3.54] *** [0.35] N 48,364 56,588 38,173 38,582 68,067

The regressions here are similar to those reported in regression (4) of Table 6. See the labeling of that that table for the definition of the sample and variables. The sample is varied here to include the following: (Regression 1) only venture capital firms who have ever invested in the industry in the past, (Regression 2) VCs which have ever made an investment in that industry in the sample (past or future), (Regression 3) only VCs with above median fund size, and (Regression 4) only VCs with above median success rates. Regression 5 is based on the full sample, but uses as the measure of public market performance the average first-day returns of all venture-backed IPOs in that sector in the previous year. Only a portion of the regression coefficients are reported in this table. T-statistics in italics below coefficient estimates are based on robust standard errors allowing for data clustering by venture capital organization. ***, **, * indicate statistical significance at the 1%, 5% and 10% level, respectively.