Managerial Short-Termism and Corporate Innovation Strategies*

Huasheng Gao Nanyang Business School

Nanyang Technological University S3-B1A-06, 50 Nanyang Avenue, Singapore 639798

65.6790.4653 hsgao@ntu.edu.sg

Po-Hsuan Hsu

Faculty of Business and Economics University of Hong Kong

Pokfulam Road, Hong Kong 852.2859.1049

paulhsu@hku.hk

Kai Li Sauder School of Business

University of British Columbia 2053 Main Mall, Vancouver, BC V6T 1Z2

604.822.8353 kai.li@sauder.ubc.ca

First version: September, 2013 This version: April, 2014

Abstract

In this paper we employ a sample of public and private firms to examine the impact of managerial short-termism associated with public capital market on innovation strategies. We show that public firms’ patents are less exploratory and more exploitative than private firms’ patents. The results based on propensity score matching, the IPO firms and their matched private (public) control firms, and the instrumental variable approach reinforce the findings from the cross-section. We then identify a number of possible underlying forces behind managerial short-termism: a higher fraction of vested stock and option grants in CEOs’ overall equity incentive portfolios, a higher likelihood of a firm becoming a takeover target due to low valuation, and a greater presence of short-term institutional investors. We conclude that managerial short-termism associated with public capital market leads to an innovation strategy that emphasizes short-term success and potentially underinvests in revolutionary technologies. Keywords: exploitative innovation; exploratory innovation; innovation strategies; managerial short-termism; patents; private firms JEL Classification: G32; O32 * We are grateful for helpful comments from Andres Almazan, Jan Bena, Bill Kerr, Alexander Ljunqgvist, Gustavo Manso, Ron Masulis, Hernan Ortiz-Molina, Jay Ritter, Cheng-Wei Wu, Ting Xu, seminar participants at Boston University, Harvard Business School, Nanyang Technological University, National Chengchi University, University of British Columbia, University of Hong Kong, University of Oklahoma, and conference participants at the 21st

Conference on the Theories and Practices of Securities and Financial Markets (Kaohsiung). We acknowledge dedicated effort of our research assistants Rita Lei Chen, Yi Du, Cherry Guo, Jialing Hu, and Chi-Yang Tsou. Gao acknowledges financial support from the Tier One Research Grant provided by the Ministry of Education of Singapore. Gao and Li acknowledge financial support from the Social Sciences and Humanities Research Council of Canada. All errors are our own.

Managerial Short-Termism and Corporate Innovation Strategies

Abstract

In this paper we employ a sample of public and private firms to examine the impact of managerial short-termism associated with public capital market on innovation strategies. We show that public firms’ patents are less exploratory and more exploitative than private firms’ patents. The results based on propensity score matching, the IPO firms and their matched private (public) control firms, and the instrumental variable approach reinforce the findings from the cross-section. We then identify a number of possible underlying forces behind managerial short-termism: a higher fraction of vested stock and option grants in CEOs’ overall equity incentive portfolios, a higher likelihood of a firm becoming a takeover target due to low valuation, and a greater presence of short-term institutional investors. We conclude that managerial short-termism associated with public capital market leads to an innovation strategy that emphasizes short-term success and potentially underinvests in revolutionary technologies. Keywords: exploitative innovation; exploratory innovation; innovation strategies; managerial short-termism; patents; private firms JEL Classification: G32; O32

1

1. Introduction

Technological innovation represents modern corporations’ endeavor to develop and accumulate

knowledge, and has long been recognized as the catalyst to economic growth and productivity increase

(Solow, 1957; Romer, 1990) and a key factor in competitive advantage of nations (Porter, 1998).

The management literature has identified two main strategies in organizational learning

trajectories: exploratory innovation and exploitative innovation. Exploratory innovation is radical

innovation to meet the needs of emerging customers or markets. It requires new knowledge or departure

from existing knowledge, and its payoffs take longer time to realize and are of higher uncertainty. In

contrast, exploitative innovation is incremental innovation to meet the needs of existing customers or

markets. It builds on existing knowledge and reinforces existing skills, processes, and structures, and its

payoffs realize faster with less uncertainty (Levinthal and March, 1993; McGrath, 2001; Benner and

Tushman, 2002; Smith and Tushman, 2005).

In reality, firms have to constantly refine their current technologies to ensure stable future profits

and maintain market shares, and in the meantime, they also must invest in new knowledge and market

opportunities to hedge against revolutionary technologies that may reshape the whole industry structure.

March (1991) posits that maintaining an appropriate balance between exploratory and exploitative

innovation is critical for firm survival and prosperity. Levinthal and March (1993, p. 105) further argue

that, “The basic problem confronting an organization is to engage in sufficient exploitation to ensure its

current viability and, at the same time, to devote enough energy to exploration to ensure its future

variability.” (Also see Lavie, Stettner, and Tushman, 2010 for a review.) Empirically, Katila and Ahuja

(2002) find that a firm’s ability to create new products is determined by both the independent and

interaction effects of exploratory and exploitative innovation. Using a sample of manufacturing firms, He

and Wong (2004) show that the interaction between explorative and exploitative innovation strategies is

positively related to sales growth rate, while the relative imbalance between explorative and exploitative

innovation strategies is negatively related to sales growth. Uotila, Maula, Keil, and Zahra (2009) further

show that both over-exploitation and over-exploration hurt firm value as measured by Tobin’s Q. Akcigit

and Kerr (2012) find that growth spillover effects are larger from exploratory innovation than from

exploitative innovation.

2

Despite the important heterogeneity in corporate innovation strategies and their distinct effects on

new product development and revenue growth, almost everything we know about innovation at the firm

level is based on patent and citation counts. Moreover, given the great independence of firms on public

stock market to finance their R&D (Brown, Fazzari, and Petersen, 2009), understanding the influence of

public stock market on corporate innovation strategy is an important but under-explored research

question. This paper aims to fill a gap in the literature by highlighting heterogeneous innovations and

providing insights into drivers of innovation strategies.

In this study we compile a database of patent and financial information on both public and private

firms to help identify key influences of public capital market on innovation strategies. We posit that both

sources of financing and managerial short-termism should affect innovation strategies. On the one hand,

access to public equity relaxes financial constraints and encourages more risk-taking learning in public

firms, resulting in them generating more patents and doing more exploratory innovation than private

firms. On the other hand, the pressure to deliver near-term results and the presence of an active takeover

market lead public firm managers to take the short-cut by engaging in low-risk learning, resulting in them

generating fewer patents and doing more exploitative innovation than private firms. Our use of both

public and private firms sharpens the contrasting effects of the financing channel and the managerial

short-termism channel on innovation strategies. Further, the innovation strategy of private firms is in

itself of great interest to financial economists and management researchers due to a lack of data prior to

our study.

Using a large sample of public and private US firms from 1997-2008, we first show that public

firms generate significantly more patents than private firms, consistent with the predictions of the

financing channel. More importantly, we show that public firms’ patent portfolios are more exploitative

(i.e., mainly making use of existing knowledge) and less exploratory (i.e., in pursuit of new knowledge)

than private firms’ patent portfolios. We also show that public firms’ patent portfolios tend to be narrower

in scope but greater in depth in terms of the knowledge domain. These findings support the predictions of

the managerial short-termism channel.

We recognize that our investigation is vulnerable to selection concerns and we take a number of

approaches to addressing them. First, we employ propensity score-matched public control firms and

3

compare them with the private firm sample. We find that private firms produce significantly more

exploratory (fewer exploitative) patents than their matched public control firms. Second, we employ a

transitioning sample of the IPO firms and their matched private (public) control firms. We find that

compared to their private peer firms, these newly-public firms experience a significant increase (decrease)

in exploitative (exploratory) patents following the IPO; while compared to their public peer firms, these

newly-public firms have significantly more (fewer) exploratory (exploitative) patents before the IPO.

Moreover, we find that innovation strategies do not explain a firm’s decision to go public, suggesting that

our findings are not mostly driven by reverse causality. Third, we apply a treatment regression approach

with the industry-level IPO underwriter concentration as the instrumental variable to address the selection

issue that companies may choose to stay private. We find that the differences in strategies—exploratory

versus exploitative innovation—are even greater between public and private firms after controlling for

selection.

Finally, we conduct within-public firm analyses to sharpen our evidence for the underlying forces

behind managerial short-termism. Our premise is that being public firms, there is a great emphasis on

short-term performance (Porter, 1992; Graham, Harvey, and Rajgopal, 2005), maybe even more so than

being private firms. If we show that the differences in innovation strategies between high and low short-

termism public firms, sorted by different sources of the short-term performance pressure, are consistent

with those between public and private firms, we can conclude that managerial short-termism associated

with public capital market is a key driver of corporate innovation strategies.

To do so we sort the public firms in our sample based on the vesting horizon of their CEOs’

equity incentive portfolios, their exposure to the market for corporate control, and their susceptibility to

transient institutional investors, and then examine the differences in innovation strategies within different

groups of the public firm sample. We classify a public firm to be of high short-termism if the ratio of its

CEO’s vested stock and option grants to her total holdings of stock and option grants is above the sample

median. CEOs with a larger fraction of their equity incentive portfolios vested are subject to more

pressure to meet short-term performance targets. We find that firms whose CEOs have a short-term focus

are associated with more exploitative and less exploratory innovation. Using the takeover exposure model

of Cremers, Nair, and John (2009) and Fang, Tian, and Tice (2013), we classify a public firm to be of

high short-termism if the (negative) coefficient on the M/B ratio is below the sample median. An

4

increased takeover threat could put pressure on managers to boost current profits and cut long-term high-

risk investments in innovation as a means to fend off takeover attempts. We find that firms with a higher

likelihood of becoming a takeover target due to low stock valuation are associated with significantly more

exploitative (fewer exploratory) patents than their lower likelihood peers. Using the classification scheme

of Bushee (1998, 2001), we classify a public firm to be of high short-termism if the ratio of transient

institutional ownership to total institutional ownership is above the sample median. A greater presence of

transient institutional investors puts increased pressure on managers to focus on current profits and reduce

long-term high-risk investments in innovation. We find that firms held by more short-term investors are

associated with significantly more exploitative (fewer exploratory) patents than firms with fewer short-

term investors. We conclude that managerial short-termism driven by CEOs’ equity incentives, the

market for corporate control, and the presence of transient institutional investors leads to less exploratory

and more exploitative innovation.

Our paper makes contribution to the literature in a number of dimensions. First, our comparison

of innovative activities in public and private firms allows us to speak to big picture questions of important

drivers of corporate innovation strategies—sources of financing and managerial short-termism, which is

the real distinction of our work from others. Previous investigations of the issues have largely ignored

innovation strategies and different organizational learning trajectories and have been primarily limited to

using public firms only. We show that on the extensive margin when the effects of the financing channel

dominate, public firm status is associated with more innovation and patent counts. On the intensive

margin when the effects of the managerial short-termism channel dominate, public firm status is

associated with fewer exploratory patents. Since it takes exploration to survive and thrive through

technological innovation waves in the long run (McGrath, 2001; Lavie, Stettner, and Tushman, 2010), our

empirical evidence highlights the real effects of sources of financing and managerial short-termism.

Second, this is one of the first studies providing large sample evidence on the innovative activities

of privately-held US firms. Most studies to-date on private firm innovation employ only a small sample of

transitioning firms such as LBO target firms or IPO firms (Lerner, Sørensen, and Strömberg, 2011;

Bernstein, 2012; Aggarwal and Hsu, 2013) or a sample of startup firms (Barrot, 2012). In

contemporaneous work, Acharya and Xu (2013) examine the relation between firms’ dependence on

external capital and innovation output—patent and citation counts—using a large sample of public and

5

private firms. Different from their work, we focus on innovation strategies—exploratory versus

exploitative innovation.

Finally, our study also adds to the management literature by highlighting the implications of

public market listing on corporate innovation strategies. Prior studies have shown that firms are more

(less) likely to engage in exploitative (exploratory) innovation when their shareholders/managers are

myopic or risk-averse (Levinthal and March, 1993; Smith and Tushman, 2005),1 when they pursue

economies of scale (Crossan, Lane, and White, 1999), when their innovative activities are more likely to

be subjects of imitation (Cohen and Levinthal, 1994), or when their environment appears to be less

volatile (McGrath, 2001). The findings from our study suggest that the public market listing status of a

firm should be part of its long-term planning on the trajectories of organizational learning and the

evolutions of its competitive advantages.

The paper is organized as follows. Section 2 reviews the related literature and develops our

hypotheses. Section 3 describes our sample and key variable construction. Section 4 presents the main

empirical analysis. Sample selection issues are addressed in Section 5. Additional investigation of the

underlying forces behind managerial short-termism is implemented in Section 6. We conclude in Section

7.

2. Literature Review and Hypothesis Development

There is a substantial literature examining the determinants of corporate investment in R&D and

innovation. For our purpose, we mainly review papers directly related to our empirical investigation,

namely the roles of financing and managerial short-termism, then proceed to develop our hypotheses.

2.1. Related Literature

Our paper is related to the large literature establishing that financial development is associated

with long-term economic growth and technological progress (see, for example, Schumpeter, 1934;

1 Levinthal and March (1993) build a conceptual framework of organizational learning and argue that managerial myopia leads to over-exploitation and under-exploration due to ignorance of the long run, the big picture, and the failure.

6

Levine, 2005; Hall and Lerner, 2010 for excellent reviews). However, less is known about how different

financing choices—equity versus debt and public versus private equity—affect corporate innovation.

Using a signaling model, Bhattacharya and Ritter (1983) show that innovative firms may lose

their informational advantage if they disclose details of their R&D projects to capital markets in order to

raise financing at better terms. Incorporating similar trade-offs, Maksimovic and Pichler (2001) further

demonstrate that both technological and competitive risks affect the timing of private and initial public

offerings in new industries: Early private financing occurs in risky industries with high development costs

and low competitive risk, while early public financing occurs in viable industries with low development

costs and low competitive risk.

Empirically, there is mixed evidence on the effect of debt versus equity financing on innovation.

On the one hand, Atanassov, Nanda, and Seru (2007) argue that banks do not have the ability to evaluate

novel technologies and hence tend to discourage investing in innovative projects and be more prone to

shut down ones that are on-going, and show that public firms that rely more on arm’s length financing

(equity and public debt) than on relational bank financing innovate more. Brown et al. (2009) find that

supply shifts in external equity financing can explain a significant portion of the 1990s R&D boom and

decline, especially among young firms that are most likely to be financially constrained. Nanda and

Rhodes-Kropf (2011, 2013) illustrate both theoretically and empirically that hot financial markets

encourage investors to engage in risky experimentation by funding innovative ideas because they expect

these innovative activities to have a higher chance of receiving subsequent funding. In a cross-country

study, Hsu, Tian, and Xu (2014) find that more developed equity markets encourage innovation in high-

tech industries while more developed debt markets discourage innovation in those industries. On the other

hand, a number of recent empirical studies suggest that increased credit supply is beneficial to innovation,

especially in small entrepreneurial firms. Using Italian data in the 1990s, Benfratello, Schiantarelli, and

Sembenelli (2008) show a positive and significant effect of banking development on the probability of

local firms’ introducing process or product innovation. Kerr and Nanda (2009) find that US banking

deregulations brought about exceptional growth in both entrepreneurship and business closures. Using the

World Bank Enterprise Surveys database, Ayyagari, Demirgüç-Kunt, and Maksimovic (2011) show that

access to bank financing is associated with greater firm innovation for small and medium enterprises

around the developing world. Using US data, Amore, Schneider, and Žaldokas (2013) show that interstate

7

banking deregulation had significant beneficial effects on corporate innovation of public firms, and

Chava, Oettl, Subramanian, and Subramanian (2013) show the same effect on innovation of young private

firms. It is worth noting that Chava et al. study is based on state-level innovation output by private firms

without knowing their identities and basic financials.

Focusing on public equity financing, Maug (1998), Admati and Pfleiderer (2009), Edmans

(2009), and Edmans and Manso (2011) argue that stock liquidity facilitates the entry of monitoring

blockholders and/or blockholders gathering and trading on private information which makes stock prices

more informative. If high liquidity leads to better monitoring and/or more efficient prices, managers may

be willing to forego short-term profits to invest in long-term projects such as innovation. On the other

hand, Ferreira, Manso, and Silva (2012) point out that private firms are less transparent to outside

investors than are public firms, leading to the tolerance-for-failure effect as the key determinant of

exploratory innovation in private firms. Longer investment horizon of private firm shareholders reinforces

the positive effect of low transparency on exploratory innovation. In contrast, the rapid incorporation of

good news into market prices creates incentives for short-termism behavior by public firms including

pursuing exploitative innovation. They predict that it is optimal to go public to exploit existing ideas and

optimal to stay private to explore new ideas. Fang et al. (2013) show that an increase in liquidity is

associated with a reduction in corporate innovation. Acharya and Xu (2013) find that public listing is

beneficial to the innovation of firms in industries with a greater need for external capital. The mixed

evidence on the role of financing in corporate innovation thus calls for further investigation.

Our paper is also related to the literature on managerial short-termism and corporate investment.

In Stein (1988), there is greater information asymmetry for long-term investments than for short-term

investments and there are takeover threats. If a takeover occurs before the payoff of long-term projects is

known, target shareholders will receive an undervalued price that does not reflect the true value of the

firm. To protect shareholders from this potential wealth loss, managers of undervalued firms attempt to

boost the firms’ current earnings by cutting long-term investments that the market cannot value

accurately, leading to managerial short-termism. Bebchuk and Stole (1993) further show that a

combination of information asymmetry on the level of investment undertaken and managers caring for

short-run performance could also lead to under-investment in long-run projects. Bolton, Scheinkman, and

Xiong (2006) show that allowing for heterogeneous beliefs by investors and consequently speculative

8

deviations of stock prices from fundamentals create a distortion in CEO compensation leading to a short-

term orientation.

Empirically, there is mixed evidence on the role of takeover markets in corporate innovation.

Atanassov (2013) shows that strong corporate governance proxied by the threat of hostile takeovers

positively affects innovation and firm value. Sapra, Subramanian, and Subramanian (2013) find that

innovation varies non-monotonically in a U-shape manner with the takeover pressure faced by firms:

Innovation is fostered either by an unhindered market for corporate control or by anti-takeover laws that

effectively deter takeovers. On the other hand, Meulbroek, Mitchell, Mulherin, Netter, and Poulsen

(1990) show that after introducing anti-takeover amendments, firms do not significantly increase their

R&D expenditures.

One challenge of testing these myopia models is the lack of a clean measure for managerial short-

termism. A number of recent papers have made progress in this direction. Using the sensitivity of a

CEO’s stock and option grants that are scheduled to vest over the coming year to proxy for her myopic

tendencies, Edmans, Fang, and Lewellen (2014) find that managerial short-termism leads to reductions in

real investments including R&D, capital expenditures, and advertising expenses. Ladika and Sautner

(2014) show that FAS 123R induces firms to accelerate the vesting period of option grants, and such

accelerated vesting leads to a reduction in capital expenditures. Our paper is complementary in that we

rely on the dichotomy between public and private firms to capture managerial short-termism and to

investigate the link between managers’ short-term focus and innovation strategies. We also explore the

underlying driving forces behind managerial short-termism.

2.2. Hypothesis Development

Being publicly-listed versus privately-held are associated with a number of important differences

that could potentially impact corporate innovation strategies. Under the financing perspective, private

firms are subject to more financing frictions than public firms and they rely mainly on debt financing

(Brav, 2009; Gao, Harford, and Li, 2013), which typically does not encourage innovative behavior

(Atanassov, Nanda, and Seru, 2007). Better access to funding also generates financial slack that protects

firms from uncertainty associated with innovative activities and thus fosters a culture of experimentation

9

(Nohria and Gulati, 1996). Levinthal and March (1993) note that the costs to exploiting activities are

lower than those to exploring activities.

Public firms’ stocks are freely traded, allowing public firm shareholders to achieve better

portfolio diversification and risk-sharing, which in turn encourages more corporate risk-taking (King and

Levine, 1993; Faccio, Marchica, and Mura, 2011). Moreover, risk management tools (such as derivatives)

available in public equity markets allow investors to make more efficient portfolio allocation and to

pursue high risk-high return projects (Levine, 2005). The above discussions suggest that public firms with

better access to financing and their shareholders with better risk-sharing tools are more likely to invest in

innovation and engage in long-term high-risk exploratory innovation, leading to our first hypothesis

regarding innovation strategies in public and private firms:

The Financing Hypothesis: Public firms’ innovation is more exploratory (less exploitative) than private firms’ innovation.

Under the managerial short-termism perspective, we expect that managers in public firm are

faced with greater short-term performance pressure than their counterparts in private firms for a number

of reasons. First, public firm managers can easily sell their equity holdings in the stock market (upon

vesting). Their ability of taking advantage of short-term price fluctuations to profit from equity sales

encourages these managers to pursue short-term projects at the expense of high long-term fundamental

values (Bolton et al., 2006; Edmans et al., 2013; Gopalan, Milbourn, Song, and Thakor, 2014). In

contrast, private firm managers have to hold their equity stakes for a long period of time and have much

more limited ways to cash out because of their non-publicly-traded stock. Second, while a temporary

undervaluation may increase the likelihood for a public firm to be taken over at an unfavorable price

(Stein, 1988), private firms are faced with very little threat of hostile takeovers due to their non-publicly-

traded stock. Lastly, short-term shareholders in public firms tend to put pressure on managers to sacrifice

long-term investments in order to meet short-term earnings targets (Bushee, 1998). Such short-term

pressure is much weaker in private firms because the lack of stock liquidity forces private firm

shareholders to take a long investment horizon (Ferreira et al., 2012). Given that exploitative innovation

results in immediate performance (Levinthal and March, 1993), we expect that public firms choose to

focus on exploitative innovation over exploratory innovation. The above discussions lead to our

alternative hypothesis:

10

The Short-Termism Hypothesis: Public firms’ innovation is more exploitative (less exploratory) than private firms’ innovation.

Our empirical analysis is designed to test these two hypotheses that have opposite predictions on

innovation strategies in public and private firms, and also to explore alternative explanations for why

innovation strategies differ between public and private firms. In the next section we describe our sample

and key variable construction, and present descriptive statistics.

3. Sample Formation and Variable Construction

Our primary data sources are the Capital IQ database, the National Bureau of Economic Research

(NBER) Patent Citations Date File (Hall, Jaffe, and Trajtenberg, 2005a, 2010 update), and the Harvard

Business School (HBS) US Patent Inventor Database (Lai, D’Amour, Yu, and Fleming, 2011).2 Capital

IQ is an affiliate of Standard & Poor’s that produces the Compustat database.3

We start with a list of unique non-financial US firms covered in Capital IQ for the period 1997-

2008. Our sample starts in 1997 because the data coverage in Capital IQ is poor before 1995 and we

employ a lead-lag specification and a three-year window to measure innovation strategies, and ends in

2008 because the patent information from the HBS database is available till 2010 but there is a well-

known two to three year lag between patent application and award dates (Hall et al., 2005a). We obtain

the patent and citation data for these firms by manually matching the NBER and HBS patent databases

2 The HBS patent database is constructed in a similar manner as the NBER patent database, and has more recent patent data (and more details about patent inventors). There are a number of important differences between the HBS patent database and the NBER patent database. First, the HBS patent database is not matched to Compustat, while the NBER patent database is. Second, the HBS patent database has scant coverage of PDPASS (the NBER assignee identifier) for the period 2007-2010. While 84.6% of awarded patents carry PDPASS in the NBER patent database in all years, the ratios of awarded patents with PDPASS in the HBS database are 67%, 65%, 62%, and 0% in 2007, 2008, 2009, and 2010, respectively. Based on both algorithm and manual matching, we are able to obtain PDPASS for 82.4% of awarded patents in the HBS database for the period 2007-2010. 3 The Capital IQ data on private US firms is based on the following mandatory disclosure requirements by the Securities and Exchange Commission (SEC). First, if a company decides on a registered public offering, the Securities Act requires it to file a registration statement (Form S-1) with the SEC that contains basic financial information. Second, and more applicable to our sample of privately-held firms, even if a company has not registered a securities offering, it must file an Exchange Act registration statement if it has more than $10 million in total assets and a class of equity securities, like common stock, with 500 or more shareholders. After that, the company is required to continue reporting via annual and quarterly reports and proxy statements. Capital IQ provides accounting information on private US firms with a similar level of details as provided by Compustat on public US firms.

11

(by first assignee name and then location) with the list of unique firms in the Capital IQ database. We end

up with an initial sample of 53,055 public firm-year observations and 148,852 private firm-year

observations.

We further require all sample firm-year observations with basic financial information. To

mitigate the reverse causality concern that innovation strategies drive going public/ private decisions

(rather than the public/private status drives innovation strategies), we remove firm-year observations

associated with going-public (1,341 IPOs) or going-private transactions (92 deals). In other words, our

sample private firms always stay private and our sample public firms always stay public.4 Public firms are

those traded on the NYSE, AMEX, or NASDAQ. The full sample consists of 50,294 public firm-year

observations (9,163 unique public firms) and 16,217 private firm-year observations (6,549 unique private

firms) for the period 1997-2008. The top five public firm industries are: Business Service (17.2%),

Pharmaceutical Products (6.9%), Retail (6.2%), Electronic Equipment (5.7%), and Wholesale (4.9%). The

top five private firm industries are: Business Service (15.7%), Utility (12.5%), Wholesale (7.9%),

Communication (6.7%), and Retail (4.6%).

Our focus is on corporate innovation strategy, so we need to examine a sample of innovative

firms (i.e., firms with patents), similar to Akcigit and Kerr (2012), Bloom, Schankerman, and Van Reenen

(2013), and Aghion, Van Reenen, and Zingales (2013). Innovative firms in year t are identified as firms

filing at least one patent over the three-year period from year t-2 to year t during the sample period 1997-

2008. The innovative firm sample consists of 13,463 public firm-year observations representing 2,426

unique public firms and 1,729 private firm-year observations representing 829 unique private firms for

the period 1997-2008.

3.2. Measures of Innovation Strategy

3.2.1. Exploratory versus exploitative patents

Our measures of innovation strategy—exploratory and exploitative patents—come from the

management literature. Following Benner and Tushman (2002), we use a firm’s existing expertise—the

4 It is worth noting that including those going-public/going-private firms in our analyses does not change our main findings. Further, those going-public firms will be used in our identification tests later on.

12

combination of its portfolio of patents and citations made by its existing patents over the past five years—

to characterize the nature of its innovative effort. Exploratory innovation goes beyond the existing

expertise, while exploitative innovation deepens the existing expertise. A patent is categorized as

“exploratory” if 60% (80%) or more of its citations are based on new knowledge outside of a firm’s

existing expertise (i.e., not citing the firm’s existing patents or the citations made by those patents); while

a patent is categorized as “exploitative” if 60% (80%) or more of its citations are based on a firm’s

existing expertise (i.e., the firm’s existing patents and the citations made by those patents).

At the firm level, the variable Explore60 (Explore80) in year t is the number of exploratory

patents applied in year t-2 to year t divided by the total number of patents applied in the same three-year

period. We use the application year (instead of the award year) to better capture the exact timing of the

underlying innovative activities behind a patent. This variable ranges between zero and one, and a higher

exploratory ratio suggests that a firm’s innovation effort focuses on expanding its technological expertise

by deviating from its current innovation trajectories and creating patents that may lead to new competitive

advantages in totally different territories.

At the firm level, the variable Exploit60 (Exploit80) in year t is the number of exploitative patents

applied in year t-2 to year t divided by the total number of patents applied in the same three-year period.

The exploitative ratio ranges between zero and one, and a higher exploitative ratio suggests that a firm’s

innovation effort focuses on deepening its technological expertise by following its current innovation

trajectories and developing patents that extend its competitive advantages in existing territories.

For a robustness check, we also employ two related alternative measures to capture the nature of

innovation, following Katila and Ahuja (2002): innovation scope and depth. Innovation scope intends to

capture the frequency a firm acquires new knowledge outside of its existing knowledge: The more often a

firm has acquired new knowledge, the broader knowledge scope it has. A firm’s existing knowledge

consists of citations made by its existing patents over the past five years. Our first alternative measure, the

variable Scope in year t is the number of new citations made by patents applied in year t-2 to year t

divided by the total number of citations made by all patents applied in the same three-year period. New

citations are citations that have never been made by the firm in the past five years. The scope variable

13

ranges between zero and one, and a higher value indicates that a firm acquires more new knowledge in

building its patent portfolio and thus may have a better chance to achieve new competitive advantages.

Innovation depth intends to capture a firm’s repetitive application of its existing knowledge: The

more often a firm has applied its existing knowledge, the more deeply it understands it. Our second

alternative measure, the variable Depth in year t is the number of repeated citations made by patents

applied in year t-2 to year t divided by the total number of citations made by all patents applied in the

same three-year period. Repeated citations are citations that have been made by the firm in the past five

years. The depth variable is equal to or greater than zero,5 and a higher value indicates that a firm

repetitively exploits its existing knowledge and thus may have a deeper understanding of its current

technological territories. More detailed discussions on constructing Explore60, Explore80, Exploit60,

Exploit80, Scope, and Depth are provided in Appendix 1.

3.2.2. Patent count

Patent count is the number of awarded patents applied by firm i in year t-2 to year t. This measure

has been widely used in the literature to capture the success of a firm’s innovative activities since Scherer

(1965) and Griliches (1981). We set this variable to zero for firm-year observations in the full sample

without any patent record. This variable is always greater than zero in the innovative firm sample by

construction.

Assessment of the innovation output of a given firm using patent count can only be made relative

to some benchmark for the following reasons. First, technology classes differ in the nature of R&D

activities and resources required to produce patentable innovation such that patent counts in two different

classes may not be directly comparable. Second, there are technology class-specific time trends in the

number of awarded patents that may not fully reflect changes in innovation output. In particular, large

increases in the number of awarded patents in some classes over time might reflect evolution of the US

Patent and Trademark Office (USPTO) practices with respect to what is patentable innovation, and hence

5 The depth ratio could be above one because some current citations may have been made more than once in the past. For example, if a firm has only one patent that is filed in year t and only cites one prior patent, and if this citation has been made by two patents filed by the firm over the past five years, then the depth ratio is two for this firm in year t.

14

patent counts from different years may not be time-consistent measures of innovation output even within

the same technology class.

We address both issues by summing up scaled number of patents, where we divide the number of

awarded patents in technology class k with application year t by the average number of awarded patents

applied by both public and private firms in the same technology class and year, following Bena and Li

(2012) and Seru (2014). Our adjusted patent count thus accounts for differences across technology classes

and differences in the USPTO practices over time. More detailed discussions on constructing Patent

count and Patent countCIQ are provided in Appendix 1.

As a robustness check, we also employ the variable Citation count, capturing the impact of

corporate innovation and its associated economic value (Trajtenberg, 1990; Hall, Jaffe, and Trajtenberg,

2005b; Aghion, Van Reenen, and Zingales, 2013). We need to make some adjustment for citation counts

because patents continue to receive citations over a long period of time, while we observe at best only up

to the last year of the available data (in our case the year of 2010). As a result, patents awarded in

different years suffer to different extent from this truncation bias in the number of citations received, and

their citation counts are not comparable and cannot be aggregated. Further, patents awarded in different

technological classes are also cited in different ways, which needs to be accounted for before aggregating.

We make adjustments to citation count following Hall, Jaffe, and Trajtenberg (2005a, henceforth HJT),

Lerner, Sørensen, and Strömberg (2011), Bena and Li (2012), and Seru (2014). More detailed discussions

on constructing Citation count and Citation countCIQ are provided in Appendix 1. In addition to these two

flow measures of innovation output, we also employ a stock measure of a firm’s patent portfolio, Patent

stock, which is the total number of granted patents applied by firm i up to year t-3.

3.3. Summary Statistics

Table 1 provides the temporal distribution of our sample. We show that there is a deceasing trend

in the number of public firms over the sample period, while there is an increasing coverage by Capital IQ

of private firms over the same period. About 30% of public firms have awarded patents, while only about

a tenth of private firms have awarded patents. The univariate statistics are suggestive of public firms

being more innovative than private firms. We further note that starting 2007 (2008), there is a decline in

15

the fraction of public (private) firms with awarded patents, probably because of the time lag for an applied

patent to be eventually awarded.

Table 2 provides summary statistics of our sample. Definitions of all variables are provided in

Appendix 1. All dollar values are in 2008 dollars. All continuous variables are winsorized at the 1st and

99th percentiles. Panel A compares innovation measures and firm characteristics between public and

private firms in the full sample. We first show that public firms are more innovative in terms of

innovation output (Patent count). On average, public firms produce 13.60 patents per three-year period,

while private firms produce 2.28 patents per three-year period. Similarly, the mean patent count (adjusted

by CIQ firm averages, Patent countCIQ) for public firms is 3.48, while the mean for private firms is 0.69.

The two-sample tests of differences (i.e., the t-test and Wilcoxon-test) reject the null that public and

private firms have the same level of innovation output at the 1% level.

The mean HJT-adjusted citation counts (Citation count) show that on average, public firms’

patents receive a total of 48.67 citations over their life time, while private firms’ patents receive 6.63 total

citations. Once we limit the window to count citations to the award year and subsequent three years, the

mean citation count (adjusted by CIQ firm averages, Citation countCIQ) for public firms’ patents is 15.05,

while the mean for private firms’ patents is 2.34. The difference is statistically significant at the 1% level

in both the t-test and Wilcoxon test, rejecting the null that public and private firms have the same level of

innovation output.

On average, public firms have a total of 69.19 patents (Patent stock), while private firms have a

total of 11.67 patents. The difference is statistically significant at the 1% level in both the t-test and

Wilcoxon test, rejecting the null that public and private firms have the same number of patent stock. In

terms of comparing firm characteristics between public and private firms,6 we show that public firms are

significantly larger than private firms. Further, we show that on average, public firms have significantly

lower leverage than private firms, consistent with the general belief that public firms have better access to

equity financing and thus rely less on debt financing. Finally, public firms have significantly faster sales

growth, higher capital expenditures, and are significantly older than private firms; while private firms

6 Note that none of the correlations among firm characteristics is high enough to warrant multi-collinearity concerns for our multivariate analyses.

16

have significantly higher fixed assets as measured by property, plant and equipment (PPE) than public

firms. Importantly, public firms spend significantly more on R&D than private firms.

Panel B compares innovation measures and firm characteristics between public and private firms

in the innovative firm sample. Again in this comparison, we show that public firms have significantly

more patent counts, citation counts, and patent stock than private firms, which is consistent with findings

in Acharya and Xu (2013). New in this panel, we examine innovation strategies in public and private

firms. We show that on average, using the variable Explore60 (Explore80), 51% (48%) of public firms’

patents are exploratory, while 60% (58%) of private firms’ patents are exploratory. Such a difference is

both economically important (representing about a 15% drop in exploratory patents when firms change

listing status from private to public) and statistically significant at the 1% level, suggesting that private

firms focus more on exploring new technological territories than public firms do. In contrast, using the

variable Exploit60 (Exploit80), 14% (12%) of public firms’ patents are exploitative, while 10% (9%) of

private firms’ patents are exploitative. Again such a difference is both economically important

(representing about a 25% increase in exploitative patents when firms change listing status from private to

public) and statistically significant at the 1% level, indicating that public firms’ innovation is more

exploitative than private firms’. Comparing innovation scope and depth between public and private firms,

we find that the average scope (depth) for public firms is 0.75 (0.21), while the average scope (depth) in

private firms is 0.81 (0.12). These differences are also significant at the 1% level, indicating that public

firms’ innovations tend to rely more on the existing knowledge (i.e., more exploitative) but private firms’

innovations tend to go beyond the existing knowledge domain (i.e., more exploratory). Overall, the

summary statistics in this panel provide some support for the short-termism hypothesis that public

ownership is associated with more exploitative innovation and less exploratory innovation.

In terms of comparing firm characteristics between innovative public and private firms, we show

that in addition to the differences we have shown before between public and private firms in general,

innovative public firms have significantly better operating performance than innovative private firms,

while both groups of innovative firms have similar sales growth.

Next we employ multivariate analysis to test our hypotheses.

17

4. Main Results

4.1. Exploratory versus Exploitative Innovation

So far, we have shown that public firms on average are more innovative than private firms in

terms of innovation output, it is premature to rave about public ownership given that there are many more

private firms than public firms contributing significantly to the US economy (Asker, Farre-Mensa, and

Ljungqvist, 2013). We need to take a careful look at these two groups of firms’ innovation strategies as

captured by the exploitative versus exploratory nature of innovation.

To empirically test our two hypotheses, we run the following panel data regression (for brevity,

we omit firm subscript i and time subscript t, and t is from 1997 to 2008):

& ,

(1)

where denotes one of the four measures for innovation strategies—Explore60,

Explore80, Exploit60, and Exploit80 of firm i in year t (which is computed using patent information over

the three-year period from year t-2 to year t). Public is an indicator variable that takes the value of one if

firm i is a public firm in year t, and zero otherwise. We include the following set of explanatory variables;

all firm financials are measured in year t-2. is the logarithm of the total number of

granted patents owned by firm i at the end of year t-3 and it controls for the size of a firm’s patent

portfolio and existing knowledge; is the logarithm of total assets of firm i controlling

the effect of firm size on innovation strategies; is the ratio of total debt to total assets

controlling for the effect of financial leverage on innovation strategies; is return on assets capturing

the influence of profitability on innovation strategies; is the growth rate of sales

controlling for business prospects and growth opportunities; is the ratio of capital expenditures to

total assets capturing the effect of physical investments on innovation strategies; & is the ratio of R&D

expenditures to total assets capturing the effect of intangible investments on innovation strategies; is

the ratio of net property, plant, and equipment to total assets controlling for asset tangibility; and

18

is the logarithm of firm i’s vintage capturing the effects of firm life cycle and information

asymmetry on innovation strategies.

We also include industry times year fixed effects to control for unobserved industry- and year-

specific heterogeneity such as competitive pressure and industry-specific business cycles. Industries are

based on Fama and French’s (1997) 48 industry classification. We do not include firm fixed effects

because our main variable of interest, , largely overlaps with firm fixed effects.7 The coefficient on

the Public indicator variable thus measures the difference in innovation strategies between public and

private firms that cannot be accounted for by differences in firm characteristics and industry and year

fixed effects. Table 3 presents the regression results based on Equation (1). Our statistical inferences are

based on standard errors clustered at the firm level that correct for time-series correlation in firm-level

innovation strategies, if any.

Panel A employs the innovative firm sample. Columns (1) and (2) compare public and private

firms in the exploratory nature of innovation. We show that the coefficients on the Public indicator

variable are negative and significant at the 1% level. In terms of the economic significance, an average

public firm produces 5%-6% fewer exploratory patents than an average private firm, consistent with the

short-termism hypothesis. This suggests that private firms are more likely to go outside of the existing

expertise in their innovation effort. We also show that patent stock is negatively associated with

exploratory patents. This is not surprising as firms with bigger patent portfolios, ceteris paribus, have

larger pools of existing knowledge (their own patents or past citations) for them to exploit and are thus

less likely to cite outside their existing knowledge. We further show that large firms and firms with good

operating performance are associated with more exploratory patents. Interestingly, we do not find any

significant association between R&D and exploratory innovation.8 Lastly, we show that older firms tend

to produce more exploratory patents. These results suggest that larger and older firms tend to reinvent

themselves to explore new technological territories.

7 In our main analysis involving the innovative firm sample, all sample firms remain either public or private throughout the sample period. Nonetheless, when we conduct similar analysis using a sample of IPOs with firm fixed effects, we obtain consistent results. 8 This finding is, however, consistent with the literature that higher R&D expenditures do not necessarily lead to more innovation. Jaffe (2000) and Lanjouw and Schankerman (2004) report that the surge of R&D investment since the 1980s does not generate commensurate patents. The Economist (1990) also notes that “American industry went on an R&D spending spree, with few big successes to show for it.”

19

Columns (3) and (4) compare public and private firms in the exploitative nature of innovation.

We show that the coefficients on the Public indicator variable are positive and significant at the 1% level.

In terms of the economic significance, an average public firm produces 3% more exploitative patents than

an average private firm, consistent with the short-termism hypothesis. This suggests that public firms are

more likely to innovate within their existing expertise. We also show that patent stock is positively

associated with exploitative patents. This is again not surprising as firms with bigger patent portfolios,

ceteris paribus, are more likely to cite their own patents or past citations. We further show that firms with

poor operating performance are associated with more exploitative patents, suggesting that firms cannot

afford risky projects due to the short-term pressure from weakened profitability. Lastly, we show that

older firms tend to produce fewer exploitative patents.

We also estimate Equation (1) using private firms and their matched public control firms because

prior work has shown that patenting activities tend to vary systematically by industry (Hall et al., 2005a)

and larger firms tend to have more patents due to economies of scale (Scherer, 1965). Given that

innovative private firms are typically smaller than innovative public firms (as shown in Table 2 Panel B),

we match each private firm with a public firm in the same industry, year, and closest in total assets

following the approach in Gao et al. (2013). To ensure that the matched public firm is truly comparable to

the private firm, we also require that the initial size of the matched public firm be within [50%, 150%] of

the size of the private firm. The matching (without replacement) procedure helps mitigate large

differences in the size distribution between the public and private firm samples and the smaller, but

potentially important difference in sample firm distribution across industries. Panel B presents the

regression results based on Equation (1) using the private firm sample and their industry- and size-

matched public control firms.

We show that compared to innovative private firms, matched innovative public firms produce

significantly fewer exploratory patents and significantly more exploitative patents, consistent with the

short-termism hypothesis. Again, we show that patent stock is negatively associated with exploratory

innovation and positively associated with exploitative innovation. Further, we show that high leverage is

negatively associated with exploratory innovation, while large and old firms produce significantly fewer

exploitative patents.

20

In Appendix 2, we use innovation scope and depth (Katila and Ahuja, 2002, defined earlier in

Section 3.2.1) as alternative measures of the exploratory and exploitative nature of innovation. Innovation

with broader scope (greater depth) can be broadly viewed as producing more exploratory (exploitative)

patents. We find that public firms’ innovation is of narrower scope but of greater depth than private firms’

innovation, broadly consistent with the findings in Table 3 and in support of the short-termism

hypothesis.

Overall, we show that public firms focus more on exploiting their existing expertise than private

firms, which may help public firms to maintain competitive advantages and appropriate economic rents

from their existing technologies. In contrast, we show that private firms are more likely to expand their

expertise and to create new competitive advantages compared to public firms. All results are in strong

support of the short-termism hypothesis that private firms’ patents are more exploratory (less exploitative)

than public firms’.

5. Dealing with Self-Selection

Being public, of course, is not an exogenous event: Most firms stay public (or private) for reasons

that correlate with their financing or investment decisions (see for example, Brav, 2009; Asker et al.,

2013; Gao et al., 2013). To account for the possible selection effect, we adopt a number of different

approaches: 1) implementing propensity score-matching based on observable firm characteristics; 2)

using a transitioning IPO sample and their matched private (public) control firms; and 3) employing an

instrumental variable (IV) to estimate a selection equation.

5.1. The Propensity Score-Matched Sample

We employ a matching technique to examine the differences in innovation strategies between

public and private firms. The matching procedure controls for selection based on observable firm

characteristics. Our data are well suited to the matching approach given that we have a much larger pool

of potential matches in the innovative public firm sample compared to the treatment group—the

innovative private firm sample (2,426 unique public firms versus 829 unique private firms), which

increases the likelihood of finding close matches for the private firms among the public firms.

21

The matching procedure that we employ is a one-to-one nearest neighbor matching without

replacement (Heckman, Ichimura, and Todd, 1997). The matching starts with a probit regression where

the public firm indicator variable is the dependent variable, using three different specifications to capture

the choice between being a public or private firm (Bharath and Dittmar, 2010; Gao et al., 2013). In Model

1, the variables we use to match are Ln(Total assets), and industry times year fixed effects. In Model 2,

the variables we use to match are Ln(Total assets), Ln(Patent stock), and industry times year fixed effects.

In Model 3, the variables we use to match are all the explanatory variables in Table 3 Panel A except that

we do not include the Public indicator variable. Then using the predicted probabilities—propensity

scores—from the estimated probit regressions, we match to each private firm-year observation a public

firm-year observation that minimizes the absolute value of the difference between propensity scores.

Table 4 presents differences in measures of innovation strategies between the propensity score-

matched public firms and private firms using three different matching criteria. We find that there are

statistically significant differences between public and private firms in their innovation strategies. Public

firms are associated with significantly fewer exploratory patents, and significantly more exploitative

patents compared to private firms, consistent with the short-termism hypothesis.

5.2. Using IPO Firms and Their Matching Private (Public) Control Firms

We also examine innovation strategies for a set of firms during our sample period that undergo a

transition in ownership status from private to public. Using the transitioning sample together with firm

fixed effects directly addresses the sample selection concern because we compare the same firm as both a

private and public firm, controlling for selection based on the time-invariant unobservable firm

characteristics.

Prior work has shown that IPOs tend to cluster in time and in industry (Pagano, and Panetta, and

Zingales, 1998; Lowry, 2003), and that sales, capital expenditures, and other performance variables

exhibit a consistently increasing pattern over the years before and after the IPO (Chemmanur, He, and

Nandy, 2010). We use matched private (public) control firms to difference out these observable factors to

capture the effect of going public on innovation strategies.

22

There are 392 IPOs involving innovative private firms over the sample period 1997-2008 so that

we can compare these transitioning firms’ innovation strategies before versus after IPOs. For each IPO

firm we find an innovative private (public) control firm that is in the same industry and has a similar

propensity to go public in the year prior to the IPO using a set of firm characteristics including firm size,

capital expenditures, R&D, sales growth, leverage, ROA, and firm age that are shown to be important in

the going public decision (Chemmanur et al., 2010).9 We also require that those private (public) control

firms stay private (public) over the sample period 1997-2008.

We then run the following panel data regression employing the IPO firms and their matched

private (public) control firms over the period from (up to) five years before to (up to) five years after the

IPO:

& ,

(2)

where denotes one of the four measures for innovation strategies—Explore60,

Explore80, Exploit60, and Exploit80—measured over the period from year t-2 to t. When comparing the

IPO firms with their matched private control firms, Event is PostIPO, an indicator variable that takes the

value of one if the firm-year observation is after the IPO year for the IPO firm in year t-2, and zero

otherwise. When comparing the IPO firms with their matched public control firms, Event is PreIPO, an

indicator variable that takes the value of one if the firm-year observation is before the IPO year for the

IPO firm in year t-2, and zero otherwise. We include the firm fixed effects in order to examine within-

firm variations under the public/private status. We also include the year fixed effects to account for

variations over time associated with market conditions that may influence the IPO timing. The coefficient

on the PostIPO (PreIPO) indicator variable captures the change in innovation strategies around the IPO

9 The matching criterion in Table 5 is different from that used in Table 4 because in the former, we aim to estimate the propensity to go public; while in the latter, we aim to match steady-state private firms with steady-state public firms.

23

after controlling for the firm and year fixed effects (a difference-in-differences estimation approach).10

Table 5 presents the regression results based on Equation (2).

In Panel A, we compare the IPO firms with their matched private control firms. In columns (1)

and (2) where the dependent variable is Explore60 and Explore80, respectively, we show that the

coefficients on PostIPO are all negative and significant. This finding indicates that compared to their

private control firms, the IPO firms decrease the number of exploratory patents after their transition from

private to public status. In columns (3) and (4) where the dependent variable is Exploit60 and Exploit80,

respectively, we show that the coefficients on PostIPO are all positive and significant. This finding

indicates that compared to their private control firms, the IPO firms increase the number of exploitative

patents after their transition from private to public status. We note that most of the within-firm changes in

firm characteristics are consistent with the patterns uncovered from the cross-sectional regression in Table

3 Panel A.

In Panel B, we compare each IPO firm to a steady-state public firm based on the same matching

method as in Panel A. Consistent with Panel A, we find that compared to their public control firms, IPO

firms have significantly more exploratory patents and significantly fewer exploitative patents in the years

before going public. Bernstein (2012) finds that patent quality declines after firms going public. Our

evidence on changes in innovation strategies after IPOs complements his findings and echoes his point

that there is a complex trade-off between public and private ownership.

To address the reverse causality concern that it is the innovation strategy that drives the going

public decision, but not the going public decision affecting the innovation strategy, in Appendix 3, we use

different measures of innovation strategy in addition to firm characteristics to predict the going public

decision. The sample includes the 392 IPO firms and the innovative private firm sample. We show that

none of the innovation strategy variables is significantly associated with a firm’s likelihood of going

public. We conclude that our main findings are unlikely to be mostly driven by reverse causality.

5.3. Using the Instrumental Variable Approach

10 Chemmanur et al. (2010) use a similar approach to study the change in total factor productivity and other firm characteristics around IPOs.

24

To account for the possible selection issue, we employ an instrumental variable (IV) approach by

modelling the innovation strategy measure and the public status of a firm as follows:11

,

∗ ,

1, ∗ 0; 0, . (3)

The dependent variable is one of the four innovation strategy variables used before. X is a list of firm-

level control variables used in earlier regressions. The coefficient of key interest is β1, on the public firm

indicator variable, Public. The variable Public* indicates the latent propensity of a firm staying public.

For the purpose of identification, we need instrumental variables that affect a firm’s propensity of staying

public, but do not affect its innovative activities directly other than through the effect of being public.

That is, the vector of Z in Equation (3) must contain variables in addition to a full overlap with the vector

of X. The public firm indicator variable is allowed to be endogenous in the sense that corr(, ) ≠ 0. A

positive (negative) association indicates that explanatory (exploitative) innovation in public firms is

greater (lower) based on unobservable heterogeneity. Thus, an estimate for β1 is upward (downward)

biased if the endogeneity is not properly accounted for.

To allow for time-varying unobserved heterogeneity across firms, we resort to the treatment

regression using the maximum likelihood estimator given in Maddala (1983, Chapter 5), where the Public

indicator variable is treated as endogenous.

Following Gao et al. (2013), our instrumental variable is the industry-level underwriter

concentration (constructed as the number of IPOs by the top five lead underwriters in a given industry

divided by the total number of IPOs in that industry) to proxy for the high cost of going public—IPO

underpricing (Liu and Ritter, 2011). We compute the underwriter concentration measure in year t using

the industry-level IPO data over year t-2 to year t to mitigate the concern that IPOs tend to be lumpy.12

11 See Li and Prabhala (2007) for an overview of dealing with selection issues versus treatment effects in corporate finance. 12 In Appendix 4, we run a modified version of Equation (1) by adding our instrumental variable as an explanatory variable. We show that the industry-level underwriter concentration variable is not significantly associated with any of the innovation strategy measures.

25

The results from estimating the treatment regression in Equation (3) are reported in Table 6. The

identification relies on both the instrumental variable and the non-linearity of the propensity of going

public. As shown in column (1) where the dependent variable is the Public indicator variable, the

coefficient on the time-varying industry-level underwriter concentration variable is negative and

significant, suggesting that low competition among underwriters increases firms’ IPO costs and thus

deters them from going public.

In columns (2)-(5) where the dependent variables are different measures of innovation strategy,

the coefficient on the Public indicator variable captures the effect of being a public firm on innovation

strategy, taking into account the possible selection of staying public decisions. We show that the effects

are negative and significant when the dependent variable is the ratio of exploratory patents (columns (2)

and (3)); and the effects are positive and significant when the dependent variable is the ratio of

exploitative patents (columns (4) and (5)). The size of the coefficient is much larger than that of the

corresponding coefficient in Table 3 Panel A without controlling for sample selection. Moreover, the

exogeneity test rejects the null hypothesis that = 0 at the 1% level.

These results indicate that there is a selection issue associated with being a public firm. In our

sample, firms that stay public tend to have more exploratory (fewer exploitative) patents based on

unobserved heterogeneity (i.e., is positive in columns (2) and (3) and negative in columns (4) and (5)).

To the extent that the treatment regression framework is valid, such a selection effect makes the

interpretation of a causal effect stronger as it renders the effect of being public under-estimated using an

OLS regression (as in Table 3 Panel A). We conclude that it is less likely that our main findings are

entirely attributed to the selection issue.

6. Additional Investigation

So far, we have provided evidence in support of the short-termism hypothesis. The natural

question is how managerial short-termism works through the corporate decision making process to end up

in influencing long-term high-risk investments like exploratory innovation. In this section we run several

tests to explore the cross-sectional variation within public firms in managers’ exposure to short-term

performance pressure that help establish the underlying mechanisms through which public firm status

26

may deter exploratory innovation. We also examine one other possible channel beyond financing and

managerial short-termism to explain our findings.

6.1. The Vesting Horizon of CEOs’ Equity Portfolios

Edmans et al. (2013) and Gopalan et al. (2014) show that when the vesting period of a CEO’s

stock and option grants is short, she is more concerned about the short-term performance and thus be

more short-term-oriented. Similar to their approach, we first compute a CEO’s vested equity portfolio as

the value of her vested stock and option grants normalized by the value of her portfolio of stock and

option grants. Based on the sample median of CEOs’ vested equity portfolios, we split the innovative

public firm sample into the high short-termism public firm subsample (where the CEO’s vested equity

portfolio is above the sample median) and the low short-termism public firm subsample (where the

CEO’s vested equity portfolio is below the sample median). We then estimate Equation (1) by replacing

the Public indicator variable by the new high and low short-termism public firm indicator variables. Table

7 Panel A presents the results.13

In columns (1) and (2), we show that the coefficients on both the high and low short-termism

public firm indicator variables are negative and significant, indicating that both groups of the public firms

produce significantly fewer exploratory patents than the private firms. More importantly, the effect of the

high short-termism public firm indicator variable on exploratory innovation is significantly larger than

that of the low short-termism public firm indicator variable. Take column (2) for example, the coefficient

on the high short-termism public indicator variable is −0.068 and significant at the 1% level, while the

coefficient on the low short-termism public indicator variable is −0.044 and significant at the 1% level.

The F statistic reported at the bottom indicates that these two coefficients are significantly different at the

1% level.

In columns (3) and (4), we find that while both groups of the public firms produce significantly

more exploitative patents than the private firms, this difference is more pronounced between the high

short-termism public firms and the private firms than that between the low short-termism public firms and

13 The information on CEOs’ vested (total) stock and option grants is obtained from ExecuComp. Because ExecuComp only covers S&P1500 firms, the sample size in this panel is smaller than that used in Table 3.

27

the private firms. Take column (4) for example, the coefficients on the high and low short-termism public

indicators are 0.053 and 0.043, respectively. This result indicates that the high short-termism public firms

produce 5.3% more exploitative patents than the private firms, while the low short-termism public firms

produce 4.3% more exploitative patents than the private firms. The F statistics indicates that these two

coefficients are significantly different at the 5% level.

Our results suggest that the vesting horizon of a CEO’s equity portfolio, which is associated with

her short-term incentives, may be an underlying force behind our findings that public firms do less

exploratory (more exploitative) innovation than private firms.14

6.2. The Market for Corporate Control

In the myopia model of Stein (1988, 1989), in the presence of asymmetric information,

shareholders and the stock market cannot properly evaluate managers’ investment in long-term high-risk

projects, resulting in those firms to be more likely undervalued by the market, leading to their greater

vulnerability to takeovers. To prevent misevaluation and/or from being taken over, managers are more

likely to invest more in short-term low-risk projects instead of long-term risky projects. Given that

exploratory innovation is typically more long-term and high-risk in nature, we expect that the pressure

from the market for corporate control could be another underlying force behind public firms’ low

exploratory innovation.

To capture a firm’s vulnerability to the market for corporate control, we estimate the sensitivity of

a firm’s likelihood to be taken over to its degree of mis-valuation. We obtain a sample of announced

mergers and acquisitions (M&As) deals from the SDC database over the sample period 1997-2008.

Following Cremers et al. (2009) and Fang et al. (2013), we estimate a firm’s likelihood of being taken

over by running a probit regression with the following explanatory variables: M/B, firm size, ROA,

leverage, cash holdings, sales growth, and year fixed effects. We estimate this model for public firms

industry by industry (based on the 48 Fama-French industries) and obtain the coefficient on the M/B ratio

14 Malmendier and Tate (2005) note that overconfident CEOs delay exercising their vested stock and option grants, hence mitigating the short-term incentives associated with their holdings of vested stock and option grants. Their findings actually strengthen our results because we do not exclude overconfident CEOs in our sample who are less sensitive to short-term incentives provided by their vested stock and option grants.

28

at the industry-level. The (mostly negative) coefficients on M/B capture different sensitivities of a firm’s

takeover likelihood to its M/B ratio within its industry: the more negative is the coefficient on M/B, the

greater the likelihood for a firm to be taken over for a given M/B value.

Based on the industry median M/B-takeover sensitivity, we split the innovative public firm

sample into the high short-termism public firm subsample (where the coefficient on M/B is less than the

sample median) and the low short-termism public firm subsample (where the coefficient on M/B is

greater than the sample median). Then we estimate Equation (1) by replacing the Public indicator variable

by the new high and low short-termism public firm indicator variables. Table 7 Panel B presents the

results.

In columns (1) and (2), we show that the coefficients on both the high and low short-termism

public firm indicator variables are negative and significant, indicating that both groups of the public firms

produce significantly fewer exploratory patents than the private firms. More importantly, the effect of the

high short-termism public firm indicator variable on exploratory innovation is significantly larger than

that of the low short-termism public firm indicator variable. The F statistic reported at the bottom

indicates that these two coefficients are significantly different at the 1% level. In columns (3) and (4), we

find that both groups of the public firms produce significantly more exploitative patents than the private

firms but the difference between the high short-termism public firms and the private firms is more

pronounced than that between the low short-termism public firms and the private firms.

Our results suggest that the market for corporate control, which is associated with the short-term

pressure, may be another underlying force behind our findings that public firms do less exploratory (more

exploitative) innovation than private firms.

6.3. Transient Institutional Investors

Bushee (1998) shows that transient institutional investors create short-term pressure on managers

to cut R&D expenditures in order to meet short-term earnings targets. Following Bushee’s (1998, 2001)

definition of transient institutional investors, we first compute the transient institutional ownership as the

number of shares owned by transient institutional investors normalized by the number of shares owned by

all institutional investors. We then sort our innovative public firms into the high short-termism public

29

firm subsample and the low short-termism public firm subsample based on the sample median of the

transient institutional ownership, with the former greater than and the latter less than the median. We

estimate Equation (1) by replacing the Public indicator variable by the new high and low short-termism

public firm indicator variables. Table 7 Panel C presents the results. We show that both groups of the

public firms produce significantly fewer exploratory (more exploitative) patents than the private firms,

and the difference in exploratory (exploitative) patents is greater between the high short-termism public

firms and the private firms than that between the low short-termism public firms and the private firms.

Our results suggest that the presence of the transient institutional investors, which is associated

with the short-term pressure, may be the third underlying force behind our findings that public firms do

less exploratory (more exploitative) innovation than private firms.

6.4. Can Differences in Information Environment Be Another Driving Force?

Public firms are required to disclose more information than private firms, leading public firms to

be more transparent. This difference in information environment may play a role in explaining their

different innovation strategies. On the one hand, firms tend to focus on exploitations when they know that

their innovation effort is closely scrutinized and do not want to share the knowledge from their

explorations with other firms (Cohen and Levinthal, 1994). Given that public firms’ innovative activities

are closely watched by the market participants and their peer firms (Bhattacharya and Ritter, 1983;

Aggarwal and Hsu, 2013), public firms may strategically choose to focus on exploitative innovation than

exploratory innovation.

On the other hand, greater disclosure reduces the information asymmetry between corporate

insiders and outsider, which in turn reduces cost of financing and lowers the likelihood of the firm being

undervalued by the stock market and thus being taken over. Therefore, greater transparency can facilitate

public firms to engage in long-term exploratory innovations. A priori, it is unclear whether more

transparent information environment of public firms makes them focus more or less on exploratory

innovations than private firms.

To empirically examine the relation between the quality of firm information environment and

innovation strategy, we divide our innovative public firm sample into the high and low analyst coverage

30

subsamples based on the sample median number of analyst earnings forecasts. We then estimate Equation

(1) by replacing the Public indicator variable by the new high and low analyst-coverage public firm

indicator variables. Table 8 presents the results.

We find that while both groups of the public firms have significantly fewer exploratory and more

exploitative patents than the private firms, this difference is more pronounced between the low analyst-

coverage (i.e., less transparent) public firms and the private firms than that between the high analyst-

coverage (i.e., more transparent) public firms and the private firms.

Overall, these results suggest that more transparent information environment in public firms

actually encourages these firms to focus more on exploratory innovation and less on exploitative

innovation compared to private firms. Thus, the difference in information environment between public

and private firms does not explain our main finding that public firms’ innovation is less exploratory and

more exploitative than private firms’ innovation.

6.5. Can the Entrepreneur’s Risk Preference Be Another Driving Force?

It is natural to expect that private firm CEOs are more likely to be entrepreneurs, while public

firm CEOs are more likely to be professional managers. Given that entrepreneurs could be more risk-

loving than professional managers, can this difference in risk preferences of public and private firm CEOs

explain these firms’ differences in innovation strategies? We employ two approaches to investigate this

possibility.

First, in Appendix 5 Panel A, we re-estimate Table 3 Panel A using a subsample of firms

incorporated prior to 1980. A firm incorporated prior to 1980 is more likely to be a mature firm by the

2000s (our sample period) and less likely to be in the “entrepreneurial” stage of its life cycle. We find

that in this subsample of mature firms, public firms still have more exploitative and less exploratory

innovation than private firms. Second, in Appendix 5 Panels B and C, we re-estimate Table 5 using a

subsample of IPOs whose founders keep the CEO position till year +5 after the IPOs (108 IPO cases),

which allows us to clearly delineate the impact of listing status on innovation strategy by keeping the

entrepreneur’s risk preference constant. We still find that after going public, these firms’ innovation

become less exploratory and more exploitative. We conclude that our findings of more exploratory and

31

less exploitative innovation in private firms than in public firms are unlikely to be mainly driven by their

CEOs’ risk preferences.

7. Conclusions

This paper examines the effect of short-termism associated with public capital market on

corporate innovation strategies by exploiting a new dataset on private US firm patents and citations.

Using a large sample of public and private firms and their patent records over the period 1997-2008, we

show that public firms appear to be more innovative than private peer firms using a number of measures

including patent and citation counts. More importantly, we show that private firms are more novel in their

innovation strategies with patents more exploratory and broader in scope, while public firms are more

conservative in their innovation strategies with patents more exploitative and relying more on their

existing knowledge. The evidence from propensity score matching, a sample of IPO firms and their

matched private (public) control firms, and the instrumental variable approach suggests that these findings

are unlikely driven by endogeneity. We further show that the vesting horizon of CEOs’ equity portfolios,

the market for corporate control, and the presence of short-term investors are likely to be the underlying

forces behind different innovation strategies between public and private firms.

The findings of our paper provide new insights into a number of important corporate finance

questions: How does ownership structure—public versus private—affect corporate innovation? Do IPOs

really encourage high-tech firms to pursue exploratory innovation? Should we change regulations to limit

the market for corporate control? Further theoretical research and empirical evidence are called for and

answers to these questions are relevant not only to firm stakeholders (such as shareholders, creditors,

managers, and inventors) but also to policy makers and governments.

32

Appendix 1 Variable definitions

Variable Definition

Firm Characteristics

Public An indicator variable that takes the value of one if the firm is a public firm, and zero otherwise.

Leverage Total debt normalized by total assets.

ROA Return on assets, computed as EBIT normalized by total assets.

Sales growth Sales (t-2) / Sales (t-3) – 1.

Capex Capital expenditures normalized by total assets.

R&D R&D expenditures normalized by total assets.

PPE Net property, plant, and equipment normalized by total assets.

Firm age Firm age since a firm’s incorporation.

Underwriter concentration Based on completed IPOs from 1995-2008, the variable in year t is computed as the ratio of the number of IPOs by the top five lead underwriters in a given industry to the total number of IPOs in that industry from year t-2 to year t.

IPO An indicator variable that takes the value of one if a private firm goes public in year t, and zero otherwise.

PostIPO An indicator variable that takes the value of one if the firm-year observation is after the IPO (year 0) for the IPO firm, and zero otherwise.

PreIPO An indicator variable that takes the value of one if the firm-year observation is before the IPO (year 0) for the IPO firm, and zero otherwise.

High short-termism public

An indicator variable that takes the value of one if the value of a public firm CEO’s vested stock and option grants as the percentage of the value of her total holdings of stock and option grants is above the sample median, and zero otherwise; alternatively, an indicator variable that takes the value of one if the coefficient on M/B in the probit regression of a public firm becoming a takeover target is below the sample median (i.e., more negative), and zero otherwise; alternatively, an indicator variable that takes the value of one if the ownership by transient institutional investors in a public firm is greater than the sample median, and zero otherwise. The ownership by transient institutional investors is the number of shares owned by transient institutional investors normalized by the number of shares owned by all institutional investors.

Low short-termism public

An indicator variable that takes the value of one if the value of a public firm CEO’s vested stock and option as the percentage of the CEO’s total stock and option holdings is below the sample median, and zero otherwise; alternatively, an indicator variable that takes the value of one if the coefficient on M/B in the probit regression of a public firm becoming a takeover target is above the sample median, and zero otherwise; alternatively, an indicator variable that takes the value of one if the

33

ownership by transient institutional investors in a public firm is below the sample median, and zero otherwise.

High analyst-coverage public

An indicator variable that takes the value of one if the number of analyst earnings forecasts made on a public firm is above the sample median, and zero otherwise.

Low analyst-coverage public

An indicator variable that takes the value of one if the number of analyst earnings forecasts made on a public firm is below the sample median, and zero otherwise.

Measures of Innovation Output

Patent count For firm i in year t, the number of patent applications that are made by firm i in year t-2 to year t (our sample period t starts in 1997 and ends in 2008) and are later awarded by the USPTO by the end of 2010. We set this variable to zero for firm-year observations without any patent record in our patent database.

Patent countCIQ This measure is constructed in three steps following Bena and Li (2012). First, for each technology class k and patent application year t, we compute the mean value of the number of awarded patents in technology class k with application year t across all Capital IQ firms that were awarded at least one patent in technology class k with application year t. Second, we scale the number of awarded patents to firm i in technology class k with application year t by the corresponding (technology class- and application year-specific) mean value from the first step. Third, for firm i in year t, we sum up the scaled number of awarded patents from the second step across all technology classes with application year t-2 to year t.

Citation count This measure is constructed in three steps. First, for each patent with application year t (our sample period t starts in 1997 and ends in 2008), we count the number of citations it receives since the award year to the end of 2010. The truncation bias in the citations received by a patent arises because citations are received many years after the innovation was created. So in the second step, we account for this bias by scaling the number of citations from the first step by a corresponding adjustment factor shown in Table 5 of Hall, Jaffe, and Trajtenberg (2005a). Third, for firm i in year t, we sum up the scaled number of citations from the second step across all patents with application year t-2 to year t.

Citation countCIQ This measure is constructed in three steps following Lerner, Sørensen, and Strömberg (2011), Bena and Li (2012), and Seru (2014). First, for each patent with application year t, we count the number of citations it receives in the award year and subsequent three years (year t to year t+3). Second, we scale the number of citations from the first step by the mean value of the number of citations received by awarded patents in technology class k with application year t across all Capital IQ firms over the same three-year window. Third, for firm i in year t, we sum up the scaled number of citations from the second step across all patents with application year t-2 to year t.

Patent stock The total number of granted patents applied by firm i up to year t-3. All patents recorded in the NBER/HBS merged patent dataset are included. We set this variable to zero for firm-year observations without any patent record in our patent database.

Measures of Innovation Strategy

Explore60 (80) This measure is constructed in three steps following Benner and Tushman (2002). First, for each patent applied by firm i in year t-2 to year t, we calculate the

34

percentage of its citations that are based on the existing expertise—the combination of firm i’s portfolio of patents and citations made by its portfolio of patents over the past five years (i.e., year t-7 to year t-3). Second, a patent is categorized as “exploratory” if more than or equal to 60% (80%) of its citations are outside of firm i’s existing expertise as defined in the first step. Finally, we calculate firm i’s exploratory ratio in year t as the number of exploratory patents applied in year t-2 to year t divided by the total number of patents applied over the same period.

Exploit60 (80) This measure is constructed in three steps following Benner and Tushman (2002). First, for each patent applied by firm i in year t-2 to year t, we calculate the percentage of its citations that are based on the existing expertise—the combination of firm i’s portfolio of patents and citations made by its portfolio of patents over the past five years (i.e., year t-7 to year t-3). Second, a patent is categorized as “exploitative” if more than or equal to 60% (80%) of its citations are based on firm i’s existing expertise as defined in the first step. Finally, we compute firm i’s exploitative ratio in year t as the number of exploitative patents applied in year t-2 to year t divided by the total number of patents applied over the same period.

Scope This measure is constructed following Katila and Ahuja (2002). We compute firm i’s scope ratio in year t as the number of new citations made by awarded patents applied by firm i in year t-2 to year t divided by the total number of citations made by awarded patents applied by firm i over the same period. New citations are citations that are made by one of firm i’s awarded patents applied in year t-2 to year t but were never cited by firm i’s awarded patents applied in the past five years (i.e., year t-7 to year t-3). The scope ratio is between zero and one.

Depth This measure is constructed following Katila and Ahuja (2002). We compute firm i’s depth ratio in year t as the number of repeated citations made by awarded patents applied by firm i in year t-2 to year t divided by the total number of citations made by awarded patents applied by firm i over the same period. Repeated citations are citations that are made by both one of firm i’s awarded patents applied in year t-2 to year t and one of firm i’s awarded patents applied in the past five years (i.e., year t-7 to year t-3). If one particular citation was cited twice over the five-year window, it is counted as two repeated citations. The depth ratio is equal to or greater than zero.

35

Appendix 2 Differences in innovation scope and depth between public and private firms This table replicates Table 3 by using scope and depth as the dependent variables. Columns (1) and (2) present the results based on the innovative firm sample. Columns (3) and (4) present the results based on the matched sample. Definitions of all variables are provided in Appendix 1. Robust standard errors clustered at the firm level are reported in brackets. Superscripts ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. The innovative firm sample The matching firm sample (1) (2) (3) (4) Scope Depth Scope Depth Public -0.045*** 0.031** -0.033** 0.027 [0.011] [0.015] [0.014] [0.019] Ln(Patent stock) -0.052*** 0.094*** -0.060*** 0.065*** [0.002] [0.005] [0.005] [0.006] Ln(Total assets) 0.022*** -0.022*** 0.012*** 0.001 [0.003] [0.004] [0.004] [0.005] Leverage -0.010 -0.024 -0.031 0.010 [0.016] [0.029] [0.026] [0.031] ROA 0.068*** -0.059* 0.019 -0.015 [0.021] [0.033] [0.027] [0.035] Sales growth -0.008** 0.015** -0.002 0.008 [0.003] [0.007] [0.007] [0.010] Capex -0.044 0.127 -0.014 0.198 [0.075] [0.110] [0.120] [0.169] R&D -0.049 0.037 -0.063 0.109 [0.039] [0.074] [0.055] [0.089] PPE 0.007 -0.076* 0.023 -0.047 [0.026] [0.045] [0.042] [0.060] Ln(Firm age) 0.018*** -0.038*** 0.013** -0.036*** [0.004] [0.005] [0.006] [0.008]

Industry × Year FEs Yes Yes Yes Yes

Constant 0.365*** 0.428*** 0.655*** 0.106 [0.096] [0.102] [0.092] [0.113] Observations 15,192 15,192 3,312 3,312 Adj R2 23% 16% 29% 22%

36

Appendix 3 Are firms with more exploratory/exploitative patents more likely to go public? The sample consists of 2,809 private firm-year observations (including both innovative private firms and innovative IPO firms) from 1997-2008. This table reports the probit regression results where the dependent variable is the IPO indicator variable. Definitions of all variables are provided in Appendix 1. Robust standard errors clustered at the firm level are reported in brackets. Superscripts ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

(1) (2) (3) (4)

Explore60 -0.222 [0.203] Explore80 -0.282 [0.203] Exploit60 -0.027 [0.272] Exploit80 0.193 [0.280] Ln(Patent stock) 0.054 0.049 0.066 0.061 [0.044] [0.045] [0.044] [0.044] Ln(Total assets) 0.325*** 0.325*** 0.323*** 0.328*** [0.043] [0.043] [0.043] [0.043] Leverage -2.296*** -2.299*** -2.284*** -2.294*** [0.330] [0.330] [0.331] [0.333] ROA 0.382* 0.388* 0.360 0.370* [0.218] [0.218] [0.219] [0.216] Sales growth 0.275*** 0.276*** 0.272*** 0.274*** [0.066] [0.066] [0.066] [0.066] Capex 0.508 0.501 0.534 0.529 [1.137] [1.138] [1.128] [1.132] R&D 0.331 0.325 0.304 0.298 [0.530] [0.529] [0.531] [0.531] Ln(Firm age) -0.123* -0.122* -0.125* -0.123* [0.065] [0.065] [0.065] [0.065] Industry × Year FEs Yes Yes Yes Yes Constant -11.556*** -11.542*** -11.685*** -11.816*** [0.939] [0.940] [0.940] [0.946] Observations 2,809 2,809 2,809 2,809 Pseudo R2 45% 45% 45% 45%

37

Appendix 4 The relation between our instrumental variable and innovation strategies This table replicates Table 3 Panel A by adding one more explanatory variable—underwriter concentration. Definitions of all variables are provided in Appendix 1. Robust standard errors clustered at the firm level are reported in brackets. Superscripts ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. (1) (2) (3) (4)

Explore60 Explore80 Exploit60 Exploit80

Public -0.054*** -0.062*** 0.026*** 0.028***

[0.012] [0.012] [0.009] [0.008]

Ln(Patent stock) -0.056*** -0.063*** 0.027*** 0.022***

[0.002] [0.003] [0.002] [0.002]

Ln(Total assets) 0.016*** 0.018*** -0.014*** -0.013***

[0.003] [0.003] [0.002] [0.002]

Leverage -0.022 -0.013 0.002 0.007

[0.017] [0.018] [0.012] [0.011]

ROA 0.054** 0.066*** -0.042*** -0.042***

[0.022] [0.022] [0.016] [0.014]

Sales growth -0.005 -0.007* 0.002 0.000

[0.004] [0.004] [0.003] [0.002]

Capex 0.054 0.059 0.052 0.035

[0.083] [0.084] [0.055] [0.051]

R&D -0.026 -0.040 0.026 0.016

[0.041] [0.042] [0.031] [0.027]

PPE -0.045 -0.043 -0.020 -0.006

[0.029] [0.029] [0.019] [0.018]

Ln(Firm age) 0.005 0.009** -0.010*** -0.008***

[0.004] [0.004] [0.003] [0.002]

Underwriter concentration -0.271 -0.326 0.213 0.247

[0.285] [0.274] [0.165] [0.176]

Industry × Year FEs Yes Yes Yes Yes

Constant 0.071 -0.019 0.487** 0.396**

[0.282] [0.288] [0.214] [0.197]

Observations 15,192 15,192 15,192 15,192

Adj R2 32% 32% 11% 11%

38

Appendix 5 Can the entrepreneur’s risk preference explain the difference in innovation strategy? Panel A replicates Table 3 Panel A by limiting to a subsample of firms that were incorporated prior to 1980. The sample consists of 6,663 public firm-year observations and 909 private firm-year observations. Panels B and C replicate Table 5 Panels A and B, respectively, by limiting to a subsample of 108 IPO firms whose founders are the CEOs till year +5 (the IPO year being year 0). Definitions of all variables are provided in Appendix 1. Robust standard errors clustered at the firm level are reported in brackets. Superscripts ***, **, * denote statistical significance at the 1, 5, and 10 percent levels, respectively. Panel A: Cross-sectional comparison of public and private firms incorporated prior to 1980 (1) (2) (3) (4) Explore60 Explore80 Exploit60 Exploit80 Public -0.060*** -0.068*** 0.016 0.022** [0.017] [0.017] [0.013] [0.011] Ln(Patent stock) -0.067*** -0.073*** 0.030*** 0.024*** [0.004] [0.004] [0.003] [0.002] Ln(Total assets) 0.015*** 0.015*** -0.014*** -0.013*** [0.004] [0.004] [0.003] [0.003] Leverage 0.001 0.007 -0.022 -0.011 [0.022] [0.024] [0.017] [0.015] ROA 0.082*** 0.102*** -0.039** -0.043** [0.025] [0.026] [0.020] [0.017] Sales growth -0.002 -0.004 0.002 0.000 [0.004] [0.004] [0.003] [0.003] Capex -0.035 -0.013 0.020 0.007 [0.105] [0.105] [0.071] [0.064] R&D 0.034 0.017 0.007 -0.008 [0.046] [0.048] [0.036] [0.031] PPE 0.015 0.005 -0.042 -0.024 [0.042] [0.043] [0.028] [0.025] Ln(Firm age) 0.006 0.003 -0.010* -0.008* [0.007] [0.007] [0.005] [0.005]

Industry × Year FEs Yes Yes Yes Yes

Constant 0.143** 0.159** 0.276*** 0.270*** [0.071] [0.072] [0.051] [0.045] Observations 7,572 7,572 7,572 7,572 Adj R2 38% 37% 15% 15%

39

Panel B: IPO firms whose founder keeps the CEO position till year +5 and matched private control firms (1) (2) (3) (4) Explore60 Explore80 Exploit60 Exploit80 PostIPO -0.034* -0.025 0.023 0.029**

[0.020] [0.020] [0.018] [0.014]

Ln(Patent stock) -0.071*** -0.078*** 0.060*** 0.046***

[0.013] [0.013] [0.012] [0.009]

Ln(Total assets) 0.001 -0.004 -0.018** -0.010

[0.009] [0.009] [0.008] [0.006]

Leverage 0.015 0.004 -0.040 0.022

[0.039] [0.039] [0.035] [0.027]

ROA -0.001 0.003 -0.022 -0.019

[0.024] [0.023] [0.021] [0.016]

Sales growth -0.002 -0.003 0.002 0.002

[0.003] [0.003] [0.002] [0.002]

Capex 0.317** 0.183 0.009 -0.007

[0.158] [0.155] [0.142] [0.109]

R&D -0.110** -0.129*** 0.018 0.030

[0.050] [0.049] [0.045] [0.034]

PPE 0.122 0.166* -0.108 -0.072

[0.097] [0.095] [0.087] [0.067]

Ln(Firm age) -0.020 -0.018 0.001 -0.011

[0.027] [0.027] [0.024] [0.019]

Firm FEs and Year FEs Yes Yes Yes Yes

Constant 0.722*** 0.786*** 0.298* 0.189

[0.175] [0.172] [0.157] [0.121]

Observations 977 977 977 977

Adj R2 30% 30% 14% 15%

40

Panel C: IPO firms whose founder keeps the CEO position till year +5 and matched public control firms (1) (2) (3) (4) Explore60 Explore80 Exploit60 Exploit80 PreIPO 0.029** 0.029** -0.018* -0.016*

[0.012] [0.011] [0.010] [0.009]

Ln(Patent stock) -0.090*** -0.092*** 0.057*** 0.047***

[0.014] [0.013] [0.011] [0.011]

Ln(Total assets) -0.006 -0.009 -0.014 -0.014*

[0.010] [0.010] [0.008] [0.008]

Leverage -0.025 -0.018 0.046 0.051

[0.042] [0.041] [0.034] [0.032]

ROA 0.019 0.022 -0.032 -0.029

[0.026] [0.025] [0.022] [0.020]

Sales growth -0.005 -0.006* 0.006** 0.005*

[0.003] [0.003] [0.003] [0.002]

Capex 0.335* 0.047 0.199 0.274**

[0.181] [0.177] [0.149] [0.139]

R&D -0.081 -0.083* 0.000 -0.019

[0.051] [0.050] [0.042] [0.039]

PPE 0.164 0.229** -0.217** -0.206**

[0.106] [0.104] [0.088] [0.082]

Ln(Firm age) -0.028 -0.035 0.011 0.005

[0.031] [0.030] [0.025] [0.024]

Firm FEs and Year FEs Yes Yes Yes Yes

Constant 0.723*** 0.729*** 0.322** 0.341**

[0.182] [0.178] [0.151] [0.141]

Observations 1,060 1,060 1,060 1,060

Adj R2 22% 23% 10% 10%

41

References:

Acharya, Viral V., and Zhaoxia Xu, 2013. Financial dependence and innovation: The case of public versus private firms, New York University working paper.

Admati, Anat, and Paul Pfleiderer, 2009. The Wall Street walk and shareholder activism: Exit as a form

of voice, Review of Financial Studies 22, 2645-2685. Aggarwal, Vikas A., and David H. Hsu, 2013. Entrepreneurial exits and innovation, Management Science

forthcoming. Aghion, Philippe, John Van Reenen, and Luigi Zingales, 2013. Innovation and institutional ownership,

American Economic Review 103, 277-304. Akcigit, Ufuk, and William R. Kerr, 2012. Growth through heterogeneous innovations, University of

Pennsylvania working paper. Amore, Mario Daniele, Cédric Schneider, and Alminas Žaldokas, 2013. Credit supply and corporate

innovation, Journal of Financial Economics 109, 835-855. Asker, John, Joan Farre-Mensa, Alexander Ljungqvist, 2013. Corporate investment and stock market

listing: A puzzle? New York University working paper. Atanassov, Julian, 2013. Do hostile takeovers stifle innovation? Evidence from antitakeover legislation

and corporate patenting, Journal of Finance 68, 1097-1131. Atanassov, Julian, Vikram Nanda, and Amit Seru, 2007, Finance and innovation: The case of publicly

traded firms, University of Oregon working paper. Ayyagari, Meghana, Asli Demirgüç-Kunt, and Vojislav Maksimovic, 2011. Firm innovation in emerging

markets: The role of finance, governance and competition, Journal of Financial and Quantitative Analysis 46, 1545-1580.

Barrot, Jean-Noël, 2012. Investor horizon and innovation: Evidence from private equity funds, HEC Paris

working paper. Bebchuk, Lucian Arye and Lars A. Stole, 1993. Do short-term objectives lead to under- or

overinvestment in long-term projects? Journal of Finance 48, 719-729. Bena, Jan, and Kai Li, 2012. Corporate innovations and mergers and acquisitions, Journal of Finance

forthcoming. Benfratello, Luigi, Fabio Schiantarelli, and Alessandro Sembenelli, 2008. Banks and innovation: Micro

econometric evidence on Italian firms, Journal of Financial Economics 90,197-217. Benner, Mary J., and Michael Tushman, 2002. Process management and technological innovation: A

longitudinal study of the photography and paint industries, Administrative Science Quarterly 47, 676-706.

Bernstein, Shai, 2012. Does going public affect innovation? Stanford University working paper.

42

Bharath, Sreedhar T., and Amy K. Dittmar, 2010. Why do firms use private equity to opt out of public markets? Review of Financial Studies 23, 1771-1818.

Bhattacharya, Sugato, and Jay R. Ritter, 1983. Innovation and communication: Signaling with partial

disclosure, Review of Economic Studies 50, 331-346. Bloom, Nicholas, Mark Schankerman, and John Van Reenen, 2013. Identifying technology spillovers and

product market rivalry, Econometrica 81, 1347-1393. Bolton, Patrick, Jose Scheinkman, and Wei Xiong, 2006. Executive compensation in speculative markets,

Review of Economics Studies 73, 577-610. Brav, Omer, 2009. Access to capital, capital structure, and the funding of the firm, Journal of Finance 64,

263-308. Brown, James R., Steven M. Fazzari, and Bruce C. Petersen, 2009. Financing innovation and growth:

Cash flow, external equity, and the 1990s R&D boom, Journal of Finance 64, 151-185. Bushee, Brian J., 1998. The influence of institutional investors on myopic R&D investment behavior,

Accounting Review 73, 305-333. Bushee, Brian, J., 2001. Do institutional investors prefer near-term earnings over long-run value?

Contemporary Accounting Research 18, 207-246. Chava, Sudheer, Alexander Oettl, Ajay Subramanian, and Krishnamurthy Subramanian, 2013. Bank

deregulation and innovation, Journal of Financial Economics 109, 759-775. Chemmanur, Thomas J., Shan He, and Debarshi K. Nandy, 2010. The going-public decision and the

product market, Review of Financial Studies 23, 1855-1908. Cohen, Wesley M., and Daniel A. Levinthal, 1994. Fortune favors the prepared firm, Management

Science 40, 227-251. Cremers, Martijn, Vinay Nair, and Kose John, 2009. Takeovers and the cross-section returns, Review of

Financial Studies 22, 1409-1445. Crossan, Mary M., Henry W. Lane, and Roderick E. White, 1999. An organizational learning framework:

From intuition to institution, Academy of Management Review 24, 522-537. Economist, 1990, Out of the ivory tower, February 3. Edmans, Alex, 2009. Blockholder trading, market efficiency, and managerial myopia, Journal of Finance

64, 2481-2513. Edmans, Alex, and Gustavo Manso, 2011. Governance through trading and intervention: A theory of

multiple blockholders, Review of Financial Studies 24, 2395-2428. Edmans, Alex, Vivian W. Fang, and Katharina A. Lewellen, 2014. Equity vesting and managerial

myopia, London Business School working paper.

43

Faccio, Mara, Maria-Teresa Marchica, and Roberto Mura, 2011. Large shareholder diversification and corporate risk-taking, Review of Financial Studies 24, 3601-3641.

Fama, Eugene, and Kenneth French, 1997. Industry costs of capital, Journal of Financial Economics 43,

153-193. Fang, Vivian W., Xuan Tian, and Sheri Tice, 2013. Does stock liquidity enhance or impede firm

innovation? Journal of Finance forthcoming. Ferreira, Daniel, Gustavo Manso, and Andre C. Silva, 2012. Incentives to innovate and the decision to go

public or private, Review of Financial Studies forthcoming. Gao, Huasheng, Jarrad Harford, and Kai Li, 2013. Determinants of corporate cash policy: Insights from

private firms, Journal of Financial Economics 109, 623-639. Gopalan, Radhakrishnan, Todd Milbourn, Fenghua Song, and Anjan Thakor, 2014. Duration of executive

compensation, Journal of Finance forthcoming. Graham, John R., Campbell R. Harvey, and Shivaram Rajgopal, 2005. The economic implications of

corporate financial reporting, Journal of Accounting and Economics 40, 3-73. Griliches, Zvi, 1981. Market value, R&D, and patents. Economics Letters 7, 183-167. Hall, Bronwyn H., Adam B. Jaffe, and Manuel Trajtenberg, M., 2005a. The NBER patent citation data

file: Lessons, insights and methodological tools. In: Jaffe, A.B., Trajtenberg, M. (Eds.), Patents, Citations and Innovations: A Window on the Knowledge Economy. MIT Press, Cambridge, MA, pp. 403-470.

Hall, Bronwyn H., Adam Jaffe, and Manuel Trajtenberg, 2005b. Market value and patent citations, RAND

Journal of Economics 36, 16-38. Hall, Bronwyn H., and Josh Lerner, 2010. The financing of R&D and innovation, in: Handbook of the

Economics of Innovation, eds. Bronwyn H. Hall and Nathan Rosenberg, 609-639. Elsevier-North Holland.

He, Zi-Lin, and Poh-Kam Wong, 2004. Exploration vs. exploitation: An empirical test of the

ambidexterity hypothesis, Organization Science 15, 481-494. Heckman, James J., Hidehiko Ichimura, and Petra E. Todd, 1997. Matching as an econometric evaluation

estimator: Evidence from evaluating a job training programme, Review of Economic Studies 64, 605-654.

Hsu, Po-Hsuan, Xuan Tian, and Yan Xu, 2014. Financial development and innovation: Cross-country

evidence, Journal of Financial Economics 112, 116 -135. Jaffe, Adam B., 2000. The U.S. patent system in transition: Policy innovation and the innovation process.

Research Policy 29, 531-557. Katila, Riitta, and Gautam Ahuja, 2002. Something old, something new: A longitudinal study of search

behavior and new product introduction, Academy of Management Journal 45, 1183-1194.

44

Kerr, William R., and Ramana Nanda, 2009. Democratizing entry: Banking deregulations, financing constraints, and entrepreneurship, Journal of Financial Economics 94, 124-149.

King, Robert, and Ross Levine, 1993. Finance, entrepreneurship, and growth: Theory and evidence,

Journal of Monetary Economics 32, 513-542. Ladika, Tomislav, and Zacharias Sautner, 2014. The effect of managerial short-termism on corporate

investment, University of Amsterdam working paper. Lai, Ronald, Alexander D’ Amour, Amy Yu, and Lee Fleming, 2011. Disambiguation and co-authorship

networks of the U.S. patent inventor database (1975-2010), Harvard University working paper. Lanjouw, Jean O., and Mark Schankerman, 2004. Patent quality and research productivity: Measuring

innovation with multiple indicators, Economic Journal 114, 441-465. Lavie, Dovev, Uriel Stettner, and Michael Tushman, 2010. Exploration and exploitation within and across

organizations, Academy of Management Annals 4, 109-155. Lerner, Josh, Morten Sørensen, and Per Strömberg, 2011. Private equity and long-run investment: The

case of innovation, Journal of Finance 66, 445-477. Levine, Ross, 2005. Finance and growth: Theory and evidence, Handbook of Economic Growth Vol. I,

ed. P. Aghion and S.D. Durlauf, 865-934. Elsevier/Amsterdam. Levinthal, Daniel A., and James G. March, 1993. The myopia of learning, Strategic Management Journal

14, 95-112. Li, K., and N. R. Prabhala, 2007. Self-selection models in corporate finance. In Handbook of Corporate

Finance: Empirical Corporate Finance Vol. I, ed. B. E. Eckbo, 37-86. Elsevier/North-Holland. Liu, Xiaoding, and Jay R. Ritter, 2011. Local underwriter oligopolies and IPO underpricing, Journal of

Financial Economics 102, 579-601. Lowry, Michelle, 2003. Why does IPO volume fluctuate so much? Journal of Financial Economics 65, 3-

40. Maddala, G. S., 1983. Limited-Dependent and Qualitative Variables in Economics, New York:

Cambridge University Press. Maksimovic, Vojislav, and Pegaret Pichler, 2001. Technological innovation and initial public offerings.

Review of Financial Studies 14, 459-494. Malmendier, Ulrike, and Geoffrey Tate, 2005. CEO overconfidence and corporate investment. Journal of

Finance 60, 2661-2700. March, James G., 1991. Exploration and exploitation in organizational learning, Organization Science 2,

71-87. Maug, Ernst, 1998. Large shareholders as monitors: is there a trade-off between liquidity and control?

Journal of Finance 53, 65-98.

45

McGrath, Rita Gunther, 2001. Exploratory learning, innovative capacity and managerial oversight, Academy of Management Journal 44, 118-131.

Meulbroek, Lisa K., Mark L. Mitchell, J. Harold Mulherin, Jeffry M. Netter, and Annette B. Poulsen,

1990. Shark repellents and managerial myopia: An empirical test, Journal of Political Economy 98, 1108-1117.

Nanda, Ramana, and Matthew Rhodes-Kropf, 2011. Financing risk and innovation, Harvard University

working paper. Nanda, Ramana, and Matthew Rhodes-Kropf, 2013. Investment cycles and startup innovation, Journal of

Financial Economics 110, 403-418. Nohria, Nitin, and Ranjay Gulati, 1996. Is slack good or bad for innovation? Academy of Management

Journal 39, 1245-1264. Pagano, Marco, Fabio Panetta, and Luigi Zingales, 1998. Why do companies go public? An empirical

analysis, Journal of Finance 53, 27-64. Porter, Michael E., 1992. Capital disadvantage: America’s failing capital investment system, Harvard

Business Review, September-October. Porter, Michael E., 1998. Competitive Advantage of Nations, Free Press. Romer, Paul, 1990. Endogenous technological change, Journal of Political Economy 98, S71-S102. Sapra, Haresh, Ajay Subramanian, and Krishnamurthy Subramanian, 2013. Corporate governance and

innovation: Theory and evidence, Journal of Financial and Quantitative Analysis forthcoming. Scherer, F. M., 1965. Firm size, market structure, opportunity, and the output of patented inventions.

American Economic Review 55, 1097-1125. Schumpeter, Joseph, 1934. The Theory of Economic Development. Translated by Redvers Opie. Harvard

University Press, Cambridge, MA. Seru, Amit, 2014. Firm boundaries matter: Evidence from conglomerates and R&D activity, Journal of

Financial Economics 111, 381-405. Smith, Wendy K., and Michael L. Tushman, 2005. Managing strategic contradictions: A top management

model for managing innovation streams, Organization Science 16, 522-536. Solow, Robert M., 1957. Technical change and the aggregate production function, Review of Economics

and Statistics 39, 312-320. Stein, Jeremy, 1988. Takeover threats and managerial myopia, Journal of Political Economy 96, 61-80. Stein, Jeremy, 1989. Efficient capital market, inefficient firms: A model of myopic corporate behavior,

Quarterly Journal of Economics 104, 655-669. Trajtenberg, Manuel, 1990. A penny for your quotes: Patent citations and the value of innovations, RAND

Journal of Economics 21, 172-187.

46

Uotila, Juha, Markku Maula, Thomas Keil, and Shaker A. Zahra, 2009. Exploration, exploitation, and

financial performance: Analysis of S&P 500 corporations, Strategic Management Journal 30, 221-231.

47

Table 1 Sample distribution over time The full sample consists of 50,294 public firm-year observations (9,163 unique public firms) and 16,217 private firm-year observations (6,549 unique private firms) from 1997-2008, obtained from matching Capital IQ with the NBER/HBS patent databases. The innovative firm sample consists of 13,463 public firm-year observations (2,426 unique public firms) and 1,729 private firm-year observations (829 unique private firms) from 1997-2008. Innovative firms in year t are required to have at least one patent over the three-year period from year t-2 to year t.

Public firms Private firms

Year # of firms # of

innovative firms

% of innovative

firms

# of firms # of

innovative firms

% of innovative

firms

1997 4,231 1,199 28% 805 101 13% 1998 4,902 1,307 27% 1,058 126 12% 1999 5,023 1,308 26% 1,235 126 10% 2000 4,828 1,249 26% 1,521 153 10% 2001 4,582 1,229 27% 1,276 153 12% 2002 4,521 1,253 28% 1,220 133 11% 2003 3,991 1,191 30% 1,245 128 10% 2004 3,737 1,108 30% 1,336 131 10% 2005 3,678 1,008 27% 1,330 112 8% 2006 3,694 951 26% 1,606 174 11% 2007 3,599 870 24% 1,856 235 13% 2008 3,508 790 23% 1,729 157 9%

Total 50,294 13,463 27% 16,217 1,729 11%

48

Table 2 Summary statistics The full sample consists of 50,294 public firm-year observations (9,163 unique public firms) and 16,217 private firm-year observations (6,549 unique private firms) from 1997-2008, obtained from matching Capital IQ with the NBER/HBS patent databases. The innovative firm sample consists of 13,463 public firm-year observations (2,426 unique public firms) and 1,729 private firm-year observations (829 unique private firms) from 1997-2008. Innovative firms in year t are required to have at least one patent over the three-year period from year t-2 to year t. Definitions of all variables are provided in Appendix 1. All dollar values are in 2008 dollars. All continuous variables are winsorized at the 1st and 99th percentiles. Panel A is based on the full sample. Panel B is based on the innovative firm sample. Test statistics of the t-test and the Wilcoxon-test of the differences in innovation measures and firm characteristics between public firms and private firms are given in superscript ***, **, and * denoting statistical significance at the 1%, 5%, and 10% levels, respectively. Panel A: The full sample

Public firms

Private firms

Mean

StdDev

25th percentile

Median

75th percentile

Mean

StdDev

25th percentile

Median

75th percentile

Patent count 13.60*** 55.37 0.00 0.00*** 2.00 2.28 18.91 0.00 0.00 0.00

Patent countCIQ 3.48*** 13.59 0.00 0.00*** 0.56 0.69 5.09 0.00 0.00 0.00

Citation count 48.67*** 203.99 0.00 0.00*** 2.98 6.63 64.18 0.00 0.00 0.00

Citation countCIQ 15.05*** 61.65 0.00 0.00*** 0.87 2.34 20.46 0.00 0.00 0.00

Patent stock 69.19*** 302.26 0 0*** 11.00 11.67 95.97 0.00 0.00 0.00

Total assets ($M) 1680*** 4928 41 177*** 844 1557 4361 7 130 893

Leverage 23.59%*** 24.27% 1.74% 18.28%*** 36.57% 39.26% 33.88% 2.52% 35.21% 64.85%

ROA 4.87%*** 15.76% -0.15% 7.33%*** 13.37% 6.32% 15.98% 0.00% 6.30% 11.92%

Sales growth 29.08%*** 80.91% 0.00% 9.86%*** 29.50% 23.03% 78.38% 0.00% 1.04% 15.60%

Capex 6.37%** 7.56% 1.90% 4.00%*** 7.70% 6.17% 9.39% 0.45% 3.18% 7.08%

R&D 6.39%*** 14.08% 0.00% 0.00%*** 6.97% 4.97% 17.28% 0.00% 0.00% 0.00%

PPE 26.47%*** 23.53% 7.92% 18.71%*** 38.74% 34.50% 28.80% 8.79% 26.86% 58.54%

Firm age 32*** 34 9 19*** 42 26 35 3 9 35

49

Panel B: The innovative firm sample

Public firms Private firms

Mean

StdDev

25th percentile

Median

75th percentile

Mean

StdDev

25th percentile

Median

75th percentile

Patent count 78.12*** 234.52 3.00 10.00*** 38.00 19.33 66.42 1.00 3.00 10.00

Patent countCIQ 302.09*** 974.15 6.58 27.62*** 130.63 64.81 315.19 1.69 7.55 29.28

Citation count 18.59*** 51.56 0.90 2.82*** 10.28 5.84 17.95 0.44 1.08 3.30

Citation countCIQ 86.86*** 263.31 2.03 8.89*** 41.74 19.21 70.18 0.42 2.61 9.48

Patent stock 390.28*** 1263.70 8.00 34.00*** 141.00 76.06 263.63 1.00 8.00 35.00

Explore60 0.51*** 0.29 0.31 0.50*** 0.71 0.60 0.34 0.33 0.58 1.00

Explore80 0.48*** 0.30 0.26 0.44*** 0.67 0.58 0.35 0.30 0.50 1.00

Exploit60 0.14*** 0.19 0.00 0.08*** 0.20 0.10 0.21 0.00 0.00 0.12

Exploit80 0.12*** 0.18 0.00 0.06*** 0.17 0.09 0.19 0.00 0.00 0.09

Scope 0.75*** 0.27 0.61 0.81*** 1 0.81 0.27 0.72 1 1

Depth 0.21*** 0.46 0.00 0*** 0.21 0.12 0.37 0 0 0.03

Total assets ($M) 3425*** 9017 106 416*** 1977 2664 7642 22 176 1121

Leverage 17.20%*** 18.03% 0.69% 12.99%*** 28.18% 40.80% 30.76% 13.70% 38.13% 62.69%

ROA 4.57%*** 16.89% -0.86% 8.00%*** 14.40% 1.08% 25.04% 0.00% 5.66% 13.74%

Sales growth 26.53% 69.12% 0.00% 11.00%*** 29.20% 28.34% 86.71% 0.00% 0.99% 17.50%

Capex 5.32%*** 4.65% 2.26% 4.02%*** 6.80% 4.42% 5.58% 0.00% 2.82% 6.30%

R&D 8.24%*** 10.12% 1.07% 5.20%*** 11.80% 7.40% 20.03% 0.00% 0.00% 3.91%

PPE 21.43%*** 16.48% 8.78% 17.17%*** 29.82% 27.98% 20.16% 11.85% 23.28% 40.74%

Firm age 40* 39 13 23*** 61 38 40 6 20 64

50

Table 3 Differences in innovation strategies between public and private firms The innovative firm sample consists of 13,463 public firm-year observations and 1,729 private firm-year observations from 1997-2008. Innovative firms in year t are required to have at least one patent over the three-year period from year t-2 to year t. Definitions of all variables are provided in Appendix 1. All dollar values are in 2008 dollars. All continuous variables are winsorized at the 1st and 99th percentiles. Panel A presents the results based on the innovative firm sample. Panel B presents the results based on the matched sample. We follow Gao, Harford, and Li (2013) to match each private firm to a public firm in the same Fama and French 48 industry and closest in total assets in the same year. To ensure that the matched public firm is truly comparable to the private firm, we also require that the initial size of the matched public firm be within [50%, 150%] of the size of the private firm. Robust standard errors clustered at the firm level are reported in brackets. Superscripts ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. Panel A: The innovative firm sample (1) (2) (3) (4) Explore60 Explore80 Exploit60 Exploit80 Public -0.054*** -0.062*** 0.026*** 0.028*** [0.012] [0.012] [0.009] [0.008] Ln(Patent stock) -0.056*** -0.063*** 0.027*** 0.022*** [0.002] [0.003] [0.002] [0.002] Ln(Total assets) 0.016*** 0.018*** -0.014*** -0.013*** [0.003] [0.003] [0.002] [0.002] Leverage -0.022 -0.013 0.002 0.007 [0.017] [0.018] [0.012] [0.011] ROA 0.054** 0.066*** -0.042*** -0.042*** [0.022] [0.022] [0.016] [0.014] Sales growth -0.005 -0.007* 0.002 0.000 [0.004] [0.004] [0.003] [0.002] Capex 0.054 0.059 0.052 0.035 [0.083] [0.084] [0.055] [0.051] R&D -0.026 -0.040 0.026 0.016 [0.041] [0.042] [0.031] [0.027] PPE -0.045 -0.043 -0.020 -0.006 [0.029] [0.029] [0.019] [0.018] Ln(Firm age) 0.005 0.009** -0.010*** -0.008*** [0.004] [0.004] [0.003] [0.002]

Industry × Year FEs Yes Yes Yes Yes

Constant 0.361*** 0.332*** 0.294*** 0.267*** [0.123] [0.123] [0.060] [0.057] Observations 15,192 15,192 15,192 15,192 Adj R2 32% 32% 12% 12%

51

Panel B: Private firms and their industry- and size-matched public firms (1) (2) (3) (4) Explore60 Explore80 Exploit60 Exploit80 Public -0.033** -0.047*** 0.023** 0.025** [0.016] [0.016] [0.012] [0.011] Ln(Patent stock) -0.066*** -0.071*** 0.028*** 0.023*** [0.006] [0.006] [0.004] [0.003] Ln(Total assets) 0.005 0.005 -0.007* -0.007** [0.006] [0.006] [0.004] [0.003] Leverage -0.060** -0.051* 0.017 0.027 [0.028] [0.030] [0.021] [0.019] ROA 0.003 0.013 -0.024 -0.018 [0.030] [0.032] [0.024] [0.023] Sales growth -0.001 -0.003 -0.001 -0.005 [0.008] [0.008] [0.007] [0.004] Capex 0.184 0.208 0.060 0.021 [0.158] [0.166] [0.097] [0.088] R&D -0.007 -0.041 0.018 0.013 [0.062] [0.066] [0.049] [0.044] PPE 0.004 -0.006 -0.054 -0.040 [0.054] [0.056] [0.033] [0.030] Ln(Firm age) -0.006 -0.002 -0.009* -0.007 [0.007] [0.007] [0.005] [0.005]

Industry × Year FEs Yes Yes Yes Yes

Constant 0.856*** 0.738*** 0.223*** 0.262*** [0.115] [0.118] [0.074] [0.066] Observations 3,312 3,312 3,312 3,312 Adj R2 36% 35% 18% 19%

52

Table 4 Differences in innovation strategies between propensity score-matched public firms and private firms The innovative firm sample consists of 13,463 public firm-year observations and 1,729 private firm-year observations from 1997-2008. Innovative firms in year t are required to have at least one patent over the three-year period from year t-2 to year t. This table presents differences in the measures of innovation strategy between the propensity score-matched public firms and the private firms. We match each private firm to a public firm using the nearest neighbor algorithm. In Model 1, the variables we use to match are Ln(total assets), and industry times year fixed effects. In Model 2, the variables we use to match are Ln(total assets), Ln(patent stock), and industry times year fixed effects. In Model 3, the variables we use to match are all the explanatory variables in Table 3 Panel A except the public firm indicator variable. To test pairwise differences in means between the two samples (i.e., matched public firms – private firms), we use bootstrapped standard errors based on 50 replications without replacement which are reported in brackets. Superscripts ***, **, * denote statistical significance at the 1, 5, and 10 percent levels, respectively.

(1) (2) (3) (4) Explore60 Explore80 Exploit60 Exploit80 Model 1 -0.066*** -0.079*** 0.035*** 0.037***

[0.011] [0.012] [0.008] [0.007]

Model 2 -0.025** -0.034*** 0.018** 0.021***

[0.012] [0.012] [0.008] [0.007]

Model 3 -0.028** -0.036*** 0.016** 0.020***

[0.012] [0.012] [0.007] [0.007]

53

Table 5 Changes in innovation strategies around IPOs The sample consists of 392 IPOs by innovative firms over the sample period 1997-2008. Innovative firms in year t are required to have at least one patent over the three-year period from year t-2 to year t. We match each IPO firm to an innovative private (public) firm that (1) is in the same industry, (2) stays private (public) during the entire sample period, and (3) has the closest propensity score in the year prior to the IPO using the nearest neighbor algorithm. The variables we use to match are Ln(total assets), leverage, ROA, sales growth, capital expenditures, R&D, and firm age. We track IPO firms’ and matched private (public) control firms’ patenting activities from year -5 to year +5 centered at the IPO year (year 0). Panel A presents the results based on the sample including the IPO firms and their matched private control firms. Panel B presents the results based on the sample including the IPO firms and their matched public control firms. Definitions of all variables are provided in Appendix 1. All dollar values are in 2008 dollars. All continuous variables are winsorized at the 1st and 99th percentiles. Robust standard errors clustered at the firm level are reported in brackets. Superscripts ***, **, * denote statistical significance at the 1, 5, and 10 percent levels, respectively. Panel A: IPO firms and matched private control firms (1) (2) (3) (4) Explore60 Explore80 Exploit60 Exploit80 PostIPO -0.048*** -0.037*** 0.036*** 0.036***

[0.010] [0.010] [0.009] [0.007]

Ln(Patent stock) -0.062*** -0.066*** 0.053*** 0.042***

[0.006] [0.006] [0.005] [0.004]

Ln(Total assets) -0.002 -0.007 -0.010** -0.004

[0.005] [0.005] [0.004] [0.003]

Leverage 0.029 0.027 -0.061*** -0.015

[0.019] [0.019] [0.016] [0.013]

ROA 0.020 0.026** -0.021** -0.012

[0.013] [0.013] [0.011] [0.008]

Sales growth -0.001 -0.001 -0.001 -0.000

[0.001] [0.001] [0.001] [0.001]

Capex 0.092 0.060 -0.100 -0.072

[0.079] [0.079] [0.068] [0.052]

R&D -0.031 -0.062** 0.001 0.015

[0.027] [0.027] [0.023] [0.018]

PPE 0.021 0.061 0.019 0.005

[0.046] [0.046] [0.039] [0.030]

Ln(Firm age) -0.032** -0.032** 0.016 0.011

[0.014] [0.014] [0.012] [0.009]

Firm FEs and Year FEs Yes Yes Yes Yes

Constant 0.821*** 0.868*** 0.141* 0.059

[0.097] [0.097] [0.083] [0.064]

Observations 3,605 3,605 3,605 3,605

Adj R2 27% 26% 14% 15%

54

Panel B: IPO firms and matched public control firms (1) (2) (3) (4) Explore60 Explore80 Exploit60 Exploit80 PreIPO 0.028*** 0.025*** -0.019*** -0.016***

[0.007] [0.006] [0.005] [0.005]

Ln(Patent stock) -0.070*** -0.073*** 0.058*** 0.042***

[0.006] [0.006] [0.005] [0.005]

Ln(Total assets) 0.005 0.001 -0.007 -0.007*

[0.005] [0.005] [0.004] [0.004]

Leverage -0.012 -0.010 -0.013 -0.002

[0.019] [0.019] [0.016] [0.014]

ROA 0.018 0.024* -0.014 -0.012

[0.013] [0.013] [0.011] [0.010]

Sales growth -0.005*** -0.005*** 0.003* 0.002

[0.002] [0.002] [0.001] [0.001]

Capex -0.034 -0.085 0.066 0.155**

[0.082] [0.082] [0.069] [0.062]

R&D 0.008 -0.014 0.012 0.019

[0.026] [0.026] [0.022] [0.020]

PPE 0.063 0.108** -0.063 -0.095***

[0.047] [0.046] [0.039] [0.035]

Ln(Firm age) -0.004 -0.003 0.014 0.019*

[0.015] [0.015] [0.013] [0.012]

Firm FEs and Year FEs Yes Yes Yes Yes

Constant 0.633*** 0.665*** 0.125 0.115

[0.094] [0.093] [0.079] [0.071]

Observations 3,982 3,982 3,982 3,982

Adj R2 23% 21% 11% 11%

55

Table 6 The instrumental variable approach The innovative firm sample consists of 13,463 public firm-year observations and 1,729 private firm-year observations from 1997-2008. Innovative firms in year t are required to have at least one patent over the three-year period from year t-2 to year t. Column (1) reports the first-stage probit regression with the public firm indicator variable as the dependent variable and underwriter concentration as the instrumental variable. Columns (2)-(5) report the second-stage treatment regression of the innovation strategy model where the dependent variables are different measures of innovation strategies and we replace the public firm indicator variable by its predicted value from the first stage. Definitions of all variables are provided in Appendix 1. All dollar values are in 2008 dollars. All continuous variables are winsorized at the 1st and 99th percentiles. Robust standard errors clustered at the firm level are reported in brackets. Superscripts ***, **, * denote statistical significance at the 1, 5, and 10 percent levels, respectively.

(1) First stage predicting

Public

(2) Explore60

(3) Explore80

(4) Exploit60

(5) Exploit80

Public -0.259*** -0.263*** 0.184*** 0.167***

[0.029] [0.030] [0.028] [0.026]

Ln(Patent stock) 0.122*** -0.053*** -0.060*** 0.024*** 0.019***

[0.010] [0.002] [0.002] [0.001] [0.001]

Ln(Total assets) 0.117*** 0.021*** 0.023*** -0.017*** -0.015***

[0.010] [0.002] [0.002] [0.001] [0.001]

Leverage -2.291*** -0.132*** -0.122*** 0.002 0.008

[0.066] [0.019] [0.019] [0.008] [0.008]

ROA 0.434*** 0.055*** 0.066*** -0.052*** -0.050***

[0.100] [0.015] [0.015] [0.011] [0.010]

Sales growth -0.033 -0.007** -0.008*** 0.003 0.001

[0.020] [0.003] [0.003] [0.002] [0.002]

Capex 5.026*** 0.250*** 0.251*** -0.140** -0.133***

[0.380] [0.064] [0.065] [0.055] [0.051]

R&D 0.006 -0.032 -0.046 0.009 0.001

[0.178] [0.028] [0.029] [0.021] [0.020]

PPE -1.567*** -0.109*** -0.107*** 0.055*** 0.059***

[0.105] [0.021] [0.021] [0.019] [0.018]

Ln(Firm age) 0.128*** 0.011*** 0.015*** -0.014*** -0.012***

[0.015] [0.003] [0.003] [0.002] [0.002]

Underwriter concentration -0.431***

[0.053]

Industry × Year FEs No Yes Yes Yes Yes

Constant -0.895*** 0.443*** 0.412*** 0.229*** 0.210***

[0.180] [0.115] [0.117] [0.086] [0.080]

Observations 15,192 15,192 15,192 15,192 15,192

Endogeneity test

Rho 0.45*** 0.44*** -0.45*** -0.43***

56

Table 7 Managerial short-termism within public firms The innovative firm sample consists of 13,463 public firm-year observations and 1,729 private firm-year observations from 1997-2008. Innovative firms in year t are required to have at least one patent over the three-year period from year t-2 to year t. Definitions of all variables are provided in Appendix 1. All dollar values are in 2008 dollars. All continuous variables are winsorized at the 1st and 99th percentiles. Panel A presents the results based on managerial short-termism caused by CEOs’ vested equity portfolios. We first compute a CEO’s vested equity portfolio as the value of her vested stock and option grants normalized by the value of her total stock and option grants. We then divide the innovative public firm sample into high and low managerial short-termism subsamples based on median of the CEOs’ vested equity portfolios and run Table 3 Panel A regression by replacing the public firm indicator variable with the high short-termism public firm (above the median) and low short-termism public firm (below the median) indicator variables. Panel B presents the results based on managerial short-termism caused by the takeover market. We first run a probit specification to predict a public firm’s likelihood to become a takeover target using Ln(total assets), M/B, leverage, ROA, sales growth, cash holdings, and year fixed effects. The coefficient on M/B captures the sensitivity of a public firm’s likelihood of becoming a target to its M/B and is our proxy for managerial short-termism: the more negative is this coefficient, the more sensitive of a firm’s likelihood of becoming a target to its (mis-)valuation and hence more pressure on managers to achieve short-term results. We then divide the innovative public firm sample into high and low managerial short-termism subsamples based on their industry-level takeover sensitivity and run Table 3 Panel A regressions by replacing the public firm indicator variable with the high short-termism public firm (below the median) and low short-termism public firm (above the median) indicator variables. Panel C presents the results based on managerial short-termism caused by the presence of transient institutional investors. We first compute the ownership by transient institutional investors in a public firm following Bushee (1998, 2001). We then divide the innovative public firm sample into high and low managerial short-termism subsamples based on their ownership by transient institutional investors and run Table 3 Panel A regressions by replacing the public firm indicator variable with the high short-termism public firm (above the median) and low short-termism public firm (below the median) indicator variables. The F-statistics to test the equality of coefficients on High short-termism public = Low short-termism public are provided at the bottom of the table. Robust standard errors clustered at the firm level are reported in brackets. Superscripts ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

57

Panel A: Managerial short-termism caused by CEOs’ vested equity portfolios (1) (2) (3) (4) Explore60 Explore80 Exploit60 Exploit80 High short-termism public (a) -0.060*** -0.068*** 0.052*** 0.053***

[0.014] [0.014] [0.012] [0.011]

Low short-termism public (b) -0.037** -0.044*** 0.041*** 0.043***

[0.015] [0.015] [0.012] [0.011]

Ln(Patent stock) -0.046*** -0.052*** 0.023*** 0.018***

[0.003] [0.003] [0.002] [0.002]

Ln(Total assets) 0.009** 0.010*** -0.011*** -0.009***

[0.004] [0.004] [0.003] [0.002]

Leverage -0.044** -0.040* 0.038*** 0.044***

[0.020] [0.021] [0.015] [0.013]

ROA -0.007 -0.002 -0.025 -0.021

[0.026] [0.027] [0.020] [0.018]

Sales growth -0.005 -0.005 0.006 0.006

[0.009] [0.009] [0.008] [0.006]

Capex 0.099 0.078 0.032 0.011

[0.110] [0.114] [0.066] [0.060]

R&D 0.013 0.007 0.060 0.049

[0.060] [0.062] [0.048] [0.041]

PPE -0.044 -0.036 -0.011 0.003

[0.036] [0.038] [0.023] [0.020]

Ln(Firm age) 0.007 0.011** -0.010*** -0.008***

[0.005] [0.005] [0.003] [0.003]

Industry × Year FEs Yes Yes Yes Yes

Constant 0.329*** 0.308*** 0.158*** 0.131***

[0.095] [0.100] [0.049] [0.047]

Observations 8,576 8,576 8,576 8,576

Adj R2 37% 36% 14% 14%

F-statistic of the test: (a) = (b) 9.92*** 11.12*** 4.77** 5.21**

58

Panel B: Managerial short-termism caused by the takeover market (1) (2) (3) (4) Explore60 Explore80 Exploit60 Exploit80 High short-termism public (a) -0.109*** -0.114*** 0.041*** 0.043***

[0.013] [0.013] [0.009] [0.008]

Low short-termism public (b) -0.030** -0.036*** 0.015* 0.018**

[0.013] [0.013] [0.009] [0.008]

Ln(Patent stock) -0.059*** -0.065*** 0.025*** 0.020***

[0.002] [0.002] [0.002] [0.001]

Ln(Total assets) 0.017*** 0.019*** -0.012*** -0.011***

[0.003] [0.003] [0.002] [0.002]

Leverage -0.055*** -0.048*** 0.007 0.015

[0.018] [0.018] [0.012] [0.011]

ROA 0.100*** 0.105*** -0.058*** -0.056***

[0.022] [0.022] [0.015] [0.014]

Sales growth -0.010*** -0.011*** 0.003 0.001

[0.004] [0.004] [0.003] [0.002]

Capex 0.291*** 0.278*** -0.029 -0.053

[0.083] [0.083] [0.055] [0.050]

R&D -0.117*** -0.103** 0.031 0.034

[0.041] [0.041] [0.030] [0.026]

PPE -0.125*** -0.121*** 0.029 0.044**

[0.028] [0.028] [0.020] [0.019]

Ln(Firm age) 0.000 0.004 -0.008*** -0.006**

[0.004] [0.004] [0.003] [0.002]

Year FEs Yes Yes Yes Yes

Constant 0.537*** 0.467*** 0.259*** 0.213***

[0.047] [0.047] [0.032] [0.029]

Observations 15,192 15,192 15,192 15,192

Adj R2 24% 25% 7% 6%

F-statistic of the test: (a) = (b) 108.84*** 106.22*** 22.65*** 27.57***

59

Panel C: Managerial short-termism caused by the presence of transient institutional investors (1) (2) (3) (4) Explore60 Explore80 Exploit60 Exploit80 High short-termism public (a) -0.062*** -0.068*** 0.031*** 0.034***

[0.012] [0.013] [0.009] [0.008]

Low short-termism public (b) -0.050*** -0.061*** 0.022** 0.023***

[0.012] [0.013] [0.009] [0.008]

Ln(Patent stock) -0.055*** -0.062*** 0.026*** 0.021***

[0.002] [0.003] [0.002] [0.002]

Ln(Total assets) 0.016*** 0.018*** -0.014*** -0.013***

[0.003] [0.003] [0.002] [0.002]

Leverage -0.024 -0.017 0.003 0.009

[0.017] [0.018] [0.012] [0.011]

ROA 0.066*** 0.080*** -0.048*** -0.047***

[0.021] [0.022] [0.016] [0.014]

Sales growth -0.006* -0.008** 0.003 0.001

[0.004] [0.004] [0.003] [0.003]

Capex 0.054 0.061 0.055 0.042

[0.084] [0.085] [0.056] [0.051]

R&D -0.013 -0.020 0.023 0.012

[0.041] [0.042] [0.032] [0.028]

PPE -0.045 -0.046 -0.020 -0.007

[0.029] [0.030] [0.020] [0.018]

Ln(Firm age) 0.006* 0.010*** -0.012*** -0.009***

[0.004] [0.004] [0.003] [0.002]

Industry × Year FEs Yes Yes Yes Yes

Constant 0.361*** 0.327*** 0.293*** 0.266***

[0.123] [0.123] [0.061] [0.058]

Observations 14,855 14,855 14,855 14,855

Adj R2 32% 32% 11% 11%

F-statistic of the test: (a) = (b) 4.16** 1.41 4.10** 5.89**

60

Table 8 The difference in information environment and innovation strategies The innovative firm sample consists of 13,463 public firm-year observations and 1,729 private firm-year observations from 1997-2008. Innovative firms in year t are required to have at least one patent over the three-year period from year t-2 to year t. Definitions of all variables are provided in Appendix 1. All dollar values are in 2008 dollars. All continuous variables are winsorized at the 1st and 99th percentiles. We divide the innovative public firm sample into high and low analyst coverage subsamples based on the median number of analyst earnings forecasts. We then run Table 3 Panel A regressions by replacing the public firm indicator variable with the high analyst coverage public firm (above the median) and low analyst coverage public firm (below the median) indicator variables. The F-statistics to test the equality of coefficients on High analyst-coverage public = Low analyst-coverage public are provided at the bottom of the table. Robust standard errors clustered at the firm level are reported in brackets. Superscripts ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. (1) (2) (3) (4) Explore60 Explore80 Exploit60 Exploit80 High analyst-coverage public (a) -0.038*** -0.045*** 0.018* 0.021** [0.014] [0.014] [0.010] [0.009] Low analyst-coverage public (b) -0.067*** -0.076*** 0.032*** 0.032*** [0.012] [0.012] [0.009] [0.008] Ln(Patent stock) -0.056*** -0.062*** 0.027*** 0.022*** [0.002] [0.003] [0.002] [0.002] Ln(Total assets) 0.015*** 0.016*** -0.013*** -0.012*** [0.003] [0.003] [0.002] [0.002] Leverage -0.019 -0.010 0.002 0.007 [0.017] [0.018] [0.012] [0.011] ROA 0.059*** 0.071*** -0.044*** -0.044*** [0.021] [0.022] [0.016] [0.014] Sales growth -0.004 -0.006 0.002 -0.000 [0.004] [0.004] [0.003] [0.003] Capex 0.048 0.055 0.064 0.049 [0.084] [0.085] [0.057] [0.052] R&D -0.015 -0.025 0.028 0.017 [0.041] [0.042] [0.032] [0.028] PPE -0.050* -0.049* -0.018 -0.005 [0.029] [0.030] [0.020] [0.018] Ln(Firm age) 0.005 0.009** -0.010*** -0.008*** [0.004] [0.004] [0.003] [0.002]

Industry × Year FEs Yes Yes Yes Yes

Constant 0.382*** 0.357*** 0.276*** 0.254*** [0.125] [0.125] [0.062] [0.059] Observations 14,599 14,599 14,599 14,599 Adj R2 33% 32% 12% 12% F-statistic of the test: (a) = (b)

10.65*** 11.35*** 4.65** 3.71*