TOLERANCE FOR FAILURE AND CORPORATE INNOVATION

Xuan Tian Kelley School of Business

Indiana University tianx@indiana.edu

(812) 855-3420

Tracy Y. Wang Carlson School of Management

University of Minnesota wangx684@umn.edu

(612) 624-5869

First version: April, 2009 Key words: corporate culture, tolerance for failure, innovation, patents, venture capital, IPO JEL classification: M14, O31, G24, G34

TOLERANCE FOR FAILURE AND CORPORATE INNOVATION

Abstract

In this paper, we examine whether a failure-tolerant corporate culture spurs

corporate innovation and increases firm value based on a sample of venture capital (VC)

backed IPO firms. We develop a novel firm-specific measure of a failure-tolerant

corporate culture hinging on the idea that the formation of corporate culture is largely

determined by the attitudes and beliefs of its founders and early active investors. VC

firms are such investors and their attitudes towards failures can have a profound impact

on the formation of a failure-tolerant culture in the entrepreneurial firms in which they

invest. We find that firms with a more failure-tolerant culture are significantly more

innovative. The failure tolerance effect is persistent and robust to controlling for other

VC firm characteristics. Further, the failure tolerance effect on firm innovation is stronger

in industries in which innovation is more difficult to achieve. We also find opposite roles

of insider equity ownership and failure tolerance in motivating innovation. Finally, we

find that a failure-tolerant corporate culture increases firm value in industries in which

innovation is important. Overall, our findings are consistent with implications in recent

theories on corporate innovation that failure tolerance is critical in motivating innovation.

Our work also contributes to the literature on corporate culture by making the first

attempt at measuring a specific aspect of corporate culture and providing empirical

evidence that corporate culture does matter in significant ways for firms’ decisions and

performances.

TOLERANCE FOR FAILURE AND CORPORATE INNOVATION

1. INTRODUCTION

Corporate culture, the shared beliefs and organizational preferences within a firm about

the optimal course of action, has received growing attention in the economics and finance

literature.1 Economic theories have long argued that corporate culture is an important part of

corporate internal incentivizing and governance mechanisms and thus should have profound

impact on corporate decisions and performances (see an extensive review of this literature in

Hermalin 2001). However, empirical study on the effect of corporate culture remains sparse for

two reasons. First, corporate culture is difficult to define or measure in econometric analysis.

Second, corporate culture has different aspects and there is inadequate theoretical guidance on

how a specific cultural aspect affects firm performance (Ostroff, Kinicki, and Tamkins 2003).

The recent theoretical literature on corporate innovation, however, provides a specific

context to examine the effect of corporate culture on firm performance. This literature shows that

tolerance for failure is critical in motivating innovation (see, e.g., Hellmann and Thiele 2008,

Manso 2008). This is because unlike standard tasks, exploration of new untested approaches is

subject to a high probability of failure. However, there has been little empirical work in this area

because measuring failure tolerance in a firm is difficult.

This paper contributes to both the literature on corporate culture and the literature on

corporate innovation by constructing a novel measure of a failure-tolerant corporate culture. This

measure enables us to examine how such an important aspect of corporate culture affects a firm’s

innovation productivity, which is vital for the long-term competitive advantage of the firm.

The starting point of our intuition is that for a firm the formation of corporate culture is

largely determined by the attitudes and beliefs of its first-generation insiders, namely, founders

and early active investors. Therefore, our empirical approach hinges on the idea that early active

investors’ attitudes towards failure have a profound impact on the formation of a failure-tolerant

culture in the firm. Venture-capital-backed IPO firms provide an ideal research setting. First,

venture capital (hereafter VC) investors provide not only capital but also intensive monitoring

and value-added services to entrepreneurial firms from the very beginning stages to the firms’

1 See, e,g., Kreps (1990), Cremer (1993), Lazear (1995), Hermalin (2001), Van den Steen (2005a, 2005b), and Bernhardt, Hughson, and Kutsoati (2006).

1

maturity. Second, entrepreneurial firms receiving VC financing typically have high growth

potential but also substantial failure risk, especially during the earlier stages of the firms’ life

cycle. Thus VC investors’ attitudes towards early failures may significantly influence the

entrepreneurs’ attitudes towards failures and the projects’ eventual outcomes. Put differently,

entrepreneurial firms backed by failure-tolerant VC investors are more likely to develop a

failure-tolerant culture.

We infer a VC firm’s failure tolerance by examining how long the VC firm invests in its

eventually failed projects (i.e., entrepreneurial firms that were written off by their VC firms).

Failure tolerance is captured by the VC firm’s average investment duration in those projects from

the first participation investment round to the time when it decides to terminate its follow-on

investments. Longer investment duration in eventually failed projects indicates larger tolerance

for early failure. We examine how a VC firm’s failure tolerance is correlated with its other

characteristics. We find that VC firms with more past investment experience, more industry-

specific skills and expertise, and tougher local competition are more tolerant of early failures.

However, a VC firm’s past successful experience does not impact its failure tolerance.

We then link a VC firm’s failure tolerance to the IPO firms backed by the VC firm. An

IPO firm’s failure tolerance is determined by its investing VC firm’s failure tolerance at the time

when the VC firm makes the first round investment. The key intuition is that IPO firms

internalize their investing VC investors’ attitudes towards failure into their own corporate culture,

which in turn affects their subsequent innovation productivity.

Our empirical analysis generates a number of interesting new findings. First, we find that

firms with a more failure-tolerant culture are significantly more innovative. They not only

produce a larger number of patents but also produce patents with significantly larger impact

(measured by the number of citations each patent receives), after controlling for other firm and

industry characteristics. The failure tolerance effect on innovation is robust after controlling for

other VC firm characteristics such as VC firm experience and industry expertise and lead VC

firm fixed effects. This implies that both observable and non-observable VC firm characteristics

are unlikely to be the omitted variables that cause the correlation between failure tolerance and

innovation of IPO firms. We also find that the failure tolerance effect persists long after the VC

firms exit their investments in the IPO firms. This suggests that the failure tolerance effect does

2

not simply reflect a temporary VC firm influence, but is likely to reflect a persistent cultural

effect.

Second, we find that the effect of failure tolerance on firm innovation is stronger in

industries in which innovation is more difficult to achieve in the sense that time and resources for

innovation are more demanding and chances of success are lower. For example, the impact of

failure tolerance on innovation productivity is almost four times higher in industries developing

new drugs than in industries producing new computer software.

Third, we contrast the effect of standard incentive schemes such as insider equity

ownership with the effect of a failure-tolerant corporate culture in motivating innovation. We

find that large insider equity ownership is associated with lower innovation productivity, which

is opposite to the effect of failure tolerance. Our finding is consistent with the implication of the

recent theoretical literature on corporate innovation that motivating innovation is very different

from motivating efforts on standard activities such as production and marketing.

Finally, we examine the effect of failure tolerance on firm value. We find that a failure-

tolerant corporate culture increases firm value (measured by Tobin’s Q) in industries in which

innovation is important. The evidence suggests that a failure-tolerant corporate culture increases

firm value through its impact on a firm’s innovation productivity and such impact is more

pertinent when innovation is more important.

The most important contribution of our paper is the construction of a firm-specific

measure for an important aspect of corporate culture–tolerance for failure. Our paper contributes

to the literature on corporate innovation because this measure allows us to directly test the

implications in economic theories about motivating innovation, which is an important topic

given the vital role of innovation in creating competitive advantages for a firm and for an

economy. We discuss the relevant theories on innovation in detail in Section 2. There is also a

growing empirical literature in corporate finance on corporate innovation. Several papers show

that the legal system matters for innovation (see, e.g., Acharya and Subramanian 2009 on the

effect of a creditor-friendly bankruptcy code, Armour and Cumming 2008 on “forgiving”

personal bankruptcy laws, Acharya, Baghai-Wadji, and Subramanian 2009 on the effect of

stringent labor laws, and Sapra, Subramanian, and Subramanian 2009 on the effect of anti-

takeover laws). Another set of papers find that a firm’s ownership structure and financing matter

for innovation (see, e.g., Ayyagari, Demirgüc-Kunt, and Maksimovic 2007 on private versus

3

state ownership and access to external financing, Aghion, Van Reenen, and Zingales 2009 on the

role of institutional equity ownership, Belenzon and Berkovitz 2007 on the effect of business

group affiliation and internal capital market, and Atannassov, Nanda, and Seru 2007 on the

effects of arm’s length financing versus relationship-based bank financing). Our paper

contributes to this literature by documenting the effect of a failure-tolerant corporate culture on

firm innovation.

Our paper is also one of the very first papers that empirically examine the effect of

corporate culture on corporate performance. The only other empirical paper we are aware of that

is related to corporate culture is Cronqvist, Low, and Nilsson (2009).2 The authors find that spin-

off firms’ investment and financing decisions are similar to those of their parent firms, and such

similarity persists over a long period. They also find that the commonality is stronger for

internally grown spin-offs, and the commonality exists even when a new outside CEO is hired to

run the spin-off firms. The authors argue that their findings are consistent with a corporate

culture effect. The common culture between the parent firm and the spin-off firm leads to

persistent similarities in their decisions. Their work is an interesting and important attempt at

understanding corporate culture effects.

While Cronqvist et al. identify the corporate culture effect by examining the role of a

common firm origin between spin-off firms and their parent firms in their decision-making, our

approach of identifying corporate culture starts at the very beginning of a firm’s history. We

show that the failure tolerance of a firm’s first-generation insiders (VC investors) have a

significant and long-lasting impact on the firm’s innovation productivity.

Another advantage of our approach is that our failure tolerance measure is least subject to

the reverse causality problem in corporate finance research. This is because our failure tolerance

measure captures the investing VC firm’s attitude towards failure before its very first investment

in an IPO firm, which is well before the observed innovation activities of the IPO firm. Therefore,

our empirical design allows us to examine the causal effect of failure tolerance on an

entrepreneurial firm’s innovation productivity and firm value.

2 The organizational behavior literature has also studied corporate culture (see Ostroff, Kinicki, and Tamkins 2003 for a review). But this literature tends to rely on case studies or surveys of employees on some selected dimensions, and thus the studies are limited to a small sample of firms. This literature also does not focus on how corporate culture affects a firm’s investment and financing decisions and performance.

4

Finally, our paper also contributes to a broader literature on culture and corporate finance.

This literature has focused on the effects of national culture, examining how differences in

principal religion, language, social normal, and history across countries affect cross-country

variation in corporate finance decisions and policies (see, e.g., Stulz and Williamson 2003 on the

effect of a country’s principal religion on creditor rights, Guison, Sapienza, and Zingales 2009

on perceptions rooted in culture and economic growth among European countries, Bottazzi, Da

Rin, and Hellmann 2007 on the effect of trust in venture capital investment, and Griffin, Li, Yue,

and Zhao 2009 on the role of national culture in corporate risk-taking.) Although national culture

must influence corporate culture, the two are not the same. Exploring the effect of corporate

culture may help us understand the substantial heterogeneity among firms within a country.

Understanding the corporate culture effect may also help us understand the economic meaning of

the significant firm fixed effects in corporate financing and investment decisions documented in

the corporate finance research (see, e.g., Lemmon, Roberts and Zender 2008).

The rest of the paper is organized as follows. Section 2 develops the hypotheses. Section

3 discusses the construction of our failure tolerance measure. Section 4 describes the empirical

specification. Section 5 reports the results regarding the effects of failure tolerance on firm

innovation and the robustness tests. Section 6 contrasts the effect of standard incentive schemes

with that of a failure-tolerant corporate culture on firm innovation. Section 7 presents the results

on the effect of failure tolerance on firm value. Section 8 concludes.

2. HYPOTHESIS DEVELOPMENT

We derive our hypotheses from the recent theoretical literature on motivating innovation.

Holmstrom (1989), in a simple principle-agent model, shows that innovation activities may mix

poorly with relatively routine activities in an organization because these two types of activities

require different incentive schemes. Innovation activity requires exceptional tolerance for failure

and a weak incentive scheme because of the high-risk and low-predictability nature of innovation.

Manso (2008) explicitly models the innovation process and the trade-off between

exploration of new untested actions and exploitation of well known actions. Manso shows that

while standard pay-for-performance incentive scheme can motivate exploitation, the optimal

contracts that motivate exploration involve a combination of tolerance for failures in the short-

5

run and reward for success in the long-run. Ederer and Manso (2009) conduct a controlled

laboratory experiment and provide evidence supporting the implications in Manso (2008).

While Manso (2008) assumes contractibility of innovation activities, Hellmann and

Thiele (2008) argue that innovative tasks are better characterized by incomplete contracts and

ex-post bargaining (also see Aghion and Tirole 1994). The authors focus on the interaction

between standard tasks and innovation in a multi-task model in which employees choose

between the two types of activities, and show that the amount of innovation is negatively related

to the strength of incentives provided for the standard task. The optimal amount of failure

tolerance also reflects the trade-off between the two types of tasks: Failure tolerance may

encourage innovation, but may also undermine incentives for standard tasks.

The essential message from the above theories is that motivating innovation is very

different from motivating efforts on standard tasks. Tolerance for failure is critical in motivating

innovation, and standard incentive schemes such as pay-for-performance can fail to do so. These

insights form the theoretical foundation of our empirical analysis.

Our main task in this study is to examine the effect of failure tolerance on corporate

innovation and firm value. Tolerance for failure may be an explicit contractual specification as in

Manso (2008), or an implicit organizational norm as in Hellmann and Thiele (2008). In our study

we take the latter aspect and examine the effect of a failure-tolerant corporate culture. Our main

hypothesis is summarized below.

Hypothesis 1: A failure-tolerant corporate culture increases a firm's innovation

productivity.

If failure tolerance is important for innovation because innovation activities often involve

substantial risk of failure, then a natural cross-sectional implication is that the effect of failure

tolerance should be particularly strong in industries in which innovation is difficult to achieve.

The difficulty can come from a low probability of success and large resources demanded. This

insight leads to our next hypothesis.

Hypothesis 2: The effect of failure tolerance on corporate innovation is stronger in

industries in which innovation is more difficult to achieve.

The theoretical work discussed in this section contrasts the effect of tolerance for failure

with that of standard incentive schemes in motivating innovation. Although standard pay-for-

performance incentive schemes are effective at motivating efforts on standard tasks, they may

6

fail to motivate innovation. In particular, Hellmann and Thiele (2008) show that when employees

can choose between standard tasks and innovation, high-powered incentives on standard tasks

will discourage innovation. We thus empirically examine and compare the effects of these two

incentive schemes.

Hypothesis 3: Failure tolerance and standard incentive schemes have opposite effects on

a firm’s innovation productivity.

If failure tolerance enhances a firm’s innovation productivity and if such productivity is

priced by investors, then we expect a failure-tolerant culture to increase firm value. But if the

firm’s operation focuses on standard tasks such as production and marketing rather than

innovation, then as argued in Hellmann and Thiele (2008), failure tolerance may decrease firm

value by undermining incentives for standard tasks. This insight leads to our last hypothesis.

Hypothesis 4: A failure-tolerant corporate culture increases firm value in industries in

which innovation is important.

With these four hypotheses in hand, we turn to the specifics of our empirical research

design.

3. FAILURE TOLERANCE

3.1 The Idea

Corporate culture is defined as the shared beliefs and organizational preferences among a

firm’s employees about the optimal course of action (see, e.g., Kreps 1990, Cremer 1993, Lazear

1995, and Hermalin 2001). As Edgar Schein points out, the process of culture formation in an

organization begins with the founding of the group. “Culture is created by shared experience, but

it is the leader who initiates this process by imposing his or her beliefs, values and assumptions

at the outset” (Schein 2004). Thus our starting point is that the formation of a corporate culture is

largely determined by the beliefs and values of its first-generation leaders–founders and early

active investors.

VC-backed IPO firms provide an ideal research setting for our study. Unlike other

passive investors such as angel investors and banks, VC investors not only provide capital but

also interact intensively with the entrepreneurial firms they back. They get heavily involved in a

firm’s everyday business: on-site visiting the firm, helping the professionalization of the

management team, and sitting in board meetings (e.g., Sahlman 1990, Gompers 1995, Lerner

7

1995, Hellmann and Puri 2000 and 2002, Chemmanur and Loutskina 2006, and Bottazzi, Da Rin,

and Hellmann 2008). During the process of active and intensive interactions with the

entrepreneurs, VC firms, usually the first-generation active investors in an entrepreneurial firm,

can help the firm build up its own corporate culture.

Tolerance for early failure is one of the most important aspects of corporate culture that

VC investors can help their portfolio firms build up. Entrepreneurial firms receiving VC

financing are typically premature and highly innovative with high growth potential but also

significant failure risk. The failure risk is especially high during the earlier stages of the

entrepreneurial firm’s life cycle. A VC investor’s attitude towards early failure of the

entrepreneurial firm’s projects may significantly influence the entrepreneur’s attitude towards

failure later on. For example, if the VC investor is very much tolerant for early failure and

willing to give the entrepreneur a second chance when the entrepreneurial firm fails to meet

stage financial or non-financial targets, such failure tolerance from the VC investor gives the

entrepreneur greater incentives to continue his innovation activities. It may also significantly

influence the entrepreneur’s attitude towards failure.

In sum, our intuition is that VC investor’s attitude towards failure can be “inherited” or

internalized by the entrepreneurs and become a part of the entrepreneurial firm’s culture, which

in turn affects the firm’s subsequent innovation activities and performance.

3.2 VC Firm’s Failure Tolerance

In this section we construct a time-varying measure of a VC firm’s tolerance for failure.

In Section 3.4 we link a VC firm’s failure tolerance to the IPO firms it invests in.

We construct our measure for a VC investor’s failure tolerance based on a sample of

failed investments the VC investor made in the past. Failure tolerance measures how long a VC

firm waits before terminating its follow-on investments in an under-performing entrepreneurial

firm that eventually failed. It is well know that the VC industry is full of failure risks and the

organizational structure of VC partnerships gives VC firms limited investment horizon. A typical

VC fund is constructed for 10 years and can be extend to a maximum of 12 years. If a project

does not show immediate progress towards stage targets after the VC firm’s initial few rounds of

investment, the choice between giving the entrepreneur a second chance by continuing to infuse

capital and writing off the project reflects a VC investor’s attitude towards early failure. We

8

therefore use the VC firm’s investment duration between its first capital infusion and the time

when it decides to terminate the investment in an eventually failed project to capture its tolerance

for failure.

We obtain data on round-by-round VC investments from the Thomson Venture

Economics database for entrepreneurial firms that received VC financing between 1980 and

2006.3 From the initial set of 282,752 VC investment round observations, we exclude investment

rounds obtained by financial firms and those with missing or inconsistent data.4 We also exclude

entrepreneurial firms that are in their late or buyout stage when they receive the first round of

VC financing. This is because entrepreneurial firms at later stages are more mature and the

failure risk is significant reduced. Thus a VC firm’s investment duration may not reflect its

failure tolerance. We correct for the Venture Economics’ over-reporting problem, which leaves

us with 228,805 individual financing rounds made by 7,384 distinct VC firms in 46,875 distinct

entrepreneurial firms.5

To construct our VC failure tolerance measure, we focus on the sub-sample of VC firms’

failed investments, i.e., entrepreneurial firms that were written off by their investing VC

investors. Venture Economics provides detailed information on the date and type of the eventual

outcome of each entrepreneurial firm (i.e., IPO, acquisition, or write-off). However, the database

does not mark all written-down firms as write-offs. Therefore, based on the fact that the VC

industry requires investment liquidation within 10 years from the inception of the fund in the

majority of the cases, in addition to the write-offs market by Venture Economics, we classify a

firm as a written-off firm if it did not receive any financing rounds within a 10-year span after

the last financing round it received.

3 We choose 1980 as the beginning year of our sample period because of the regulatory shift in the U.S. Department of Labor’s clarification of the Employee Retirement Income Security Act’s “prudent man” rule in 1979. This Act allowed pension funds to invest in venture capital partnerships, leading to a large influx of capital to venture capital funds and a significant change of venture capital investment activities. 4 For example, some entrepreneurial firms’ first VC financing round dates occur before the founding dates of their investing VC firms, and some entrepreneurial firms’ founding dates occur later tan their IPO dates. 5 Gompers and Lerner (2004) document that the Venture Economics database reports 28% more financing rounds than actually occurred because Thomson frequently splits financing rounds. To correct Venture Economics’ over-reporting of financing rounds, we collect financial information from IPO prospectuses and S-1 registration statements for firms that eventually go public. For firms acquired by public firms, we collect financial information from the acquirers’ proxy, 10-K, or 10-Q statements, which are generally available in the Securities and Exchange Commission’s (SEC) EDGAR database. For firms that are written off or remain private, we eliminate repeated rounds within three months if they share the same amount of round financing.

9

There are 18,546 eventually failed entrepreneurial firms receiving 67,367 investment

rounds from 4,910 VC firms in our sample. For each eventually failed entrepreneurial firm a VC

firm invested in, we calculate the VC firm’s investment duration (in years) from its very first

investment round date to its last participation round date. If the entrepreneurial firm continues to

receive additional financing rounds from other VC investors after the VC firm’s last participation

round, we then calculate the duration from the VC firm’s very first investment round date until

the next financing round date after its last participation round. This is because the VC firm may

make the decision not to invest follow-on funds, i.e., not to tolerate the poor performance of the

entrepreneurial firm any more, only after its detailed evaluations of the firm, which typically

happens right before the next round of capital infusions.6

We then calculate our main variable of interest, Failure Tolerance, by taking the

weighted average of a VC firm’s investment duration in its eventually failed projects up to a

given year. The weight is the VC firm’s investment in a project as a fraction of its total

investment up to that year. For example, if a VC firm has invested in 12 eventually failed

entrepreneurial firms between 1980 and 1987, the weighted average of the investment duration in

these 12 firms is assigned as the value of its Failure Tolerance measure in 1987. Our intention is

to use the variation in VC firms’ investment duration in their eventually failed entrepreneurial

firms to capture their tolerance for failure. Intuitively, the longer the duration between the VC

firm’s first participation round and the time when it decides to terminate its follow-on capital

infusions, the more tolerant the VC firm is for its eventually failed investment.

Table 1 provides the descriptive statistics for our VC failure tolerance variable. There are

18,993 VC firm-year observations with Failure Tolerance information available in our sample.

On average, VC investors wait for about one and half years before deciding to terminate their

follow-on investments in an unpromising deal. Half of the VC firms intend to write-off their

under-performing entrepreneurial firms within one year and three months since its very first

participation financing round. The distribution of Failure Tolerance is right skewed with

skewness of 2.43. Moreover, from an economic perspective, there is a large difference between

waiting for two years rather than one year before terminating an investment, but probably a

6 It is important to note that based on our current algorithm of calculating the VC’s investment duration in an eventually failed entrepreneurial firm, the duration measure for the investing VC firm will be missing if the entrepreneurial firms receives only one round of VC financing in its entire life. This is because we do not know when exactly the VC investor decides to terminate his investments in the under-performing entrepreneurial firm.

10

smaller difference between waiting for ten years versus nine years. Both the skewness and the

likely nonlinearity of the economic impact of VC’s tolerance for failure suggest that a logarithm

transformation of the failure tolerance measure is appropriate. We then use the natural logarithm

of Failure Tolerance as the main measure in the rest of the analysis.

3.3 Failure Tolerance and Venture Capital Firm Characteristics

In this section, we examine how a VC firm’s failure tolerance is correlated with its other

characteristics such as its investment experience, portfolio concentration, past successful

experience, local competition, and macro-level VC industry conditions.

The VC literature suggests that a VC firm’s past investment experience is an important

determinant of the performance of IPO firms it invests in (e.g., Lerner 1994, Gompers 1996,

Chemmanur and Loutskina 2006, Hochberg, Ljungqvist, and Lu 2007, and Sorensen 2007). Are

more experienced VCs also more tolerant of early failure in the entrepreneurial firms they back?

In theory it is not clear. A VC may learn to be more tolerant over time due to the high failure rate

in the VC industry. But past experience may also make a VC more decisive at cutting losses in

unpromising projects and moving on.

We first measure a VC firm’s overall past experience. For each VC firm and each year

we compute the following four VC experience measures: a) the total dollar amount the VC firm

has invested since 1980 (“Past Amount Invested”); b) the total number of firms the VC firm has

invested in since 1980 (“Past Firms Invested”); c) the total dollar amount the VC firm has raised

since 1965 (“Past Fund Raised”); and d) the age of the VC firm measured as the number of years

since its date of inception (“VC Age”).

It is also possible that not only a VC’s overall past experience but also its past successful

experience matter in determining its attitude towards failure. Thus for each VC firm and each

year, we compute “Past Successful Exit” as the proportion of entrepreneurial firms financed by

the VC firm that exited successfully through either going public or acquisition by another

company since 1980. Previous literature also suggests that going public is a more desirable

outcome than acquisitions for both entrepreneurs and VC firms (see, e.g., Sahlman 1990, Brau,

Francis, and Kohers 2003). Therefore, only firms of the best quality may access the public

capital markets through an IPO (Bayar and Chemmanur 2008). Therefore, we calculate “Past

11

IPO Exit” as the proportion of entrepreneurial firms financed by the VC firm that went public

since 1980.

Another important dimension of a VC firm’s experience is its expertise in certain

industries. VC firms are generally highly specialized in a few industries. The concentration of a

VC’s portfolio firms across industries can to some extend reflect such industry expertise. For

example, if a VC firm specializes in investing in biotechnology firms, its portfolio firms tend to

be highly concentrated in the biotechnology industry. Following Kacperczyk, Sialm, and Zheng

(2005), we construct an investment concentration index for each VC firm and each year. The

Venture Economics database assigns each entrepreneurial firm in a VC’s investment portfolio to

one of 18 industries.7 The investment concentration of VC firm-i at year t is defined as the sum

of the squared deviations of the weights (the number of portfolio firm) for each of the 18

different industries held by the VC firm relative to the industry weights of the total venture

investment

tiw ,

tiw , : 2,,

18

1

)( titii

ww −∑=

. It measures how much a VC firm’s portfolio deviates from a

hypothetical VC market portfolio, which consists of all entrepreneurial firms in the industry in

which a VC firm could have invested. The measure equals zero if the VC firm’s portfolio has

exactly the same industry composition as the hypothetical VC market portfolio and increases as a

VC firm’s portfolio becomes more concentrated in a few industries.

The last VC-specific characteristic we consider is the degree of local competition a VC

firm faces. On the one hand, tough competition means that good projects are less available to

each VC firm, which may make VCs more tolerant of early failures in projects they invest in. On

the other hand, competitive pressure may make VCs less tolerant for projects that do not show

immediate promises. To capture the toughness of the local competition in the VC industry, we

count the number of VC firms located within the same zip code area (“Local Competition”). The

intuition is that the more VC firms located within the same zip code area, the tougher the

competition.

Gompers and Lerner (2000) find that a large VC industry fund supply increases the

valuation of a VC firm’s new investments. To see whether ample supply of venture funds make

7 The 18 industries assigned by Venture Economics database are Agriculture/Forestry/Fish, Biotechnology, Business Services, Communications, Computer Hardware, Computer Other, Computer Software, Construction, Consumer-Related, Financial Services, Industrial/Energy, Internet-Specific, Manufacture, Medical/Health, Other, Semiconductor/Electronics, Transportation, and Utilities.

12

VC investors more failure tolerant, we construct a VC industry condition variable, “Industry

Fund Inflow”, which measures the total capital inflows into the VC industry each year.

Table 1 provides the descriptive statistics for these VC characteristics. The average VC

firm in a given year is 7.5 years old and has invested 395 million dollars in 24 entrepreneurial

firms. Among all entrepreneurial firms the average VC firm has financed, 53% had a successful

exit but only 16% went public. The average VC’s portfolio firms are concentrated in a few

industries with the investment concentration index of 0.37 and it needs to compete with about six

other VC firms that are located within the same zip code area.

Table 2 reports the panel regression results with the natural logarithm of a VC firm’s

Failure Tolerance as the dependent variable. In all regressions, we include VC investor’s

investment concentration, past successful experience, local competition, VC fund inflows one

year and two years prior to the financing round year, and VC firm fixed effects and year fixed

effects. To control for potential differences in investment cycles in different industries, in all

regressions we control for the entrepreneurial firms’ industry fixed effects based on the 18-

industry classification in Venture Economics.8 Robust standard errors are clustered by VC firms

because the residuals may be correlated across observations from the same VC firm. Since the

four VC experience variables are highly correlated with each other, we examine their correlation

with Failure Tolerance one by one in the regressions.

In models (1) and (2), the coefficient estimates of the total dollar amount and the total

number of entrepreneurial firms a VC firm has invested in are positive and significant. This

suggests that more past investment experience makes the VC more tolerant of its portfolio firm’s

early failure. The coefficient estimate of the total funds raised by a VC firm is positive and

significant as reported in model (3), suggesting that the VC firm’s tolerance for failure increases

with their fundraising experience. The coefficient estimate of VC age reported in model (4) is not

statistically significant, which suggests that the length of time the VC firm has been in the

profession does not have a significant impact on its attitude towards failure.

The coefficient estimate of a VC firm’s investment concentration index is positive and

significant across all regressions. Since the measure is meant to capture a VC firm’s industry

8 If a VC firm invests in multiple industries in a given year, we choose the industry in which the VC firm invests the largest amount of capital in that year for the industry fixed effect.

13

expertise, the result suggests that the more specialized VC firms are more tolerant of early

failures in their portfolio firms.

A concern one may have is that our measure that is meant to capture a VC firm’s attitude

towards early failure may simply measure how smart a VC firm is in the sense that “smarter” VC

investors should be able to terminate unpromising projects earlier and less sophisticated ones

wait for too long. The positive relationship between our Failure Tolerance measure and VC past

investment experience presented in Table 2 is inconsistent with this argument. To further

examine the validity of this concern, we include the VC’s past successful investment experience

in all regressions. If the argument was true, we should expect a negative relationship between

Failure Tolerance and the VC’s past successful experience. That is, a more successful VC

investor is smarter and won’t wait for too long before abandoning a bad project. However, the

coefficient estimates of a VC’s past successful exit rate are statistically insignificant in all

regressions, which suggests that the VC’s attitude towards failure is independent of its past

successful experience. For further robustness, we run the same regressions with the VC’s past

successful exit rate replaced by its past IPO exit rate. We continue to find statistically

insignificant coefficient estimates. Overall, the Failure Tolerance measure seems not to capture

the smartness of the VC investors but rather the VC’s attitude towards failure.

Table 2 also shows that local VC competition is significantly and positively related to VC

failure tolerance. This suggests that tougher competition in the local area makes a VC firm more

tolerant of its portfolio firms’ early failures. This may be because really promising projects are

relatively scarce due to tough competition.

Finally, a larger VC industry fund inflow in the prior year does not increase VC firms’

failure tolerance, but a larger fund inflow two years earlier does. This suggests that it takes some

time for an increase in the capital supply of the industry to increase VC firms’ failure tolerance.

In sum, we find that a VC firm’s past investment experience, but not past successful

experience, and its industry expertise and competition in the local VC industry make the VC

more tolerant of early failure in the projects it invests in.

3.4 IPO Firm’s Failure Tolerance

14

In this section we link a VC firm’s failure tolerance to the IPO firms it invests in. Our

empirical analysis hinges on the idea that IPO firms internalize their investing VCs’ attitudes

towards failure into their own corporate culture.

Figure 1 illustrates how we match a VC’s failure tolerance with an IPO firm backed by

the VC. Let us consider year t. VC firm-i’s failure tolerance in year t is the average investment

duration in all of its failed projects up to this year. In the same year the VC firm makes its first-

round investment in a new start-up firm-j, and this firm later goes public. Then IPO firm-j’s

failure tolerance is determined by the VC firm-i’s failure tolerance in year t. In sum, an IPO

firm’s failure tolerance is determined by its investing VC firm’s failure tolerance at the time

when the VC firm makes the first round investment in it.

Figure 1: IPO Firm’s Failure Tolerance

1980 t t+k

Firm-j’s IPO VC-i’s starts to invest in Firm-j

VC-i’s failure tolerance at t

VC-i has N failed projects. Compute average investment duration in them.

Firm-j’s failure tolerance

We obtain the list of VC-backed IPOs between 1985 and 2006 from the Securities Data

Company (SDC) Global New Issues database.9 Following the IPO literature, we exclude from

our initial IPO sample spin-offs, closed-end fund, REITs, ADRs, unit offerings, reverse LBOs,

foreign issues, offerings in which the offer price is less than $5, finance (SIC code between 6000

and 6999), and utilities (SIC code between 4900 and 4999). We also corrected for mistakes and

typos in the SDC database following Jay Ritter’s "Corrections to Security Data Company’s IPO

database" (http://bear.cba.ufl.edu/ritter/ipodata.htm). In the end, we have 2,669 VC-backed IPO

firms. We then merge the IPO sample with the Venture Economics database, and exclude IPO

firms from consideration if the identities of the investing VC firms are unavailable through SDC.

9 We choose 1985 as the beginning year of our IPO sample because we want to have a long enough time gap between the beginning year of our VC sample in which the Failure Tolerance measure is constructed and the beginning year of our IPO sample in which the Failure Tolerance measure is utilized. By doing so, we minimize the possibility that the VC-backed IPO firm has no Failure Tolerance information available.

15

For each IPO firm in our sample, we can observe the identity of its investing VC firms,

the value of each VC firm’s Failure Tolerance measure, and other characteristics of those VC

firms as reported in Table 1 at their first participation round dates. If an IPO firm receives

funding from a VC syndicate (about 86% of our sample), we then calculate the weighted-average

of our Failure Tolerance measure. The weight is the investment by a VC firm as a fraction of the

total VC investment received by the IPO firm. Consequently, there is a fixed failure tolerance

measure for each IPO firm in our final sample. Panel A of Table 3 reports the descriptive

statistics of the IPO firms’ Failure Tolerance measure. The mean Failure Tolerance measure is

about one year and ten months and it could be as large as six years and four months.

For each IPO firm we also calculate the weighted-average of VC characteristics including

past investment experience, past successful experience, investment concentration, and local

competition. All these VC characteristics are parallel to our IPO firm Failure Tolerance measure

by construction. The descriptive statistics of VC characteristics are reported in Panel A of Table

3. Compared with the summary statistics of all VC firms that has financed failed projects as

reported in Table 1, the VC investors of our IPO sample are older, have invested more money,

have participated in more firms, have raised more fund, have more diversified investment

portfolio, and have more successful past investment experience. There seems to be no significant

difference regarding the toughness of competition in the local market.

Finally, we extract financial information for our IPO firms from Standard & Poor’s

COMPUSTAT files, stock price and shares outstanding data from CRSP, insider ownership from

the Compact Disclosure database, and institutional investors’ ownership from the Thomson

Financial 13f institutional holdings database. Finally, there are 1,848 VC–backed IPO firms in

our sample with non-missing VC investor identity, financial and ownership information.

4. EMPIRICAL SPECIFICATION

Having constructed a measure of an IPO firm’s failure-tolerant culture, we now proceed

to the next step and examine how failure tolerance affects the firm’s performance after IPO. As

discussed in Section 2, we focus on the effect of failure tolerance on the firm’s innovation

productivity and firm value. Specifically, we estimate the following two models:

titjtiiti uYearIndZeranceFailureTolLnInnovationLn ,,0, )()( ++++×+= δβα (1)

titjtiiti vYearIndZeranceFailureTolLnQLn ,,1, )()( ++++×+= δγα (2)

16

Innovation and Q are our dependent variables and we discuss their constructions in detail in

Sections 4.1 and 4.2. Failure Tolerance is a measure of an IPO firm’s failure-tolerant culture and

is defined in Section 3.4. Following the innovation literature, we control for a vector of firm

characteristics (Z) that may affect a firm’s innovation productivity and firm value. In the baseline

regressions, Z includes firm size (measured by the logarithm of sales), profitability (measured by

ROA), growth opportunities (measured by Tobin’s Q) investments in innovative projects

(measured by R&D expenditures over total assets), capital expenditure, leverage, institutional

ownership, firm age (measured by years since IPO), asset tangibility (measured by net PPE

scaled by total assets), and industry concentration (measured by the sales Herfindahl index).

Finally, Indj and Yeart capture two-digit SIC industry fixed effects and fiscal year fixed effects,

respectively.

One of the advantages of our empirical design is that the main variable of interest,

Failure Tolerance, is least subject to endogeneity or reverse causality concerns. This is because

Failure Tolerance captures the investing VC firm’s attitude towards failure before its very first

investment round in an IPO firm, which happens well before we observe the innovation activities

and valuation of the IPO firm. Therefore, Failure Tolerance is a relative “clean” exogenous

variable in the above empirical specifications.

4.1 Proxies for Innovation

We construct our innovation variables from the latest version of the NBER patent

database created initially by Hall, Jaffe, and Trajtenberg (2001), which contains updated patent

and citation information from 1976 to 2006. The patent database provides annual information

regarding patent assignee names, the number of patents, the number of citations received by each

patent, the technology class of the patent, the year when a patent application was filed, and the

year when the patent was granted. As suggested by the innovation literature (e.g., Griliches,

Pakes, and Hall 1987), the application year is more important than the grant year since it is closer

to the time of the actual innovation than the grant year. We then construct the innovation

variables based on the year when the patent applications are filed.

Although we use the application year as the relevant year for our analysis, the patents

appear in the database only after they are granted. Since there is a significant lag between patent

applications and patent grants (about two year on average), the patent database is subject to two

17

types of truncation problems. The first one is regarding patent counts. As we approach the last

few years for which there are patent data available (e.g., 2005 and 2006 in the data used here),

we observe a smaller number of patent applications that are eventually granted. This is because

many patent applications filed during these years were still under review and had not been

granted until 2006. Following Hall, Jaffe, and Trajtenberg (2001, 2005), we correct for the

truncation bias in patent counts using the “weight factors” computed from the application-grant

empirical distribution. The second type of truncation problem is regarding the citation counts.

This is because patents keep receiving citations over a long period of time, but we observe at best

only the citations received up to 2006. Following Hall, Jaffe, and Trajtenberg (2001, 2005), the

truncation in citation counts is corrected by estimating the shape of the citation-lag distribution.

We then use two proxies constructed from the truncation-adjusted NBER patent database to

measure an IPO firm’s innovative activity.

The first measure we employ is the truncation-adjusted patent count for an IPO firm each

year. Specifically, this variable counts the number of patent applications filed in a year that are

eventually granted. However, a simple count of patents may not distinguish breakthrough

innovations from incremental technological discoveries. Therefore, we construct our second

measure that intends to capture the importance of each patent by accounting for the number of

citations each patent receives in subsequent years.

We merge the NBER patent database with our IPO sample. Following the previous

literature (e.g., Atanassov, Nanda, and Seru 2007), we replace the patent and citation count with

zero for IPO firms that have no patent and citation information available from the NBER dataset.

Panel B of Table 3 presents the summary statistics of the innovation variables in the IPO sample.

The observation unit is IPO firm-year. As shown in the table, the distribution of patent grants in

the full IPO sample is very right skewed. Firm-year observations with zero patent represent

roughly 73% of the sample. But this percentage is still significantly lower than that reported in

Atanassov, Nanda, and Seru (2007) (84%) whose sample includes the universe of COMPUSTAT

firms. This suggests that VC-backed IPO firms are on average more innovative than firms

represented by the COMPUSTAT universe. On average, an IPO firm has 3.11 granted patents in

a given year and each patent receives 2.5 citations.

We also report summary statistics for the subsample of firm-year observations with

positive patent counts. This reduces our sample size to 5,264 firm-year observations. Half of the

18

firm-year observations in this subsample have 3 granted patents and the mean is about 11.5

patents. The table also shows that more than a quarter of the firm-year observations with positive

patent counts receive no citation. On average, each patent receives 9.4 citations in this subsample.

Not surprisingly, the distribution of patent counts and that of citations per patent are

highly right skewed. Moreover, from an economic perspective, there is a large difference

between receiving one patent rather than none, but probably a smaller difference between having

101 versus 100 patents. Both the skewness and the likely nonlinearity in the economic impact of

patent counts and citations per patent suggest that a logarithm transformation of these two

measures is appropriate. We then use the natural logarithms of patent counts and citations per

patent as the main innovation measures in our analysis.

4.2 Firm Value and Control Variables

Following the corporate finance literature (e.g., Kaplan and Zingales 1997), we use

Tobin’s Q as our proxy for firm value. Q is measured as a firm’s market value of assets divided

by the book value of assets at each fiscal year end. The market value of assets is defined as the

market value of equity plus the book value of assets minus the book value of equity.

All the financial variables in our analysis are winsorized at the 1st and 99th percentiles to

mitigate the influence of outliers on our results. Details of variable constructions are described in

the appendix table. Panel C of Table 3 reports the summary statistics of IPO firm characteristics.

In our IPO sample, Q ranges from 0.58 to 19.44 with a mean value of 3.01, a median value of

2.08, and a standard deviation of 2.94. The average firm has total book assets of 485.5 million

dollars, sales of 375 million dollars, leverage of 34.6%, and net PPE ratio of 17.36%.

5. FAILURE TOLERANCE AND CORPORATE INNOVATION

5.1 Baseline Results

Table 4 reports the baseline results on how a corporate culture that tolerates early failure

affects a firm’s innovation productivity. The dependent variables are the logarithm of patents in a

year and the logarithm of citations per patent. To facilitate the economic interpretation of our

regression results, we use the logarithm of our failure tolerance measure as our independent

variable so that the regression coefficient estimate gives us the elasticity of innovation to Failure

Tolerance. All regressions include year fixed effect and industry fixed effect, and report

19

coefficient estimates and the Huber-White-Sandwich robust standard errors clustered by IPO

firms (in parentheses).

Model (1) of Table 4 shows that firms with a failure-tolerant culture tend to produce

more patents. The elasticity of patents to failure tolerance is 0.258. This means that a one

percent increase in failure tolerance on average leads to more than a quarter percent increase in

the number of patents in a year. To be more concrete, consider a VC firm at the 25th percentile

of the failure tolerance distribution. According to Table 3 Panel A, this VC firm on average

invests for 1.3 years in a project before acknowledging failure and liquidating it. If this VC firm

is willing to invest for 2.3 years before giving up a project (roughly the 75th percentile of the

failure tolerance distribution), then everything else equal the IPO firms backed by this VC firm

tend to have 20% ( 258.0*3.1

3.13.2 −= ) more patents a year later on.

In model (2) we restrict our analysis to firms with at least one patent during our sample

period (1985-2006) and are thus included in the NBER patent dataset.10 We expect the effect of

failure tolerance to be stronger in the subsample of firms for which innovation is absolutely

relevant. This is exactly what we find. The elasticity of patents to failure tolerance increases to

0.422 in this subsample and is even more statistically significant.

Models (3) and (4) of Table 4 show that firms with a failure-tolerant culture also tend to

produce patents of higher quality (measured by the number of citations per patent). Model (3)

shows that a one percent increase in failure tolerance on average leads to a 0.2 percent increase

in citations per patent. Again, the effect of failure tolerance is much stronger in the subsample of

firms with nonzero patents. In un-tabulated regressions, we also exclude self-citations when

computing citations per patent. Our results are robust to such modification.11

As we have discussed in Section 3, by construction our failure tolerance measure is

determined long before an entrepreneurial firm starts to produce patents. Therefore, the effect of

failure tolerance on innovation we document here is not subject to any reverse causality issue.

We control for a comprehensive set of firm characteristics that may affect a firm’s

10 The number of observations in Table 4 model 2 is 7,607, while it is 5,264 in Table 3 Panel B for the subsample with patents>0. The discrepancy is due to a difference in the definition of subsample with patents>0. In Table 4 model 2, “Patents>0” means that the firm has at least one patent over the entire sample period (but not necessarily in each year). In Table 3 Panel B, “Patents>0” means that the patent count is nonzero for a firm-year observation. 11 For example, the coefficient estimate of Ln(Failure Tolerance) is 0.238 (p-value=0.03) in model 3 of Table 4 when the natural logarithm of the modified citations per patent is the dependent variable.

20

innovation productivity. We find that firms that are larger (higher sales), more profitable (higher

ROA), have more growth potential (higher Q) and lower exposure to financial distress (lower

leverage) are more innovative. A larger R&D spending, which can be viewed as a larger

innovation input, is associated with more innovation output. Higher investment (higher capital

expenditures) is also associated with higher innovation productivity. Further, higher institutional

ownership is associated with more innovation, which is consistent with the findings in Aghion,

Van Reenen and Zingales (2009). Finally, firm age, asset tangibility (measured by the net PPE

over assets), and industry competition (measured by the Herfindahl index) do not significantly

impact a firm’s innovation productivity.

Overall, our baseline results suggest that a corporate culture that is tolerant of early

failure can increase a firm’s innovation productivity. These results provide support for the

implications of Holmstrom (1989) and Manso (2008) that tolerance for failure is critical in

motivating innovation.

5.2 Robustness

5.2.1 Alternative Econometric Specifications

We conduct a comprehensive set of robustness tests for our baseline results. The first set

of robustness is on alternative econometric specifications. Besides the pooled OLS specification

reported in Table 4, we have used the Fama-MacBeth regression adjusting for auto-correlations

of coefficient estimates. We have also used a Tobit regression that takes into consideration the

non-negative nature of patent data and citation data. We have run a Poisson regression when the

dependent variable is the number of patents to take care of the discrete nature of patent counts.

We have also run a firm random-effect model, which is generally more efficient than a pooled

OLS regression (Wooldridge 2009, page 493). Our baseline results are robust in all the above

alternative models. We thus do not report these additional results.

5.2.2 Controlling for Other VC Firm Characteristics

The second set of tests is to check whether the failure tolerance effect is robust to

controlling for other characteristics of an IPO firm’s VC investors. Table 2 shows that VC

investors that are more experienced are more tolerant of early failure in projects. VC firms’

portfolio concentration and local competition also make them more tolerant of failures. Does our

21

failure tolerance measure simply reflect these characteristics? In other words, can other VC firm

characteristics be the omitted variables in our baseline regressions? The answer is no.

We first control for the observable VC firm characteristics that are correlated with the VC

firm’s failure tolerance as indicated in Table 2. As discussed in Section 3.4, for each IPO firm in

our sample we compute the weighted average characteristics of its VC firms at their first

investment in the firm. These VC characteristics are parallel to our failure tolerance measure by

construction. To save space, we only report results for the key explanatory variables in the table.

Table 5 Panel A shows that the failure tolerance effect on a firm’s innovation

productivity is robust to controlling for observable VC firm characteristics. If more experienced

VCs are more tolerant of failure, and at the same time are better at selecting more innovative

entrepreneurial firms, then we would expect the effect of VC experience to be positive and

significant and the effect of failure tolerance to disappear or substantially weaken. But we find

that failure tolerance still has a positive and significant effect on patent generation and patent

quality in all models. The magnitude of the effect also remains stable. For example, the average

elasticity of patents to failure tolerance in the four models is 0.264, which is comparable to the

magnitude in model (1) of Table 4 (0.258) in which we do not control for the other VC firm

characteristics. In addition, after controlling for the failure tolerance effect, none of the other VC

firm characteristics significantly affect the innovation productivity of the IPO firms.

Another possibility is that the omitted variables are unobservable VC firm characteristics.

Including a VC firm fixed effect can absorb the effect of any time-invariant unobservable VC

firm characteristics. However, controlling for VC firm fixed effect in our study is tricky because

86% our IPO firms are financed by VC syndicates rather than a single VC firm. We therefore

control for lead VC firm fixed effects as well as observable VC firm characteristics in Table 5

Panel B. We define the lead VC firm as the one that invests the most in an IPO firm.12 We find

that the failure tolerance effect is still positive and significant. The average estimate for the

elasticity of patents to failure tolerance is 0.248 in Panel B, which is comparable to the

magnitude in Panel A where we just control for observable VC firm characteristics.

Taken together, the results in Table 5 suggest that the effect of failure tolerance on

innovation documented in Table 4 does not simply reflect the effects of observable time-varying

12 In our IPO sample, about 60% of the lead VC firms (based on our definition) had participated since the very first VC financing round received by the IPO firm, and about 90% of the lead VC firms started their investment in the first three VC financing rounds received by the IPO firm.

22

VC firm characteristics or unobservable time-invariant VC characteristics. Even though these

characteristics affect the VC firms’ attitudes towards early failure as shown in Table 2, they

generally do not have any significant and direct effect on innovation.

5.2.3 Persistence of the Failure Tolerance Effect

Our last set of robustness tests is related to the persistence of the failure tolerance effect.

Economic theories suggest that a corporate culture, once formed, can persist over time (see e.g.,

Lazear 1995, Akerlof and Kranton 2000, 2005). This implies that a corporate culture should have

a persistent effect on corporate decisions and performances. Our empirical approach to the effect

of corporate culture hinges on the idea that a VC firm’s attitude towards early failure has a

profound impact on the formation of a failure-tolerant culture in the entrepreneurial firms backed

by the VC firm.

However, VC investors do not stay forever in the IPO firms they invest in. Existing

studies show that VC investors on average cash out about 70% of their investment in an IPO firm

within two years after the IPO (see, e.g., Gompers and Lerner 1998). If the VC investors’

tolerance for failure is not internalized into the corporate culture of the entrepreneurial firm, then

we should expect the effect of failure tolerance to wane after the VC investors exit their

investment. In this case, the failure tolerance effect we have documented simply reflects a

transitory VC investor influence, not the effect of a corporate culture. As Hermalin (2001) puts it,

corporate culture “resides with the firm, not an individual”.

Our full sample panel regression analysis includes innovations generated long after a

firm’s IPO and after the exit of venture capital investment.13 In the Fama-Macbeth approach, we

run year by year regressions and then take the average failure tolerance effect across years. This

approach tells us whether the failure tolerance effect is stable over time. Both approaches give us

a significant failure tolerance effect during the entire sample period, and thus provide some

support for a persistent cultural effect.

To further check the persistence of the failure tolerance effect as an IPO firm ages, in

Table 6 we restrict the sample to firms that existed for at least seven years after their IPOs. We

then examine how the failure tolerance effect on innovation evolves within the first seven years

13 In our sample we have nonzero patent and citation data for 1,513 firms that went public between 1985 and 2006. We have nonzero patent and citation data for 693 of these firms more than four years after their IPOs and up to 22 years after IPO.

23

after these firms’ IPOs. 14 The sample restriction mitigates the survivorship bias when we

compare the effect over time.

Models (1)–(3) of Panel A report the average failure tolerance effect on patent counts for

the same set of firms in the first two years, five years, and seven years after their IPOs,

respectively. Panel B reports the results for the number of citations per patent. We find that the

failure tolerance effects on both patent counts and patent quality are positive and significant in

all three windows. This suggests that the effect persists for at least seven years after IPO and five

years after the bulk of the venture capital investment exits.

It is true that there is some attenuation in the magnitude of the failure tolerance effect

over time. In both panels the effect is the strongest in the first two years, and then it gradually

declines as we extend the window. However, the differences across the three windows are not

statistically significant, as shown in the bottom row of each panel. Also, some attenuation can be

expected and can still be consistent with a persistent cultural effect. As a firm matures over time,

its operational focus may gradually shift from innovation to standard mass production.

Overall, our analysis suggests that the failure tolerance effect on firm innovation persists

after VC firms exit the IPO firms. This implies that the failure tolerance effect we document does

not simply reflect a transitory influence of VC firms. VC firms’ attitude towards failure seems to

be internalized by the IPO firms they invest in.

5.3 Difficulty of Innovation and Effect of Failure Tolerance

Failure tolerance is critical in motivating innovation because exploration of new methods

often involves a large chance of early failure compared to exploitation of well-known methods.

This insight implies that in industries in which innovation is more difficult to achieve (i.e., the

chance of failure is high), having a corporate culture that tolerates failure is more important for

innovation productivity.

Different types of patents can involve different degrees of difficulty. Following the work

of Hall, Jaffe, and Trajtemberg (2005), we classify patents in our sample into four categories

based on the nature of the patents: (1) drugs, medical instrumentation, and chemicals; (2)

14 We choose seven years as our cutoff for two reasons. First, if VC firms largely cash out of the IPO firms within two years after the IPOs as shown in Gompers and Lerner (1998), then we still have at least five years to examine the persistence of the failure tolerance effect. Second, this cutoff leaves us with a good sample size for the analysis. We have used other cutoffs such as five years, eight years and ten years, and results are similar.

24

computers, communications, and electrical; (3) software programming; (4) other miscellaneous

patents.15 If a firm has no patent, then we classify it into one of the above four categories based

on the most frequent type of patents produced by the firm’s 3-digit SIC industry. For example, if

a firm in the industry with 3-digit SIC 283 has no patent in our sample period, then it is classified

under category (1) because 77% of the patents generated by the firm’s industry are related to

drugs and chemicals.

Common sense suggests that among the above four categories patents of new drugs are

probably the most difficult to produce. A new drug development process involves many steps

requiring different levels of experimentation. Existing studies suggest that the cost of developing

a new drug varies from $500 million to $2 billion (see, e.g., Adams and Brantner 2006). Hall,

Jaffe, and Trajtemberg (2005) also show that the market value impact of drug patents is much

higher than that of all other types of patents. Thus we expect that having a failure-tolerant culture

is most important in industries producing new drugs.

Table 7 reports the baseline regressions within each patent category. Panel A shows that

in each category a failure-tolerant culture has a significantly positive effect on patent generation.

But as we have expected, the effect of failure tolerance is the strongest in industries producing

drugs, medical instrumentation, and chemical patents. The elasticity of patents to failure

tolerance is 0.743 in category (1), almost triples the effect in the computers and electrical

category (0.255), and almost quadruples the effect in the software programming category (0.190).

As shown in the bottom row of Panel A, the differences in the failure tolerance effect between

category (1) and other categories are highly statistically significant. Table 7 Panel B shows

similar results for citations per patent. Failure tolerance is most important for patent quality in

industries producing drugs-related patents.

In sum, the results in Table 7 suggest that having a more failure-tolerant corporate culture

is more important for innovation productivity when innovation is more difficult to achieve. This

cross-sectional comparison provides further support for our empirical measure of failure

15 Hall, Jaffe, and Trajtemberg (2005) have six categories: chemicals, drugs and medical instrumentation, computers and communications, electrical, metals and machinery, and miscellaneous. We group chemicals with drugs because we only have a few observations of chemical patents. Software programming patents (computer-related patents generated by the 3-digit SIC industry 737) belong to the computers and communications category. For finer comparisons between different types of patents, we single out software programming. We then group patents related to computer hardware, communications, and electrics together. Finally, we group metals, machinery and miscellaneous together because we do not have many observations of these patents and label this category as miscellaneous patents.

25

tolerance and the recent theories on motivating innovation.

6. STANDARD INCENTIVE SCHEME AND FAILURE TOLERANCE

Economic theories suggest that although the standard incentive scheme such as pay-for-

performance is effective at motivating effort on standard tasks, it can fail to motivate innovation

(see, e.g., Manso 2008). In Table 8 we contrast the role of the standard incentive scheme with

that of failure tolerance in motivating innovation.

In the classical corporate finance literature (e.g., Holmstrom 1982), insider equity

ownership plays an important role in the motivation of efforts. Thus we use insider equity

ownership as our proxy for the standard incentive scheme. Compact Disclosure database

provides annual data of equity ownership of executive officers and directors for a large fraction

of our IPO firms in our sample period.16 Also, insiders generally still hold significant amount of

their firms’ equity at IPO and for some years after IPO. The average insider ownership in the

first two years after IPO in our sample is 25%. This implies that equity ownership should

account for the bulk part of the total equity incentives provided to executive officers and

directors. To facilitate the interpretation of the regression results, we use the logarithm of insider

ownership as our independent variable.

Model (1) of Table 8 shows that insider ownership is negatively related to the number of

patents generated by a firm. The elasticity of patent counts to insider ownership is −0.073 and is

statistically significant. 17 In model (2) we include our failure tolerance measure. Insider

ownership still has a significantly negative association with innovation, while failure tolerance

has a significantly positive effect as we have shown before.

In models (3)–(4) we find similar contrasting effects of insider ownership and failure

tolerance on patent quality. While a failure-tolerant corporate culture contributes to high-impact

16 Publicly traded companies in the United States are required to report equity ownership of executive officers and directors in their annual proxy statements. The equity ownership includes equity shares held by officers and directors, underlying shares in their vested stock options, and underlying shares in their stock options exercisable within 60 days of the reporting date. Annual compensation data for most of our IPO firms is not available. Thus we cannot include the full incentive effect of stock options. However, we believe that the insider ownership data should capture the bulk part of total equity incentives provided to executive officers and directors. 17 In an unreported regression we add a firm-fixed effect in model 1 of Table 8. We find that Ln(Insider Ownership) is still negatively and significantly related to Ln(Patents). The elasticity of patents to insider ownership is -0.038 (p-value=0.012).

26

patents, insider equity ownership fails to do so.18

In unreported regressions we also examine whether the insider ownership effect on

innovation is nonlinear. When using a quadratic specification for insider ownership, we find no

nonlinearity in the effect on patent counts and a convex effect on citations per patent (but not a

U-shape effect). When using a spline regression with cutoffs being 5% and 25% (Morck, Shleifer

and Vishny 1988), we again find no nonlinearity in the effect on patent counts. The effect of

insider ownership on citations per patent is significantly negative when insider ownership is

between 5% and 25%, and is positive and insignificant in the lowest and the highest regions. In

both specifications, the effect of failure tolerance is always positive and significant.

In another unreported robustness test, we restrict our analysis to the period before 1993.

Stock options were not as commonly used back then, and insider equity ownership should be a

more precise measure of executives’ total equity incentives. We find that insider ownership is

still negatively related to patent generation (coefficient estimate= −0.091, p-value=0.04) and

citations per patent (coefficient estimate= −0.07, p-value=0.21).

Taken together, our findings in Table 8 are consistent with the implications in existing

economic theories that while tolerance for failure is critical in encouraging innovation, standard

incentive scheme such as insider equity ownership can fail to spur innovation. Clearly, insider

ownership is not as exogenous as our failure tolerance measure. Thus we do not claim any causal

effect of insider ownership on a firm’s innovation productivity.

7. FAILURE TOLERANCE AND FIRM VALUE

In Sections 5 and 6 we show that a failure-tolerant corporate culture motivates corporate

innovation. Does failure tolerance increase firm value? Is such a culture generally good for a

firm no matter whether innovation is important for the firm?

In Table 9 we examine the effect of failure tolerance on firm value measured by Tobin’s

Q. We measure Q at the first fiscal year end after IPO. To be consistent in our economic

18 Lerner and Wulf (2006) examine the relationship between the average long-term incentives (the sum of restricted stock grants and stock option grants as a fraction of total expected compensation) given to corporate R&D heads and the average innovation productivity of the firm (measured by the average patents granted, the average citations received, and the originality of patents) in a cross-section of 278 large firms. They find a positive relationship between long-term incentives given to R&D heads and innovation in 177 firms with centralized R&D organization, but long-term incentives given to CEOs, CFOs, and human resource heads in those firms are negatively and insignificantly related to innovation. Our insider equity ownership data is more likely to capture equity incentives of CEOs and CFOs, and thus our results are consistent with their findings.

27

interpretation, we use the natural logarithm of Tobin’s Q as the dependent variable and the

natural logarithm of failure tolerance as the independent variable. To see whether failure

tolerance is more important for firms in more innovative industries, we calculate two industry

level measures. “Ind. Total Patents” is the total number of patents in the entire sample period by

4-digit SIC industries. “Ind. Total Citations” is the total number of citations in the entire sample

period by 4-digit SIC industries. A higher value of Ind. Total Patents or Ind. Total Citations

indicates a more innovative industry.

Model (1) of Table 9 shows that after controlling for other firm characteristics, firms with

a failure-tolerant culture have weakly higher firm value. The elasticity of firm value to failure

tolerance is 0.081 and is marginally significant.

Next, we allow the effect of failure tolerance on firm value to vary with the importance of

innovation in an industry. We expect investors to value failure tolerance more in industries in

which innovation is more pertinent. Thus in model (2) we include the interaction between failure

tolerance and the logarithm of Ind. Total Patents. The direct effect of failure tolerance becomes

negative and insignificant (−0.077). But the interaction effect is positive and significant (0.027).

This implies that failure tolerance does not increase firm value in less innovative industries. The

effect of failure tolerance on firm value increases with the importance of innovation in an

industry. In model (3) we find similar effects when using Ind. Total Citations to identify more

innovative industries.

To see the economic impact of the interaction effect on firm value, let us consider one of

the drugs industries–the 4-digit SIC industry of 2834. The total patent count in this industry in

our sample period is 1,124. Model (2) implies that the elasticity of Q to failure tolerance in this

industry is 0.19 (= 0.027×Ln(1,124)). An average firm in this industry has book assets of $143

million and Tobin’s Q of 4.4. Thus a ten percent increase in failure tolerance can increase the

market value of the average firm by about $20 million (=10%×0.19%×143×4.4).19

In addition, the direct effect of Ln(Ind. total Patents) and Ln(Ind. Total Citations) are

significantly positive (0.044 in model (2) and 0.043 in model (3)), implying that firm value is on

average higher in more innovative industries. This is consistent with the findings in the existing

literature (see, e.g., Hall 2000). 19 In this calculation we treat the coefficient estimate of Ln(Failure Tolerance) as zero because it is insignificantly different from zero. If we treat it as it is (-0.077), then the market value impact of a 10% increase in failure tolerance for the average firm is $7 million.

28

Overall, our analysis suggests that a failure-tolerant corporate culture increases firm value

through its impact on innovation. Failure tolerance increases a firm’s innovation productivity,

which in turn increases firm value. This link is more pertinent in industries where innovation is

more important.

8. CONCLUSION

In this paper, we examine whether tolerance for failure, an important aspect of corporate

culture, spurs corporate innovation and increases firm value based on a sample of VC-backed

IPO firms between 1985 and 2006. We develop a novel firm-specific measure of a failure-

tolerant corporate culture. Our key idea is that the formation of corporate culture is largely

determined by the attitudes and beliefs of its first-generation insiders, namely, founders and early

active investors. VC investors are such investors. Thus their attitudes towards failures should

have a profound impact on the formation of a failure-tolerant culture in the entrepreneurial firms

they invest in. In other words, the entrepreneurial firms can “inherit” or internalize their VC

investors’ failure tolerance.

Our firm-level failure tolerance measure allows us to directly test the key implications in

the recent theoretical literature on motivating innovation. Our empirical analysis generates a

number of interesting new findings. First, firms with a more failure-tolerance culture are

significantly more innovative. The failure tolerance effect is persistent and robust to including

other VC firm characteristics such as VC firm’s experience and industry expertise and lead VC

firm fixed effects. Second, we examine the cross-sectional effect of failure tolerance and find

that its effect on firm innovation is stronger in industries in which innovation is more difficult to

achieve. Third, we contrast the effect of standard incentive schemes such as insider equity

ownership with the effect of a failure-tolerant corporate culture in motivating innovation. We

find that while failure tolerance spurs innovation, large insider equity ownership is negatively

related to a firm’s innovation productivity. Finally, we find that a failure-tolerant corporate

culture increases firm value in industries in which innovation is important. Overall, our findings

are consistent with the theory implications in the corporate innovation literature that tolerance for

failure can motivate corporate innovation and increase firm value.

29

Our paper also contributes to the literature on corporate culture. This literature has

remained mostly theoretical because empirically identifying the corporate culture effect is very

challenging. Our paper is the first attempt at measuring a specific aspect of corporate culture, and

our work provides empirical evidence that corporate culture does matter in significant ways for

firms’ decisions and performances.

30

REFERENCES

Acharya, Viral V., and Krishnamurthy Subramanian, 2009, “Bankruptcy Codes and Innovation,” Review of Financial Studies, forthcoming.

Acharya, Viral V., Ramin Baghai-Wadji, and Krishnamurthy Subramanian, 2009, “Labor Laws

and Innovation,” Working Paper, London Business School. Adams, Christopher P., and Van V. Brantner, 2006, “Estimating the Cost of New Drug

Development: Is It Really 802 Million Dollars?” Health Aff (Millwood) Vol. 25, 420–428. Aghion, Philippe, and Jean Tirole, 1994, “The Management of Innovation,” Quarterly Journal of

Economics Vol. 109, 1185-1209. Aghion, Philippe, John Van Reenen, and Luigi Zingales, 2009, “Innovation and Institutional

Ownership,” Working Paper, Harvard University. Akerlof, George A., and Rachel E. Kranton, 2000, “Economics and Identity,” Quarterly Journal

of Economics Vol. 115, 715-753. Akerlof, George A., and Rachel E. Kranton, 2005, “Identity and the Economics of

Organizations,” Journal of Economic Perspectives Vol. 19, 9-32. Armour, John, and Douglas J. Cumming, 2008, “Bankruptcy Law and Entrepreneurship”,

American Law and Economics Review Vol. 10, 303-350. Atanassov, Julian, Vikram Nanda, and Amit Seru, 2007, “Finance and Innovation: The Case of

Publicly Traded Firms,” Working Paper, University of Chicago. Ayyagari, Meghana, Asli Demirgüc-Kunt, and Vojislav Maksimovic, 2007, “Firm Innovation in

Emerging Markets,” Working Paper, George Washington University. Bayar, Onur, and Thomas Chemmanur, 2008. “IPOs or Acquisitions? A Theoretical and

Empirical Analysis of the Choice of Exit Strategy by Entrepreneurs and Venture Capitalists,” Working Paper, Boston College.

Belenzon, Sharon, and Tomer Berkovitz, 2007, “Innovation in Business Groups,” Working Paper,

Oxford University. Bernhardt, Dan, Eric Hughson, and Edward Kutsoati, 2006, “The Evolution of Managerial

Expertise: How Corporate Culture Can Run Amok,” American Economic Review, Vol. 96, 195-221

Bottazzi, Laura, Marco Da Rin, and Thomas Hellmann, 2007, “The Importance of Trust for

Investment: Evidence from Venture Capital,” Working Paper, University of British Columbia

31

Bottazzi, Laura, Marco Da Rin, and Thomas Hellmann, 2008, “Who are the Active Investors?

Evidence from Venture Capital,” Journal of Financial Economics, Vol. 89, 488-812 Brau, James C, Bill Francis, and Ninon Kohers, 2002, “The Choice of IPO versus Takeover:

Empirical Evidence,” Journal of Business, Vol. 76, 583-612 Chemmanur, Thomas, and Elena Loutskina, 2006, “The Role of Venture Capital Backing in

Initial Public Offerings: Certification, Screening, or Market Power?” Working Paper, Boston College

Cronqvist, Henrik, Angie Low, and Mattias Nilsson, 2009, “Persistence in Firm Policies, Firm

Origin, and Corporate Culture: Evidence from Corporate Spin-offs,” Working Paper, Claremont McKenna College.

Cremer, Jacques, 1993, “Corporate Culture and Shared Knowledge,” Industrial and Corporate

Change Vol. 2, 351-386 Ederer, Florian, and Gustavo Manso, 2009, “Is Pay-for-Performance Detrimental to Innovation?”

Working Paper, MIT. Gompers, Paul, 1995, “Optimal Investments, Monitoring, and the Staging of Venture Capital,”

Journal of Finance, Vol. 50, 1461-1489 Gompers, Paul, 1996, “Grandstanding in the Venture Capital Industry,” Journal of Financial

Economics, Vol. 43, 133-156 Gompers, Paul, and Josh Lerner, 1998, “Venture Capital Distributions: Short-run and Long-run

Reactions,” Journal of Finance Vol. 53, 2161-2183. Gompers, Paul, and Josh Lerner, 2000, “Money Chasing Deals? The Impact of Fund Inflows on

Private Equity Valuation,” Journal of Financial Economics, Vol. 55, 281–325. Gompers, Paul, and Josh Lerner, 2004, The Venture Capital Cycle, 2nd Edition, MIT Press. Griffin, Dale, Kai Li, Heng Yue, and Longkai Zhao, 2009, “Culture Values and Corporate Risk-

Taking”, Working Paper, University of British Columbia. Griliches, Zvi, Bronwyn Hall, and Ariel Pakes, 1987, “The Value of Patents as Indicators of

Inventive Activity”, In Dasgupta and Stoneman (eds.), Economic Policy and Technological Performance, Cambridge: Cambridge University Press

Guiso, Luigi, Paola Sapienza, and Luigi Zingales, 2006, “Does Culture Affect Economic

Outcomes?” Journal of Economics Perspectives, Vol. 20, 23-48

32

Guiso, Luigi, Paola Sapienza, and Luigi Zingales, 2009, “Cultural Biases in Economic Exchange,” Quarterly Journal Economics forthcoming.

Hall, Bronwyn, 2000, “Innovation and Market Value”, in Barrell, Ray, Geoffrey Mason, and

Mary O'Mahoney (eds.), Productivity, Innovation and Economic Performance, Cambridge: Cambridge University Press.

Hall, Bronwyn, Adam Jaffe, and Mannuel Trajtenberg, 2001, “The NBER Patent Citations Data

File: Lessons, Insights and Methodological Tools,” NBER Working Paper No. 8498 Hall, Bronwyn, Adam Jaffe, and Mannuel Trajtenberg, 2005, “Market Value and Patent

Citations”, Rand Journal of Economics, Vol. 36, 16-38 Hellmann, Thomas, and Manju Puri, 2000, “The Interaction between Product Market and

Financing Strategy: The Role of Venture Capital,” Review of Financial Studies, Vol. 13, 959-984.

Hellmann, Thomas, and Manju Puri, 2002, “Venture Capital and Professionalization of Start-Up

Firms: Empirical Evidence,” Journal of Finance, Vol. 57, 169-197. Hellmann, Thomas, and Veikko Thiele, 2008. “Incentives and Innovation: A Multi-tasking

Approach”, Working Paper, University of British Columbia. Hermalin, Benjamin E., 2001, “Economics and Corporate Culture,” in Cary L. Cooper, Sue

Cartwright, and P. Christopher Earley, eds.: The International Handbook of Organizational Culture and Climate (John Wiley & Sons, Chichester, England).

Hochberg, Yael, Alexander Ljungqvist, and Yang Lu, 2007, “Venture Capital Networks and

Investment Performance,” Journal of Finance, Vol. 62, 251-301 Holmstrom, Bengt, 1982, “Moral Hazard in Teams,” Bell Journal of Economics, Vol. 13, 324-

340 Holmstrom, Bengt, 1989, “Agency Costs and Innovation,” Journal of Economic Behavior and

Organization, Vol. 12, 305-327. Kacperczyk, Marcin T, Clemens Sialm, and Lu Zheng, 2005, “On the Industry Concentration of

Actively Managed Equity Mutual Funds,” Journal of Finance, Vol. 60, 1983-2011. Kaplan, Steven, and Zingales, Luigi, 1997, “Do Investment-Cash Flow Sensitivities Provide

Useful Measures of Financing Constraints?” Quarterly Journal of Economics, Vol. 112, 169-215

Kotter, John, and James Heskett, 1992, “Corporate Culture and Performance”, The Free Press.

33

Kreps, David M., 1990, “Corporate Culture and Economic Theory,” in James E. Alt, and Kenneth A. Shepsle, eds: Perspectives on positive political economy (Cambridge University Press, Cambridge, England)

Lazear. Edward P., 1995, “Corporate Culture and the Diffusion of Values,” in Horst Siebert, ed.:

Trends in business organization: Do participation and cooperating increase competitiveness? (Institutfur Weltwirtschaft an der Universitat Kiel, J.C. B. Mohr/Paul Siebeck, Tubingen, Germany)

Lemmon, Michael L, Michael R. Roberts, and Jamie F. Zender, 2008, “Back to the Beginning:

Persistence and the Cross-Section of Corporate Capital Structure,” Journal of Finance, Vol. 63, 1575-1608.

Lerner, Josh, 1994, “Venture Capital and the Decision to Go Public,” Journal of Financial

Economics, Vol. 35, 293-316 Lerner, Josh, 1995, “Venture Capitalists and the Oversight of Private Firms,” Journal of Finance,

Vol. 50, 301-318 Lerner, Josh, 2006, “The New New Financial Thing: The Origins of Financial Innovations,”

Journal of Financial Economics, Vol. 79, 223-255. Lerner, Josh, and Julie Wulf, 2007, “Innovation and Incentives: Evidence from Corporate R&D,”

Review of Economics and Statistics 89, 634-644. Manso, Gustavo, 2008, “Motivating Innovation,” Working Paper, MIT Sloan School of

Management. Morck, Randall K, Shleifer, Andrei, and Vishny, Robert, 1988 “Management Ownership and

Market Valuation: An Empirical Analysis,” Journal of Financial Economics Vol. 20, 293-315.

Ostroff, Cheri, Angelo J. Kinicki, and Melinda M. Tamkins, 2003, “Organizational Culture and

Climate,” in W. Borman and D. Ilgen (eds.) Handbook of Psychology: Industrial and Organizational Psychology, 12: 565-93.

Sahlman, William A, 1990, “The Structure and Governance of Venture Capital Organizations,”

Journal of Financial Economics, Vol. 27, 473–521. Sapra, Haresh, Ajay Subramanian, and Krishnamurthy Subramanian, 2009, “Corporate

Governance and Innovation: Theory and Evidence,” Working Paper, University of Chicago. Schein, Edgar H., 2004, “Organizational Culture and Leadership,” 3rd Edition, Jossey-Bass. Sorensen, Morten, 2007, “How Smart is Smart Money? A Two-Sided Matching Model of

Venture Capital,” Journal of Finance, Vol. 62, 2725-2762

34

Stulz, Rene M., and Rohan Williamson, 2003, “Culture, Openness, and Finance,” Journal of

Financial Economics, Vol. 70, 313-349 Van den Steen, Eric, 2005a, “On the Origin of Shared Beliefs (and Corporate Culture),” Working

Paper, MIT Sloan School of Management Van den Steen, Eric, 2005b, “Organizational Beliefs and Managerial Vision,” Journal of Law,

Economics, and Organization, Vol. 21 256-283 Wooldridge, Jeffrey M., 2009 “Introductory Econometrics: A Modern Approach,” 4th edition,

South-Western Publishing.

35

Appendix: Variable Definitions and Data Sources

Failure Tolerance and Other VC Characteristics

Failure Toleranceit The average number of years VC firm i invested in its projects that were initiated in or after year 1980 and eventually failed in or before year t (Source: Venture Economics)

Past Amount Investedit

The total dollar amount invested by VC firm i since 1980 up to year t (Source: Venture Economics)

Past Firms Investedit The total number of firms VC firm i has invested in since 1980 up to year t (Source: Venture Economics)

Past Fund Raisedit The total dollar amount raised by VC firm i since 1965 up to year t (Source: Venture Economics)

VC Ageit Age of VC firm i in year t measured as the number of years since its date of inception (Source: Venture Economics)

Investment Concentrationit

Investment concentration index of VC firm i in year t measured as the sum of the squared deviations of the weights (the number of portfolio firms) for each of the 18 different industries held by the VC firm i relative to the industry weights of the total venture investment (Source: Venture Economics)

Past Successful Exitit

The proportion of entrepreneurial firms financed by VC firm i that either went public or were acquired between year 1980 and year t (Source: Venture Economics)

Past IPO Exitit The proportion of entrepreneurial firms financed by VC firm i that went public between year 1980 and year t (Source: Venture Economics)

Local Competitionit The number of VC firms located within the same zip code of VC firm i in year t (Source: Venture Economics)

Industry Fund Inflowt

The total dollar amount raised by all VC firms in year t (Source: Venture Economics)

Innovation Variables

Patentit Number of patents firm i applied for in year t. Only patents that were later granted are included. The variable is also corrected for the truncation bias in patents granted towards the end of the sample using the “weight factors” provided by Hall, Jaffe, and Trajtenberg (2001, 2005). The weight factors are computed from an application-grant empirical distribution. (Source: NBER Patent Data)

36

Citations/Patent it The average number of citations per patent of firm i applied for in year t (Source: NBER Patent Data)

IPO Firm Characteristics

Tobin’s Q it Market to book ratio of firm i in year t: (total assets + year end closing price*year end outstanding shares - book equity)/total assets (Source: Compustat)

Sales it Sales by firm i in year t (in $million) (Source: Compustat)

Assets it Total assets of firm i in year t (in $million) (Source: Compustat)

ROA it Operating income before depreciation to total assets ratio of firm i in year t (Source: Compustat)

R&D/Assets it Research and Development expenditure to total assets ratio of firm i in year t (Source: Compustat)

CapExp/Assets it Capital expenditure to total assets ratio of firm i in year t (Source: Compustat)

Leverage it Total debt of firm i in year t divided by its total assets (Source: Compustat)

Firm Age it Age of firm i in year t since its IPO (Source: SDC)

PPE/Asset it Net property, plants and equipments to assets ratio of firm i in year t (Source: Compustat)

Institutional Ownership it

Total percentage of firm i’s equity held by institutional investors in year t (Source: Thomson Financial 13f institutional holdings database)

Insider Ownershipit Total percentage of firm i’s equity held by officers and board of directors in year t (Source: Compact Disclosure)

Herfindahl Index it Herfindahl index of firm i’s industry in year t constructed based on sales at 4-digit SIC industries (Source: Compustat)

37

Table 1: Summary Statistics for the VC Sample

This table reports the summary statistics for variables constructed based on our venture capital sample. The VC sample contains investments made by 2,857 VC firms from year 1980 to 2006. All the variables listed below are time varying characteristics of a VC firm or the conditions of the entire VC industry.

Variable 25% Median Mean 75% Std. Dev. N Failure Tolerance 0.72 1.23 1.41 1.89 0.97 18,993 Other VC Characteristics Past Amount Invested (mil.) 11.25 58.40 394.89 263.15 1298.07 35,662 Past Firms Invested 3.00 8.00 23.98 23.00 50.14 35,662 Past Fund Raised (mil.) 27.00 81.20 234.39 219.40 558.41 16,128 VC Age 2.00 5.00 7.51 11.00 8.04 28,789 Investment Concentration 0.08 0.28 0.37 0.72 0.33 34,270 Past Successful Exit (%) 25.00 60.00 52.82 77.76 34.41 35,689 Past IPO Exit (%) 0 8.33 16.18 24.05 22.74 35,689 Local Competition 1.00 2.00 7.40 7.00 12.37 27,256 VC Industry Conditions Industry Fund Inflow (bil.) 8.37 32.41 41.93 57.03 41.93 35,689

38

Table 2: Failure Tolerance and VC Characteristics The dependent variable is natural logarithm of a VC firm’s “Failure Tolerance” in a given year. Portfolio firm industry fixed effects refer to the industry a VC firm invests in. If a VC firm invests in multiple industries in a given year, we choose the industry in which the VC firm invests the largest amount of capital in that year for the industry fixed effect. Coefficient estimates and the Huber-White-Sandwich robust standard errors clustered by VC firm (in parentheses) are reported. ***, ** and * indicate significance at 1%, 5% and 10% levels respectively.

(1) (2) (3) (4) Ln(Past Amount Invested) 0.034*** (0.007) Ln(Past Firms Invested) 0.023* (0.012) Ln(Past Fund Raised) 0.020*** (0.004) Ln(VC Age) -0.002 (0.011) Investment Concentration 0.033*** 0.026*** 0.025*** 0.019** (0.008) (0.008) (0.008) (0.008) Past Successful Exit (%) -0.004 0.007 0.009 0.014 (0.011) (0.011) (0.011) (0.011) Local Competition 0.023*** 0.026*** 0.024*** 0.028*** (0.009) (0.009) (0.009) (0.009) Ln(Industry Fund Inflow 1 year prior) -0.016*** -0.002 -0.003 -0.030*** (0.004) (0.006) (0.006) (0.004) Ln(Industry Fund Inflow 2 years prior) 0.017 0.023** 0.022*** 0.061*** (0.016) (0.009) (0.008) (0.015) Constant 0.680*** 0.551*** 0.584*** 0.535*** (0.137) (0.154) (0.148) (0.138) VC firm fixed effects Yes Yes Yes Yes Portfolio firm industry fixed effects Yes Yes Yes Yes Year fixed effects Yes Yes Yes Yes Observations 18,159 18,159 18,159 18,159 R2 0.151 0.144 0.150 0.143

39

Table 3: Summary Statistics for the IPO Sample

Panel A: Failure Tolerance and Other VC Characteristics

Variable 25% Median Mean 75% Std. Dev. N Failure Tolerance 1.32 1.76 1.80 2.26 0.69 1,848 Past Amount invested (mil.) 123.92 402.93 819.71 882.87 1595.73 1,848 Past Firms invested 17.62 37.65 56.60 70.97 64.18 1,848 Past Fund Raised (mil.) 53.77 167.95 425.65 398.64 1047.76 1,848 VC Age 5.81 10.54 11.20 15.47 7.38 1,848 Investment Concentration 0.04 0.10 0.14 0.17 0.15 1,848 Past Successful Exit (%) 63.26 70.24 64.71 75.55 21.33 1,848 Past IPO Exit (%) 17.85 23.81 24.38 30.16 13.06 1,848 Local Competition 2.18 4.76 6.96 9.00 7.28 1,848

Panel B: Innovation

Variable 25% Median Mean 75% Std. Dev. N Full Sample Patents 0 0 3.11 1 23.71 19,437 Citations/Patent 0 0 2.54 0 11.56 19,437 Sub-sample with patents > 0 Patents 1.04 3 11.48 7.25 44.51 5,264 Citations/Patent 0 2.55 9.39 8.91 20.72 5,264

Panel C: Control Variables

Variable Mean Median Std. Dev. N Tobin’s Q 3.01 2.08 2.94 14,230 Sales (mil.) 375.07 51.77 2122.73 16,653 Assets (mil.) 485.46 76.78 2583.134 16,715 ROA (%) -10.43 3.81 42.17 16,521 R&D/Assets (%) 14.06 6.98 21.12 19,437 CapExp/Assets (%) 6.15 4.00 6.70 16,371 Leverage (%) 34.64 25.80 34.83 19,437 Firm Age 2.91 2.00 5.11 19,437 PPE/Assets (%) 17.36 11.19 17.46 16,670 Institutional Ownership (%) 37.58 32.31 29.01 13,061 Insider Ownership (%) 19.07 12.30 20.52 10,420 Herfindahl Index 0.24 0.11 0.31 19,437

40

Table 4: Failure Tolerance and Corporate Innovation

The dependent variable is the natural logarithm of the number of patents in a year in models (1) and (2), and is the natural logarithm of the number of citations per patent in a year in models (3) and (4). “Patents>0” refers to the subsample of firms with at least one patent during our sample period. Coefficient estimates and the Huber-White-Sandwich robust standard errors clustered by IPO firm (in parentheses) are reported. Industry fixed effect is by 2-digit SIC industries. ***, ** and * indicate significance at 1%, 5% and 10% levels respectively.

Ln(Patents) Ln(Citations/Patent) Full Sample Patents>0 Full Sample Patents>0 (1) (2) (3) (4) Ln(Failure Tolerance) 0.258** 0.422*** 0.201** 0.310** (0.113) (0.159) (0.093) (0.122) Ln(Sales) 0.097*** 0.117*** 0.030** 0.032** (0.019) (0.022) (0.014) (0.015) ROA 0.677*** 0.564*** 0.388*** -0.014 (0.141) (0.197) (0.125) (0.174) R&D/Assets 1.520*** 1.052*** 1.066*** 0.223 (0.286) (0.356) (0.248) (0.310)

2.017*** 3.071*** 1.259** 1.968** CapExp/Assets (0.586) (0.919) (0.584) (0.880)

Leverage -0.813*** -0.689*** -0.755*** -0.691*** (0.162) (0.237) (0.133) (0.195) Tobin’s Q 0.093*** 0.072*** 0.071*** 0.040*** (0.015) (0.017) (0.012) (0.014)

1.006*** 1.436*** 0.812*** 1.142*** Institutional Ownership (0.188) (0.257) (0.145) (0.188)

Firm Age 0.058*** 0.035 0.016 -0.013 (0.018) (0.024) (0.010) (0.013) PPE/Assets 0.124 0.255 -0.078 -0.198 (0.361) (0.563) (0.317) (0.461) Herfindahl Index -0.252 -0.144 -0.095 0.144 (0.181) (0.279) (0.158) (0.225) Constant 0.149 0.496 -3.139** -4.235*** (0.607) (0.836) (1.386) (0.508) Industry fixed effects Yes Yes Yes Yes Year fixed effects Yes Yes Yes Yes Observations 11,994 7,607 11,994 7,607 R2 0.315 0.257 0.253 0.261

41

Table 5: Controlling for Other VC Characteristics

In this table we control for other observable VC firm characteristics and lead VC firm fixed effects. Coefficient estimates and the Huber-White-Sandwich robust standard errors clustered by IPO firm (in parentheses) are reported. ***, ** and * indicate significance at 1%, 5% and 10% levels respectively.

Panel A: Controlling for Observable VC Characteristics

Dependent Variable: Ln(Patents) (1) (2) (3) (4) Ln(Failure Tolerance) 0.239** 0.267** 0.251** 0.299** (0.117) (0.118) (0.119) (0.117)

0.203 0.024 0.182 0.068 Investment Concentration (0.434) (0.459) (0.402) (0.397)

Past Successful Exit 0.205 0.393 0.243 0.515 (0.401) (0.391) (0.311) (0.345) Local Competition 0.083 0.102 0.088 0.103 (0.072) (0.071) (0.070) (0.068) Ln(Past Amount Invested) -0.003 (0.050)

-0.060 Ln(Past Firms Invested) (0.076)

Ln(Past Fund Raised) -0.014 (0.034) Ln(VC Age) -0.155* (0.087) Controls, industry, and year fixed effects Yes Yes Yes Yes Observations 11,994 11,994 11,994 11,994 R2 0.316 0.317 0.316 0.317

Dependent Variable: Ln(Citation/Patents) (1) (2) (3) (4) Ln(Failure Tolerance) 0.199** 0.222** 0.211** 0.239** (0.100) (0.101) (0.100) (0.099)

0.096 -0.056 0.137 0.068 Investment Concentration (0.336) (0.352) (0.316) (0.309)

Past Successful Exit 0.219 0.336 0.175 0.366 (0.305) (0.293) (0.254) (0.270) Local Competition 0.105* 0.114** 0.101* 0.108** (0.056) (0.055) (0.053) (0.054) Ln(Past Amount Invested) -0.031 (0.037)

-0.083 Ln(Past Firms Invested) (0.054)

Ln(Past Fund Raised) -0.031 (0.027) Ln(VC Age) -0.149** (0.066) Controls, industry, and year fixed effects Yes Yes Yes Yes Observations 11,994 11,994 11,994 11,994 R2 0.254 0.255 0.254 0.255

42

Panel B: Controlling for Observable VC Characteristics and Lead VC Fixed-Effects

Dependent Variable: Ln(Patents) (1) (2) (3) (4) Ln(Failure Tolerance) 0.227* 0.247** 0.246** 0.270** (0.125) (0.125) (0.125) (0.125)

0.207 0.006 0.232 0.161 Investment Concentration (0.432) (0.447) (0.397) (0.395)

Past Successful Exit 0.441 0.633 0.454 0.701* (0.432) (0.422) (0.330) (0.370) Local Competition 0.084 0.102 0.085 0.104 (0.075) (0.073) (0.073) (0.072) Ln(Past Amount Invested) -0.031 (0.052)

-0.105 Ln(Past Firms Invested) (0.078)

Ln(Past Fund Raised) -0.045 (0.039) Ln(VC Age) -0.204** (0.096) Controls, industry, and year fixed effects Yes Yes Yes Yes Lead VC fixed effects Yes Yes Yes Yes Observations 11,994 11,994 11,994 11,994 R2 0.354 0.354 0.354 0.355

Dependent Variable: Ln(Citation/Patents) (1) (2) (3) (4) Ln(Failure Tolerance) 0.185* 0.203* 0.214** 0.217** (0.109) (0.109) (0.109) (0.109)

0.177 -0.001 0.199 0.178 Investment Concentration (0.349) (0.361) (0.325) (0.320)

Past Successful Exit 0.335 0.490 0.373 0.494* (0.334) (0.320) (0.270) (0.290) Local Competition 0.074 0.087 0.079 0.083 (0.065) (0.064) (0.062) (0.063) Ln(Past Amount Invested) -0.039 (0.041)

-0.105* Ln(Past Firms Invested) (0.060)

Ln(Past Fund Raised) -0.062* (0.032) Ln(VC Age) -0.176** (0.075) Controls, industry, and year fixed effects Yes Yes Yes Yes Lead VC fixed effects Yes Yes Yes Yes Observations 11,994 11,994 11,994 11,994 R2 0.274 0.274 0.275 0.275

43

Table 6: Persistence of the Failure Tolerance Effect

The dependent variable is the natural logarithm of the number of patents in a year in Panel A, and is the natural logarithm of the number of citations per patent in Panel B. “Firm Age” is the number of years since the IPO year. All regressions are our baseline regressions as in Table 4, but are restricted to firms that existed for at least seven years after IPO (i.e., Firm Age≥7). The bottom row of each panel reports the results from the Chi-square test of the difference in the failure tolerance effect between Model (1) and another model. Coefficient estimates and the Huber-White-Sandwich robust standard errors clustered by IPO firm (in parentheses) are reported. Industry fixed effect is by 2-digit SIC industries. ***, ** and * indicate significance at 1%, 5% and 10% levels respectively.

Panel A: Ln(Patents)

(1) (2) (3) Firm Age ≤ 2 Firm Age ≤ 5 Firm Age ≤ 7 Ln(Failure Tolerance) 0.359** 0.314** 0.289** (0.161) (0.149) (0.139) Ln(Sales) 0.060* 0.090*** 0.100*** (0.034) (0.027) (0.024) ROA 0.401 0.678*** 0.798*** (0.328) (0.259) (0.231) R&D/Assets 2.270*** 1.771*** 1.751*** (0.702) (0.528) (0.456)

1.508 1.478 1.213 CapExp/Assets (1.239) (0.941) (0.862)

Leverage -0.585 -0.734** -0.770*** (0.359) (0.304) (0.248) Tobin’s Q 0.087*** 0.107*** 0.116*** (0.022) (0.018) (0.017)

0.414 0.977*** 0.919*** Institutional Ownership (0.299) (0.228) (0.207)

Firm Age -0.677 0.028 0.281 (0.692) (0.553) (0.492) PPE/Assets -0.431 -0.243 -0.295 (0.279) (0.249) (0.233) Herfindahl Index 0.060* 0.090*** 0.100*** (0.034) (0.027) (0.024) Constant -3.615* -3.407* -3.140 (2.024) (1.996) (1.920) Industry fixed effects Yes Yes Yes Year fixed effects Yes Yes Yes Observations 1,910 3,831 5,114 R2 0.323 0.323 0.309 Comparison of failure tolerance effect with (1) (p-value)

-0.045 (0.584)

-0.070 (0.449)

44

Panel B: Ln(Citations/Patent)

(1) (2) (3) Firm Age ≤ 2 Firm Age ≤ 5 Firm Age ≤ 7 Ln(Failure Tolerance) 0.323* 0.298** 0.257* (0.178) (0.149) (0.135) Ln(Sales) 0.047 0.053** 0.053** (0.035) (0.026) (0.023) ROA 0.210 0.390 0.515** (0.363) (0.265) (0.228) R&D/Assets 1.388* 1.049** 1.194*** (0.728) (0.529) (0.425)

2.427* 2.055** 1.971** CapExp/Assets (1.382) (0.989) (0.934)

Leverage -1.152*** -0.994*** -1.035*** (0.384) (0.335) (0.260) Tobin’s Q 0.108*** 0.122*** 0.129*** (0.028) (0.021) (0.020)

0.497 0.638*** 0.453** Institutional Ownership (0.376) (0.247) (0.214)

Firm Age -1.115 -0.439 -0.140 (0.746) (0.570) (0.485) PPE/Assets -0.218 -0.058 -0.134 (0.325) (0.280) (0.253) Herfindahl Index 0.323* 0.298** 0.257* (0.178) (0.149) (0.135) Constant -3.051 -2.919* -2.348 (1.980) (1.757) (1.654) Industry fixed effects Yes Yes Yes Year fixed effects Yes Yes Yes Observations 1,910 3,831 5,114 R2 0.252 0.240 0.233 Comparison of failure tolerance effect with (1) (p-value)

-0.025 (0.796)

-0.066 (0.541)

45

Table 7: Difficulty of Innovation and the Effect of Failure Tolerance

The dependent variable is the natural logarithm of patents in a year in Panel A, and is the natural logarithm of citations per patent in Panel B. The patents categories are based on the classifications in Hall, Jaffe, and Trajtemberg (2005). The bottom row of each panel reports the results from the Chi-square test of the difference in the failure tolerance effect between category (1) and another patent category. Coefficient estimates and the Huber-White-Sandwich robust standard errors clustered by IPO firm (in parentheses) are reported. ***, ** and * indicate significance at 1%, 5% and 10% levels respectively.

Panel A: Ln(Patents)

Drugs & Chemical (1)

Computers & Electrical (2)

Software (3)

Miscellaneous(4)

Ln(Failure Tolerance) 0.743** 0.255** 0.190** 0.313*** (0.306) (0.102) (0.076) (0.091) Ln(Sales) 0.021 0.249*** 0.210*** 0.071* (0.018) (0.039) (0.038) (0.038) ROA 0.774*** 1.547*** 0.291* 0.737** (0.240) (0.238) (0.149) (0.369) R&D/Assets 2.165*** 5.440*** 0.754** 10.458*** (0.395) (0.479) (0.319) (1.153) CapExp/Assets 2.866** 2.779** 0.828 -0.899 (1.329) (1.082) (1.294) (0.747) Leverage -0.680** -1.500*** -0.982*** -1.300*** (0.309) (0.185) (0.170) (0.192) Tobin’s Q 0.058** 0.121*** 0.161*** 0.066** (0.026) (0.019) (0.016) (0.032) Institutional Ownership 0.835** 1.366*** 0.110 0.194 (0.356) (0.227) (0.228) (0.212) Firm Age 0.019 0.132*** 0.082*** 0.082*** (0.025) (0.015) (0.016) (0.015) PPE/Assets -1.854** -0.924** 0.786 -0.680*** (0.749) (0.418) (0.717) (0.254) Herfindahl Index -1.401*** -0.842*** 0.717** -0.593*** (0.421) (0.160) (0.339) (0.133) Constant -5.070*** -9.562*** -9.098*** -4.258*** (0.472) (0.742) (0.635) (0.876) Year fixed effects Yes Yes Yes Yes Observations 3,177 3,580 2,911 2,277 R2 0.207 0.243 0.155 0.296 Comparison of Failure Tolerance effect with (1) (p-value)

-0.488*** (0.004)

-0.553*** (<0.001)

-0.430*** (0.004)

46

Panel B: Ln(Citations/Patent)

Drugs & Chemical (1)

Computers & Electrical (2)

Software (3)

Miscellaneous(4)

Ln(Failure Tolerance) 0.576** 0.199** 0.177** 0.185** (0.225) (0.101) (0.088) (0.089) Ln(Sales) -0.003 0.074*** 0.104*** 0.019 (0.016) (0.027) (0.026) (0.037) ROA 0.099 1.321*** 0.343** 0.720* (0.221) (0.220) (0.169) (0.384) R&D/Assets 0.739** 4.807*** 0.696* 9.389*** (0.337) (0.440) (0.362) (1.152) CapExp/Assets 3.039** 1.328 0.809 -0.388 (1.332) (1.072) (1.493) (0.791) Leverage -0.643*** -1.183*** -0.932*** -1.066*** (0.234) (0.166) (0.163) (0.186) Tobin’s Q 0.027 0.072*** 0.151*** 0.011 (0.023) (0.019) (0.020) (0.032) Institutional ownership 0.278 1.073*** 0.573*** 0.358* (0.280) (0.207) (0.208) (0.203) Firm Age 0.027 0.019 0.027** 0.057*** (0.017) (0.012) (0.013) (0.014) PPE/Assets -2.344*** -0.505 1.003 -0.918*** (0.605) (0.401) (0.776) (0.255) Herfindahl Index -0.855** -0.544*** 0.971** -0.623*** (0.338) (0.155) (0.402) (0.134) Constant -4.543*** -6.406*** -7.460*** -3.785*** (0.382) (0.543) (0.481) (0.700) Year fixed effects Yes Yes Yes Yes Observations 3,177 3,580 2,911 2,277 R2 0.249 0.189 0.132 0.262 Comparison of Failure Tolerance effect with (1) (p-value)

-0.377** (0.017)

-0.399*** (0.008)

-0.391** (0.010)

47

Table 8: Insider Ownership, Failure Tolerance and Corporate Innovation

The dependent variable is the natural logarithm of patent counts in a year in models (1) and (2), and is the natural logarithm of citations per patent in models (3) and (4). Coefficient estimates and the Huber-White-Sandwich robust standard errors clustered by IPO firm (in parentheses) are reported. Industry fixed effect is by 2-digit SIC industries. ***, ** and * indicate significance at 1%, 5% and 10% levels respectively.

Ln(Patents) Ln(Citations/Patent) (1) (2) (3) (4)

-0.073*** -0.062*** -0.062*** -0.052** Ln(Insider ownership) (0.021) (0.022) (0.019) (0.020)

Ln(Failure Tolerance) 0.264** 0.194* (0.125) (0.108) Ln(Sales) 0.105*** 0.102*** 0.030* 0.030* (0.021) (0.021) (0.017) (0.016) ROA 0.668*** 0.702*** 0.387*** 0.396*** (0.162) (0.166) (0.146) (0.149)

1.653*** 1.513*** 1.265*** 1.168*** R&D/Assets (0.337) (0.344) (0.299) (0.304)

2.170*** 2.082*** 1.131* 1.080 CapExp/Assets (0.677) (0.669) (0.681) (0.683)

Leverage -0.925*** -0.902*** -0.889*** -0.876*** (0.186) (0.190) (0.160) (0.163) Tobin’s Q 0.100*** 0.100*** 0.084*** 0.083*** (0.016) (0.016) (0.014) (0.014)

1.128*** 1.048*** 0.974*** 0.919*** Institutional Ownership (0.208) (0.212) (0.172) (0.175)

Firm Age 0.055** 0.065*** 0.005 0.011 (0.022) (0.023) (0.015) (0.015)

-0.187 0.171 -0.087 0.122 PPE/Assets (0.408) (0.405) (0.364) (0.370)

Herfindahl Index -0.530*** -0.344* -0.311* -0.156 (0.196) (0.207) (0.178) (0.191) Constant -2.272 -3.241* -2.122 -3.010** (1.955) (1.718) (1.740) (1.433) Industry fixed effects Yes Yes Yes Yes Year fixed effects Yes Yes Yes Yes Observations 9,934 9,452 9,934 9,452 R2 0.314 0.322 0.236 0.241

48

Table 9: Failure Tolerance and Firm Value

The dependent variable is the natural logarithm of a firm’s Tobin’s Q at the first fiscal year end after IPO. “Ind. Total Patents” and “Ind. Total Citations” are the total number of patents and the total number of citations in the entire sample period by 4-digit SIC industries, respectively. Coefficient estimates and the Huber-White-Sandwich robust standard errors (in parentheses) are reported. Industry fixed effect is by 2-digit SIC industries. ***, ** and * indicate significance at 1%, 5% and 10% levels respectively.

(1) (2) (3) 0.081* -0.077 -0.124 Ln(Failure Tolerance) (0.046) (0.082) (0.086)

0.027** Ln(Failure Tolerance) x Ln(Ind. Total Patents) (0.013) 0.044*** Ln(Ind. Total Patents)

(0.013) 0.032** Ln(Failure Tolerance) x Ln(Ind. Total Citations) (0.014)

Ln(Ind. Total Citations) 0.043*** (0.014)

0.029*** 0.029*** 0.029*** Ln(Insider Ownership) (0.011) (0.011) (0.011)

Ln(Sales) -0.013* -0.014* -0.013* (0.007) (0.007) (0.007) ROA 0.206* 0.166 0.175* (0.106) (0.104) (0.105)

1.040*** 0.822*** 0.863*** R&D/Assets (0.210) (0.209) (0.209)

1.063*** 1.196*** 1.168*** CapExp/Assets (0.328) (0.327) (0.323)

Leverage 0.255*** 0.287*** 0.283*** (0.086) (0.085) (0.085)

-0.106 -0.076 -0.056 Institutional Ownership (0.118) (0.117) (0.118)

-0.887*** -0.909*** -0.860*** PPE/Assets (0.168) (0.164) (0.163)

Herfindahl Index -0.091 0.090 0.089 (0.077) (0.081) (0.081) Constant 0.804*** 0.393 0.500* (0.308) (0.305) (0.294) Industry fixed effects Yes Yes Yes Year fixed effects Yes Yes Yes Observations 1,367 1,367 1,367 R2 0.277 0.294 0.292

49