Corporate Innovation Strategy and Stock Price Crash Risk

Journal of Corporate Finance 53 (2018) 155–173
Contents lists available at ScienceDirect
Journal of Corporate Finance

journal homepage: www.elsevier.com/locate/jcorpfin
Corporate innovation strategy and stock price crash risk☆

T
Ning Jia
⁎
School of Economics and Management, Tsinghua University, Weilun Building 201F, Beijing 100081, China
ARTICLE INFO ABSTRACT
Keywords: We examine the association between corporate innovation strategy and future stock price crash
Innovation strategy risk. Using a large sample of US firms for the period 1992–2012, we find that exploration-or-
Exploration iented (exploitation-oriented) firms are more (less) prone to stock price crash risk. An ex-
Exploitation amination of underlying mechanisms suggests that compared with exploitative firms, exploratory
Crash risk
firms incur a higher failure-to-success ratio and are less likely to disclose interim negative news
Agency problems
Governance quality
about their innovation projects. The documented relationships are stronger for firms with more
severe agency problems and lower governance quality. Our findings advance the understanding
JEL Classification:
of the capital market consequences of corporate innovation strategy and major catalysts for stock
O31
price crash risk.
O34
G12
1. Introduction
As competition intensifies and the pace of change accelerates, corporations continuously renew themselves and seek new sources
of growth by investing in innovation. However, their strategies have considerable differences. The management literature has
identified two generic types of innovation: exploratory innovation and exploitative innovation (Levinthal and March, 1993; McGrath,
2001; Benner and Tushman, 2002). Exploration involves disruptive changes, path breaking, and the pursuit of new opportunities.
Exploitation, in contrast, pertains to “the refinement and extension of existing technologies and paradigms” (March, 1991).1
As succinctly summarized by March (1991), the distinction between “exploration of new possibilities” and “exploitation of old
certainties” captures several fundamental differences in firm behavior that significantly affect its performance. Prior research has
examined the impact of corporate innovation strategy on new product development and financial performance (e.g., Katila and
Ahuja, 2002; Uotila et al., 2009), but less is known about the capital market implications. In this paper we focus on one important
capital market consequence—firm-specific stock price crash risk.
Crash risk captures higher moments of the stock return distribution, that is, extreme negative returns (Callen and Fang, 2015a;
Kim et al., 2014). Understanding what causes stock crashes is important because such risk cannot be mitigated through portfolio
diversification (Kim et al., 2014), and investors demand higher expected returns for stocks with more negative skewness as a reward
for accepting this risk (Harvey and Siddique, 2000). Prior studies have attempted to uncover major determinants of crash risk (see,
☆
Ning Jia gratefully acknowledge the financial support of the National Natural Science Foundation of China (Project 71,372,049) and BNP
Paribas-Tsinghua SEM Center for Globalization of Chinese Enterprises. The author remains responsible for any remaining errors or omissions.
⁎
Corresponding author.
E-mail address: [email protected].
1
Since the seminal work of March (1991), innovation strategy and the conceptual distinction between exploration and exploitation have been
studied in a wide range of management research areas, including strategic management (e.g., Winter and Szulanski, 2001), organization theory
(e.g., Holmqvist, 2004; He and Wong, 2004), and managerial economics (e.g., Ghemawat and Ricart i Costa, 1993).
https://doi.org/10.1016/j.jcorpfin.2018.10.006
Received 12 March 2018; Received in revised form 7 October 2018; Accepted 12 October 2018
Available online 21 October 2018
0929-1199/ © 2018 Elsevier B.V. All rights reserved.
N. Jia Journal of Corporate Finance 53 (2018) 155–173
e.g., Jin and Myers, 2006; Hutton et al., 2009; Kim et al., 2011a, 2011b; Kim and Zhang, 2016). The general idea is that managers
have various incentives and opportunities to withhold bad news for an extended period. When accumulated bad news is finally
revealed to the market, it happens at once and triggers a large negative drop in stock price.
In a recent survey paper, Habib et al. (2017) notes that “some stocks are potentially more prone to crash due to the fundamental
nature of their operations” (p. 8). In this study we focus on one important area of firm operation—innovation. As elaborated in the
next section, we expect exploratory firms to have a higher stock price crash risk than exploitative firms for at least two reasons. First,
the strategic management literature notes that exploration involves more asymmetric payoffs than exploitation (Levinthal and March,
1993; McGrath, 2001). Although breakthrough inventions are associated with greater financial gains when successful, they are also
characterized by higher failure-to-success ratio than innovations based on exploitation (March, 1991). As such, exploratory firms tend
to generate more bad news, which is the main antecedent of crash risk (He and Wong, 2004).
In addition to directly affecting crash risk owing to its fundamental nature, exploratory innovation may also indirectly affect crash
risk by exacerbating agency-related incentives to hide bad news. Novel inventions tend to create significant knowledge and an
information gap between the firm and outsiders (Rindova and Petkova, 2007; Kaplan and Tripsas, 2008). A more opaque information
environment makes it easier for self-interested managers to engage in bad news hoarding, which in turn can increase stock price crash
risk.
We test the above conjecture by examining a sample of publicly traded US firms for the period 1992–2012. We focus on in-
novation-intensive firms, that is, firms that filed at least one patent during our sample period. Following prior studies (e.g., Katila and
Ahuja, 2002; Benner and Tushman, 2003; Custódio et al., 2017; Balsmeier et al., 2017), our main proxy for the extent to which a firm
adopts an exploratory innovation strategy is EXPLORE, calculated as the number of exploratory patents filed in a given year divided
by the number of all patents filed by the firm in the same year.2 We classify a patent as exploratory if at least 60% of its citations are
based on new knowledge (i.e., citations not in the firm's existing knowledge base). A higher value of EXPLORE indicates a greater
intensity of exploratory innovation.
To ensure the robustness of our findings, we construct another measure for exploitative innovation intensity, EXPLOIT, which is
calculated as the number of exploitative patents filed in a given year divided by the number of all patents filed by the firm in the same
year. We classify a patent as exploitative if at least 60% of its citations are based on existing knowledge. A higher value of EXPLOIT
indicates a greater intensity of exploitative innovation.
Following prior studies (e.g., Chen et al., 2001), we employ two measures of firm-specific stock price crash risk: the negative
skewness of firm-specific weekly returns (NCSKEW) and the asymmetric volatility of negative and positive stock returns (DUVOL).
Our baseline results suggest a positive and significant relationship between exploration intensity and one-year-ahead stock price
crash risk. In contrast, we find a significant and negative relationship between exploitation intensity and stock price crash risk. The
economic effect is sizable: on average, an increase of one standard deviation in EXPLORE in year t is associated with an increase of
0.046 in NCSKEW and 0.011 in DUVOL in year t + 1. In comparison, the mean and median values of NCSKEW are 0.020 and − 0.029,
respectively. The mean and median values of DUVOL are −0.009 and − 0.020, respectively.
One challenge in interpreting our baseline findings is that the association between corporate innovation strategy and crash risk
could be driven by unobservable characteristics that are related to both a firm's choice of innovation strategy and crash risk.
Therefore, in addition to using the lagged values of innovation strategy and control variables in the regressions, we conduct two other
tests to alleviate the endogeneity concern, including firm fixed effects and a change specification (i.e., examine how changes in
exploration or exploitation intensity affect changes in future crash risk). These two analyses provide consistent and supportive
evidence of a significantly positive (negative) effect of exploration (exploitation) intensity on crash risk.
Next, we provide evidence on plausible mechanisms through which innovation strategy affects stock price crash risk. As discussed
earlier, we expect exploratory firms to have a higher failure-to-success ratio and to disclose less interim negative news about their
innovation projects. These characteristics and behavior render exploratory firms more prone to crash risk. To explore the first
mechanism, we obtain citation information for every patent obtained by our sample firms and consider patents with no future
citations as failures. In contrast, patents with least one future citation are considered successes. We find that exploratory firms incur a
significantly higher failure-to-success ratio than exploitative firms.
To examine whether exploratory firms disclose less interim negative news about their innovation projects than exploitative firms,
we search the LexisNexis Newspapers and Wires for disclosure of innovation activities made by our sample firms. In particular, we
focus on information about the progress of innovation projects to identify interim bad news (e.g., major milestone of R&D, details
about pipeline projects, implementation of R&D projects, whether R&D projects are on schedule). We find supportive evidence that
exploratory firms disclose less interim negative news about their innovation projects than exploitative firms.
We also posit that information opaqueness associated with exploration may exacerbate agency-related incentives to hide bad
news, which in turn causes crash risk. To test this conjecture, we examine two agency-related factors: CEO age and equity-based
incentives. Prior studies highlight a negative relationship between CEO age and the severity of agency problems. Serfling (2014) and
Li et al. (2017) for example, find that younger CEOs undertake bolder investment projects. Prior studies also show that CEO's equity-
based incentives are positively associated with the severity of agency problems (e.g., Cheng and Farber, 2008; Cheng and Warfield,
2
Although exploration and exploitation are two distinct types of innovation strategy, prior studies also suggest that companies rarely make
exclusive choice between them. March (1991) suggests that maintaining an appropriate balance between exploration and exploitation is critical for
firm survival and prosperity. Therefore, instead of using an indicator variable to partition between exploratory and exploitative innovation strategy,
we examine the intensity of exploration vs. exploration.
156
2005). So we expect the relationship between exploration and crash risk to be more pronounced for firms with younger CEO and
CEOs who faces stronger equity-based incentives. Our findings support these conjectures.
Hiding negative news may be easier for firms with poor governance quality. To test this conjecture we consider three proxies of
governance quality: board busyness, G-index, and analyst coverage. Prior studies have documented a negative relationship between
board busyness and governance quality (e.g., Fich and Shivdasani, 2006; Core et al., 1999). G-index, developed by Gompers et al.
(2003), counts a firm's anti-takeover provisions (ATPs). A high G-index value represents strong managerial power and thus lower
governance quality. He and Tian (2013) find that firms with lower analyst coverage are subject to less external monitoring and thus
signals lower governance quality. So we expect the relationship between exploration and stock price crash risk to be stronger for firms
with busy board, higher G-index, and lower analyst coverage. Our findings largely support these conjectures. Together, these cross-
sectional tests also help alleviate the endogeneity concern.
Our sample is limited to firms that file and obtain at least one patent during the sample period. However, not all firms file patents.
To alleviate the concern of sample selection bias, we compare an array of characteristics between firms with and without patents. We
do not find significant differences in crash risk between these two groups of firms.
Finally, we explore other sources of external financing for firms with different innovation strategies. We document a negative
relation between exploration and debt issue. However, exploratory firms are associated with a higher likelihood of corporate venture
capital financing.
Our study contributes to the literature in two ways. First, it adds to the growing literature on corporate innovation by providing
evidence from capital markets that complements prior research on the consequences of corporate innovation strategy (Katila and
Ahuja, 2002; Uotila et al., 2009; Jia, 2017). Sorescu and Spanjol (2008) study the impact of innovation strategy on corporate
valuation and find that exploratory innovation benefits shareholders as shown by increases in both profits and long-term abnormal
returns. However, our findings suggest that exploratory innovation also leads to higher incidences of extreme negative returns, which
has not been previously documented in the strategic management literature. Another closely related paper is Habib and Hasan
(2017). They focus on broad business strategy and find prospectors are more prone to future crash risk than defenders. Our study
differs from theirs in several ways. To start, we focus on strategy for a specific type of business activity. A composite business strategy
measure can introduce confounding factors. Moreover, focusing on specific business activity allows us to provide more direct evi-
dence on the underlying mechanisms (i.e., failure-to-success ratio and negative news disclosure in this case). In addition, we examine
agency-related and governance-related moderating factors, which are not studied in Habib and Hasan (2017).
Second, our study adds to the literature on stock price crashes by identifying exploratory innovation strategy as a potential
catalyst. The study also provides useful guidance for stock market investors to help them discern investment risk based on the firm's
business orientation. Retail investors tend to concentrate investments in a small number of firms, and stock price crashes of firms in
their portfolios can be highly detrimental to their personal wealth. We highlight exploratory innovation as a precursor of crash risk.
The remainder of the paper proceeds as follows. Section 2 reviews the related literature and develops the hypotheses. Section 3
describes the data and presents descriptive statistics. Section 4 reports baseline empirical results. Section 5 addresses the endogeneity
issues and provides results of additional analyses. Section 6 concludes the paper.
2. Literature review and hypothesis development
2.1. Literature on corporate innovation strategy
Since the seminal work of March (1991) and Levinthal and March (1993), exploration vis-à-vis exploitation innovation strategy
has become a major research theme in the strategic management literature. Benner and Tushman (2002) and He and Wong (2004)
differentiate between exploratory innovation, which involves a shift to different technological trajectories, and exploitative in-
novation, which involves improvements of the current approach or technological trajectory. One stream of literature studies the
determinants of corporate choice between exploratory and exploitative innovation strategy. Firms have been shown to be less (more)
likely to engage in exploration (exploitation) when they pursue economies of scale (Crossan et al., 1999), when their innovative
activities are more likely to be subjects of imitation (Cohen and Levinthal, 1994), and when their environment appears to be less
volatile (McGrath, 2001).
Another stream of literature examines the performance consequences of different innovation strategies. Uotila et al. (2009) find
an inverted U-shaped relationship between the relative share of explorative orientation and financial performance, and they also find
that the relationship is positively moderated by the R&D intensity of the industry in which the firm operates. He and Wong (2004)
examine how exploratory and exploitative innovation can jointly influence firm performance. Based on a sample of 206 manu-
facturing firms, they find that the interaction between explorative and exploitative innovation is positively related to sales growth
rate.
In comparison, extant literature has offered less insight on the capital market consequences of corporate innovation strategy.
Several studies have examined its impact on analyst coverage and forecast behavior. Benner (2010) shed early light on this topic by
examining how financial analysts respond to radical technological changes, using the shift to digital technology in photography and
the advent of Voice over Internet Protocol (VoIP) technology in wireline telecommunications. She finds that analysts are more
attentive and positive toward corporate strategies that extend and preserve the existing technology (which is exploitative in nature)
than toward strategies that respond directly to the new technology (which is exploratory in nature). Litov et al. (2012) examine
uniqueness in corporate strategy, which is measured as the distance of the sales “distribution” for each firm from the centroid of its
counterparts in the primary industry. Using a sample of firms between 1985 and 2007, they find that uniqueness in corporate strategy
157
increases the cost of collecting and analyzing information to evaluate the firm's future value and therefore results in lower analyst
coverage. They also find that uniqueness in corporate strategy is associated with higher firm value. Jia (2017) finds that firms
pursuing an exploration-oriented innovation strategy (as opposed to an exploitation-oriented innovation strategy) are associated with
lower analyst coverage and higher forecast error and dispersion. The effect is less pronounced for firms with greater disclosure of
innovation activities and for firms followed by analysts with more firm-specific experience. We add to the literature on capital market
consequences of innovation strategy by examining its impact on stock price crash risk.
2.2. Hypothesis development
Conceptually, stock price crash risk is based on the argument that corporate managers delay the disclosure of negative news for an
extended period for strategic or nonstrategic reasons. When accumulated bad news is finally released to the market, it causes a large
negative drop in stock price. A growing body of research has focused on identifying major catalysts of crash risk, which can be largely
classified into five groups (Habib et al., 2017): (i) financial reporting and corporate disclosures (e.g., Hutton et al., 2009; Chen et al.,
2017), (ii) managerial incentives and managerial characteristics (e.g., Kim et al., 2011a; He, 2015), (iii) capital market transactions
(e.g., Callen and Fang, 2015a), (iv) corporate governance mechanisms (e.g., Andreou et al., 2016; Yuan et al., 2016), and (v) informal
institutional mechanisms (e.g., Cao et al., 2016; Callen and Fang, 2015b).
In this study we take the perspective that some companies are fundamentally more prone to crash given the nature of their
business activities and operations. In particular, we focus on corporate innovation activities and posit that exploration-oriented firms
are more prone to stock price crash risk than exploitation-oriented firms for two reasons.
First, exploration entails more asymmetric payoffs (i.e., higher failure-to-success ratio) than exploitation (March, 1991). Ex-
ploratory firms that successfully innovate at a breakthrough level can increase the likelihood that they will dominate the market and
thereby build a sustainable competitive edge; however, as O'Reilly and Tushman (2003) note, “exploration, by its nature, is inefficient
and is associated with an unavoidable increase in the number of bad ideas.” Because experiments and risk taking form the foundation
of exploratory innovation, it is characterized by more failures than successes and the associated returns are “uncertain, distant, and
often negative” (March, 1991). In contrast, exploitation leverages a firm's existing technological or knowledge base (March, 1991).
Such strategy reduces variation in that fewer projects incur negative NPV realizations and the performance is more stable. In
summary, because exploration entails higher failure-to-success ratio than exploitation, exploratory firms tend to generate more bad
news, which is the main antecedent of crash risk (He and Wong, 2004; Manso, 2011).
In addition to directly affecting crash risk because of its fundamental nature, exploratory innovation may also have an indirect
effect by exacerbating agency-related incentives to hide bad news. A knowledge gap and information asymmetry between the firm
and outsiders is particularly severe for exploration (He and Wong, 2004). In contrast, exploitation pertains to incremental im-
provement of existing technologies and/or products, and more information is available about this type of innovation (e.g., track
record information or prior performance data). The more opaque information environment associated with exploration makes it
easier for self-interested managers to engage in bad news hoarding, which can also cause stock price crash risk.
Based on the above discussion, we expect exploration-oriented firms to be more prone to future stock price crashes than ex-
ploitation-oriented firms. We state our first hypothesis below:
Hypothesis 1. Exploration-oriented firms are more prone to future firm-specific stock price crash risk than exploitation-oriented
firms.
Prior studies have found that agency-related incentives are a major cause of crash risk (e.g., Kim et al., 2011a; He, 2015). We
therefore expect it to be easier for managers with greater self-interest to hide bad news in exploratory firms that are associated with a
more opaque information environment, which in turn increases stock price crash risk. We state our second hypothesis below:
Hypothesis 2. The relation between innovation strategy and future firm-specific stock price crash is stronger for firms with more
severe agency problems.
Prior studies have shown that strong corporate governance can deter managerial misconduct. Board monitoring, for example, has
been shown to effectively constrain earnings management (Beasley, 1996; Peasnell et al., 2005; Marra et al., 2011). We therefore
expect the relation between innovation strategy and crash risk to be stronger for firms with poorer governance quality. We state our
third hypothesis below:
Hypothesis 3. The relation between innovation strategy and future firm-specific stock price crash is stronger for firms with lower
governance quality.
158
3. Sample selection and summary statistics
3.1. Sample selection
Our sample includes publicly listed US firms for the period 1992–2012. Since we study innovation-intensive firms, we exclude
firms that never filed a single patent with the United States Patent and Trademark Office (USPTO) during our sample period. We
collect firm-year patent and citation information from the Google USPTO Bulk Downloads.3 This database provides rich information
on all patents granted by the USPTO, including patent application and grant date, patent assignee name, the technology class of the
patent, and detailed information on subsequent patents that cite the focal patent.
We supplement this dataset with firm financial information from Compustat and stock price data from Center for Research in
Security Prices (CRSP). Financial institutions (Standard Industrial Classification [SIC] codes from 6000 to 6999) are excluded because
their financial statements tend to be influenced by factors unique to the financial industry. Following prior research (e.g., Kim et al.,
2011b), we also exclude firms with a year-end share price lower than $1. After further excluding observations without data to
compute other variables used in the multivariate regression analyses, our final sample consists of 13,310 firm-year observations.
3.2. Variable measurement
3.2.1. Measuring stock price crash risk

We employ two measures of firm-specific stock price crash risk. Both measures are based on firm-specific weekly returns esti-
mated as the residuals from the market model. Using firm-specific returns ensures that our crash risk measures reflect firm-specific
factors rather than broad market movements. Specifically, we estimate the following regression:
r j, = j + 1, j rm, 2 + 2, j rm, 1 + 3, j rm, + 4, j rm, + 1 + 5, j rm, + 2 + j, (1)
where rj, τ is the return on stock j in week τ, and rm, τ is the return on the CRSP value-weighted market index in week τ. The lead and
lag terms for the market index return are included to allow for nonsynchronous trading (Dimson, 1979). The firm-specific weekly
return for stock j in week τ (Wj, τ) is calculated as the natural logarithm of one plus the residual return from Eq. (1).4
Our first measure of crash risk is the negative conditional skewness of firm-specific weekly returns over the fiscal year NCSKEW,
which is calculated by taking the negative of the third moment of firm-specific weekly returns for each year and normalizing it by the
standard deviation of firm-specific weekly returns raised to the third power. Specifically, for each firm j in year t, NCSKEW is
calculated as
3 3/2
NCSKEWj, t = n (n 1) 2 W 3j, /[(n 1)(n 2) ( Wj2, ) ] (2)
where Wj, τ is firm-specific weekly return as defined above, and n is the number of weekly returns during year t. The construct is
multiplied by negative one such that a higher value of NCSKEW indicates higher crash risk.
Our second measure of crash risk is the down-to-up volatility measure (DUVOL) of the crash likelihood. For each firm j over a
fiscal-year period t, firm-specific weekly returns are separated into two groups: “down” weeks, when the returns are below the annual
mean, and “up” weeks, when the returns are above the annual mean. The standard deviation of firm-specific weekly returns is
calculated separately for each of these two groups, and DUVOL is the natural logarithm of the ratio of the standard deviation in the
down weeks to the standard deviation in the “up” weeks:
DUVOLj, t = log (nu 1) Wj2, /(nd 1) Wj2,

Down Up (3)
where nu and nd are the number of up and down weeks in year t, respectively. A higher value of DUVOL indicates greater crash risk. As
suggested in Chen et al. (2001), DUVOL does not involve third moments and hence is less likely to be overly influenced by extreme
weekly returns.
3.2.2. Measuring innovation strategy

We construct two patent-based measures of innovation strategy. Following prior studies (e.g., Katila and Ahuja, 2002; Benner and
Tushman, 2003; Balsmeier et al., 2017), we define exploratory and exploitative patents according to the extent to which a firm's new
patents use current versus new knowledge. A firm's existing knowledge consists of its previous patent portfolio and the set of patents
that have been cited by the firm's patents filed over the past five years. A patent is categorized as exploitative if at least 60% of its
citations are based on current knowledge; a patent is categorized as exploratory if at least 60% of its citations are based on new
knowledge (i.e., citations not in the firm's existing knowledge base).
We then calculate a firm's exploration intensity for a given firm-year (EXPLORE) as the number of exploratory patents filed in a
given year divided by the number of all patents filed by the firm in the same year. A higher value of EXPLORE indicates a higher
exploratory innovation intensity. We also construct a counterpart measure to capture the intensity of exploitative innovation.
3
available at http://www.google.com/googlebooks/uspto.html.
4
The residuals from Eq. (1) are highly skewed, so we perform a log transformation to obtain a more symmetric distribution (Hutton et al., 2009).
159
Specifically, we define the intensity of exploitative patents for a given firm-year (EXPLOIT) as the number of exploitative patents filed
in a given year divided by the number of all patents filed by the firm in the same year. A higher value of EXPLOIT indicates higher
exploitative innovation intensity.
3.2.3. Measuring control variables

We control for a vector of variables that have been shown to impact firm-specific stock price crash risk. Chen et al. (2001) find
that firms with high stock turnovers are associated with higher stock price crash risk. So we include the detrended stock trading
volume (DTURN), which is a proxy for investor heterogeneity, or the difference of opinions among investors. We also include the
lagged value of (NCSCKEW) to control for the potential serial correlation of stock price crash risk (Kim et al., 2011b). Chen et al.
(2001) find that past returns also help to forecast crash risk. The predictive power of past returns can be explained by a bubble
buildup as indicated by high past returns, followed by a large price drop when prices fall back to fundamentals. We thus control for
past returns (RET), calculated as the mean of firm-specific weekly returns over the fiscal year. For a similar reason, we control for the
market-to-book ratio (MB) because glamour stocks (those with a high MB) are also predicted to have higher crash risk.
The predictive power of firm size has been documented in several studies (e.g., Harvey and Siddique, 2000; Chen et al., 2001),
hence we control for firm size (SIZE), calculated as the natural logarithm of total assets. Following Kim et al. (2014) and Callen and
Fang (2015a), we also control for stock volatility (SIGMA), calculated as the standard deviation of firm-specific weekly returns over
the fiscal year, because more volatile stocks are likely to be more crash prone. In addition, we control for firm leverage (LEV),
calculated as total long-term debts divided by total assets, and profitability measured by return on assets (ROA). Our last control
variable is abnormal accruals, a proxy for earnings management, because Hutton et al. (2009) show that earnings management is
positively related to future crash risk. We measure abnormal accruals as the residuals from the modified Jones model (Dechow et al.,
1995), estimated by each year and each 2-digit SIC code industry. We use the absolute value of abnormal accruals (ACCM) in our
regression analysis. Detailed variable definitions are provided in Appendix 1.
3.3. Sample description and summary statistics
Table 1 reports sample distribution by industry in Panel A, summary statistics of main variables used in subsequent analyses in
Panel B, and the associated correlation matrix in Panel C. Panel A tabulates the top ten industries of our sample, with industry
classification being based on the 2-digit SIC code. The largest sector is Electronic & Other Electric Equipment (SIC code 36), followed
by Industrial and Commercial Machinery and Computer Equipment (SIC code 35) and Chemicals and Allied Products (SIC code 28),
respectively. Firms in these industry sectors tend to be more innovative and produce more patents.
Table 1, Panel B provides summary statistics of variables used in the main analyses. To minimize the effect of outliers, we
winsorize all continuous variables at the 1st and 99th percentiles. On average, a firm has a crash risk NCSKEW of 0.02 and DUVOL of
−0.009. The average exploratory intensity EXPLORE is 0.662, while the average exploitative intensity EXPLOIT is 0.175.5 The
average change in monthly trading volume (as a percentage of shares outstanding) is 0.064. The average firm in our sample has a
firm-specific weekly return of −17.5%, a natural logarithm of total assets of 7.06, a market-to-book ratio of 3.62, a weekly return
volatility of 0.05, a leverage of 0.19, and a return on assets of 0.13. The average absolute value of abnormal accruals is 0.14.
Table 1, Panel C reports the Pearson correlation matrix for main variables used in subsequent analyses. The two crash risk
measures NCSKEW and DUVOL have a high correlation of 0.96, which is comparable to that reported in prior studies (e.g., Chen et al.,
2001; Callen and Fang, 2015a). Although these two measures are constructed differently from firm-specific daily returns, they appear
to capture the same underlying construct. More importantly, both NCSKEW and DUVOL have a significant and positive association
with exploration intensity EXPLORE, and they both have a significant and negative association with exploitation intensity EXPLOIT.
Because EXPLORE and EXPLOIT capture opposite innovation strategy, they have a significantly negative correlation. The correlation
matrix provides preliminary evidence that supports our first hypothesis. Because univariate correlation analysis does not take into
account the effects of the other correlated variables, we consider the evidence to be suggestive and rely on subsequent multivariate
analyses to draw inferences.
4. Baseline empirical results
4.1. Baseline results
To examine the effect of corporate innovation strategy on stock price crash risk, we estimate the following models using the
ordinary least squares (OLS):
NCSKEW / DUVOLi, t + 1 = + EXPLOREi, t + Controli, t + Yeart + Industryj + i, t (4)
5
Our summary statistics on exploration and exploitation intensity are similar to those reported by other studies (e.g., Jia and Tian, 2018). We
believe a plausible reason for relatively higher value of exploration is that although exploratory innovation is much more costly and risky, it is also
associated with significantly higher future financial payoff when successful. Prior studies also note that exploratory innovation is associated with
higher reputational payoffs than exploitative innovation (Schmidt and Calantone, 1998). So, firms have incentives to invest in exploratory in-
novation in order to maintain sustainable advantage in the long run.
160
Table 1
Sample distribution and summary statistics.
Panel A: Sample distribution by industry
SIC Code Industry Number of Obs. Percentage
36 Electrical and Electronic Equipment 2425 18.22%

35 Industrial and Commercial Machinery and Computer Equipment 1893 14.22%
28 Chemicals and Allied Products 1723 12.95%
38 Instruments and Related Products 1505 11.31%
73 Business Services 1389 10.44%
37 Transportation Equipment 718 5.39%
20 Food and Kindred Products 453 3.40%
24 Lumber And Wood Products, Except Furniture 349 2.62%
26 Paper And Allied Products 287 2.16%
13 Oil and Gas Extraction 259 1.95%
Other – 2309 17.35%
Total 13,310 100.00%
Panel B: Summary statistics
Variable 25% Median Mean 75% S.D. N
NCSKEWt+1 −0.416 −0.029 0.020 0.386 0.748 13,310

DUVOLt+1 −0.247 −0.020 −0.009 0.214 0.357 13,310
EXPLOREt 0.364 0.750 0.662 1 0.363 13,310
EXPLOITt 0 0 0.175 0.286 0.263 13,310
DTURNt −0.226 0.031 0.064 0.335 1.036 13,310
SIGMAt 0.032 0.045 0.053 0.065 0.029 13,310
RETt −0.210 −0.101 −0.175 −0.051 0.228 13,310
SIZEt 5.832 6.930 7.059 9.185 1.679 13,310
MBt 1.686 2615 3.615 4.217 4.519 13,310
LEVt 0.027 0.168 0.187 0.288 0.171 13,310
ROAt 0.091 0.141 0.128 0.195 0.092 13,310
ACCMt 0.048 0.087 0.143 0.152 0.252 13,310
Panel C: Correlation matrix
Variable 1 2 3 4 5 6 7 8 9 10 11 12
1 NCSKEWt+1 1
2 DUVOLt+1 0.96c 1
3 EXPLOREt 0.02b 0.02b 1
4 EXPLOITt −0.01b −0.01b −0.79a 1
5 DTURNt 0.01c 0.02c 0.01 −0.02a 1
6 SIGMAt 0.02b 0.01 0.01c −0.04c 0.13c 1
7 RETt −0.00 0.00 0.01 0.01 −0.14a −0.94a 1
8 SIZEt 0.03a 0.04a −0.07a 0.08a −0.03a −0.45a 0.35a 1
9 MBt 0.03a 0.03a −0.01 −0.01 0.08a 0.051a −0.05a 0.01 1
10 LEVt −0.01 −0.01 −0.02a 0.01 −0.01 −0.07a 0.03a 0.26a −0.07a 1
11 ROAt 0.05a 0.06a 0.05a −0.04a 0.05a −0.40a 0.38a 0.25a 0.14a −0.08a 1
12 ACCMt −0.01 −0.01 −0.02b 0.01 0.08a 0.18a −0.17a −0.15a 0.14a −0.01a −0.22a 1
This table reports sample distribution by industry (Panel A), as well as summary statistics (Panel B) and correlation matrix (Panel C) among main
variables used in this study. The sample includes all U.S. firms covered by Compustat and CRSP (excluding financial institutions and firms with a
year-end share price that is lower than $1). The final sample consists of 13,310 firm-year observations for the period of 1992–2012.
Note: Pearson correlations are reported. c, b, and a indicate significance at the 10%, 5%, and 1% levels, respectively. Definitions of variables are
provided in Appendix 1.
NCSKEW / DUVOLi, t + 1 = + EXPLOITi, t + Controli, t + Yeart + Industryj + i, t (5)
where i indexes firm, j indexes industry, and t indexes time. The dependent variables are stock price crash risk that firm i incurs in
year t + 1. The main variable of interest is EXPLORE or EXPLOIT in year t. Control is a vector of firm attributes that could affect a
firm's stock price crash risk as discussed in Section 3.2.3. Year and Industry capture year and industry (2-digit SIC) fixed effects,
respectively.6 We cluster standard errors at the firm and year level.
Table 2 reports results from the regression analysis of the relationship between corporate innovation strategy and stock price
6
As a robustness test, we also conducted the baseline analyses using 4-digit SIC for industry effects. Results are qualitatively the same.
161
Table 2
Corporate innovation strategy and stock price crash risk: Baseline results.
Panel A: Exploratory intensity
Dep Var = NCSKEWt+1 DUVOLt+1
(1) (2) (3) (4)
EXPLOREt 0.061*** 0.062*** 0.027** 0.030**

(0.024) (0.023) (0.012) (0.012)
DTURNt 0.001 −0.001
(0.006) (0.003)
NCSKEWt 0.271*** 0.125***
(0.012) (0.005)
SIGMAt 2.356*** 0.969**
(0.902) (0.436)
RETt 0.266*** 0.106**
(0.100) (0.048)
SIZEt 0.012** 0.007***
(0.006) (0.002)
MBt 0.057*** 0.003***
(0.002) (0.001)
LEVt −0.026 −0.010
(0.051) (0.024)
ROAt 0.335*** 0.204***
(0.097) (0.047)
ACCMt 0.008 0.015
(0.032) (0.016)
Constant −0.011 −0.223*** −0.027* −0.145***
(0.042) (0.071) (0.020) (0.035)
Year and Industry Fixed Effects Included Included Included Included
R2 0.02 0.09 0.02 0.09
Observations 13,310 13,310 13,310 13,310
Panel B: Exploitative intensity
Dep Var = NCSKEWt+1 DUVOLt+1
(1) (2) (3) (4)
EXPLOITt −0.062** −0.055** −0.033** −0.028*

(0.030) (0.027) (0.016) (0.016)
DTURNt 0.000 −0.001
(0.006) (0.003)
NCSKEWt 0.271*** 0.125***
(0.012) (0.005)
SIGMAt 2.360*** 0.974**
(0.902) (0.436)
RETt 0.267*** 0.107**
(0.100) (0.048)
SIZEt 0.012** 0.008***
(0.006) (0.003)
MBt 0.006*** 0.003***
(0.002) (0.001)
LEVt −0.028 −0.011
(0.051) (0.024)
ROAt 0.332*** 0.203***
(0.097) (0.047)
ACCMt 0.005 0.014
(0.032) (0.016)
Constant 0.041 −0.165** −0.004 −0.120***
(0.035) (0.067) (0.017) (0.033)
R2 0.02 0.09 0.02 0.09
Observations 13,310 13,310 13,310 13,310
This table reports OLS regression estimates of stock price crash risk on corporate innovation strategy. The key independent variable in Panel A and
Panel B is exploration intensity EXPLORE and exploitation intensity EXPLOIT, respectively. The dependent variables NCSKEW and DUVOL are
measured in year t + 1, and all independent variables are measured in year t. Definitions of variables are provided in Appendix 1. Year and industry
fixed effects are included. Robust standard errors clustered by firm and year are displayed in parentheses. ***, **, and * denote significance at the 1,
5, and 10% levels, respectively.
162
Table 3
Corporate innovation strategy and stock price crash risk: Firm fixed effects.
Dep Var = NCSKEWt+1 DUVOLt+1 NCSKEWt+1 DUVOLt+1
(1) (2) (3) (4)
EXPLOREt 0.059** 0.028** – –

(0.029) (0.014) – –
EXPLOITt – – −0.064* −0.023
– – (0.039) (0.018)
DTURNt 0.004 0.001 0.003 0.001
(0.006) (0.003) (0.006) (0.003)
NCSKEWt 0.194*** 0.090*** 0.194*** 0.090***
(0.012) (0.005) (0.012) (0.005)
SIGMAt −0.533 −0.417 −0.561 −0.430
(1.095) (0.516) (1.094) (0.515)
RETt 0.104 0.039 0.100 0.037
(0.111) (0.053) (0.111) (0.053)
SIZEt 0.081*** 0.034*** 0.082*** 0.035***
(0.017) (0.008) (0.017) (0.008)
MBt 0.007*** 0.003*** 0.007*** 0.003***
(0.002) (0.001) (0.002) (0.001)
LEVt −0.047 −0.024 −0.048 −0.025
(0.071) (0.034) (0.071) (0.034)
ROAt 0.391*** 0.210*** 0.386*** 0.208***
(0.135) (0.065) (0.135) (0.065)
ACCMt 0.008 0.020 0.006 0.019
(0.035) (0.017) (0.035) (0.017)
Constant −0.529*** −0.245*** −0.478*** −0.221***
(0.128) (0.061) (0.125) (0.059)
Year and Firm Fixed Effects Included Included Included Included
R2 0.19 0.19 0.19 0.19
Observations 13,310 13,310 13,310 13,310
This table reports OLS regression estimates of stock price crash risk on corporate innovation strategy using firm fixed effects. The dependent
variables NCSKEW and DUVOL are measured in year t + 1, and all independent variables are measured in year t. Definitions of variables are
provided in Appendix 1. Year and firm fixed effects are included. Robust standard errors clustered by firm and year are displayed in parentheses.
***, **, and * denote significance at the 1, 5, and 10% levels, respectively.
crash risk. Panel A reports results on the impact of exploratory intensity. We start with a parsimonious model in columns (1) and (3)
that only includes the main variable of interest EXPLORE as well as industry and year fixed effects. The coefficient estimate on
EXPLORE is significantly positive in both columns, suggesting that firms with higher exploration intensity are associated with higher
future stock price crash risk. In columns (2) and (4) we examine the full model as specified in Eq. (4), and we continue to observe a
positive and significant coefficient on EXPLORE. The economic effect is sizable: results in column (2) suggest that on average, an
increase of one standard deviation in EXPLORE in year t is associated with an increase of 0.046 in NCSKEW in year t + 1. In
comparison, the mean and median values of NCSKEW are 0.020 and − 0.029, respectively. Results in column (4) suggest that on
average, an increase of one standard deviation in EXPLORE in year t is associated with an increase of 0.011 in DUVOL in year t + 1. In
comparison, the mean and median values of DUVOL are −0.009 and − 0.020, respectively. We find opposite results when examining
exploitative intensity in Panel B.
The coefficients on the control variables are generally consistent with prior studies. Firms that are larger and have a higher past
price crash risk, a higher past return and return volatility, a higher market-to-book ratio, and a higher ROA are associated with higher
future crash risk. Overall, results in Table 2 support Hypothesis 1 that exploration-oriented firms are more prone to future stock price
crash risk than exploitation-oriented firms.
5. Additional tests
5.1. Endogeneity
A major concern with our baseline results is that they are driven by omitted variables that affect both a firm's choice of innovation
strategy and its crash risk. In the baseline test, we regress one-year-ahead crash risk on current-year innovation strategy as well as
control variables to mitigate such concern. In this section, we attempt to further address the endogeneity issue using two alternative
approaches.
Our first approach is the inclusion of firm fixed effects, which absorb time-invariant firm unobservable characteristics and help
mitigate the unobservable heterogeneity problem. Results are reported in Table 3. Results from the fixed effects regressions are
largely consistent with the baseline analyses.
Our second approach to the endogeneity problem is a change specification; that is, we regress changes in crash risk on changes in
163
Table 4
Corporate innovation strategy and stock price crash risk: Change specification.
Dep Var = ΔNCSKEWt+1 ΔDUVOLt+1 ΔNCSKEWt+1 ΔDUVOLt+1
(1) (2) (3) (4)
ΔEXPLOREt 0.082** 0.038**

(0.038) (0.018)
ΔEXPLOITt −0.080* −0.031
(0.047) (0.022)
ΔDTURNt 0.006 0.001 0.006 0.001
(0.008) (0.004) (0.007) (0.004)
ΔNCSKEWt 0.237*** 0.108*** 0.236*** 0.107***
(0.012) (0.005) (0.012) (0.005)
ΔSIGMAt −1.166 −0.756 −1.185 −0.762
(1.377) (0.640) (1.368) (0.641)
ΔRETt 0.077 0.017 0.074 0.015
(0.145) (0.067) (0.145) (0.067)
ΔSIZEt 0.197*** 0.093*** 0.197*** 0.093***
(0.029) (0.014) (0.029) (0.014)
ΔMBt 0.009*** 0.004*** 0.009*** 0.004***
(0.003) (0.001) (0.003) (0.001)
ΔLEVt −0.237** −0.122** −0.242** −0.124**
(0.112) (0.051) (0.112) (0.052)
ΔROAt 0.772*** 0.414*** 0.772*** 0.414***
(0.181) (0.084) (0.180) (0.085)
ΔACCMt 0.033 0.032 0.030 0.031
(0.050) (0.024) (0.050) (0.024)
Constant 0.008 0.001 0.008 0.001
(0.108) (0.001) (0.108) (0.049)
R2 0.08 0.08 0.08 0.08
Observations 8547 8547 8547 8547
This table reports OLS regression estimates of changes in stock price crash risk on changes in innovation strategy. The dependent variables
ΔNCSKEW t+1 and ΔDUVOL t+1 are measured between year t + 1 and year t + 4. Independent variables are measured between year t and year
t + 3. Definitions of variables are provided in Appendix 1. Year and industry fixed effects are included. Robust standard errors clustered by firm and
year are displayed in parentheses. ***, **, and * denote significance at the 1, 5, and 10% levels, respectively.
the intensity of exploration (exploitation) and control variables. To alleviate the concern that corporate innovation strategy can be
sticky such that annual change is small, we employ a change specification over a three-year window; that is, the dependent variables
ΔNCSKEWt+1 and ΔDUVOLt+1 are changes in crash risk between year t + 1 and year t + 4. The independent variables ΔEXPLOREt
and ΔEXPLOITt and the control variables are changes between year t and year t + 3. Results are reported in Table 4. An increase in
exploratory intensity is associated with an increase in future firm-specific crash risk. In contrast, an increase in exploratory intensity
is associated with a decrease in future firm-specific crash risk.
5.2. Innovation strategy and failure-to-success ratio
In this section we test the conjecture that exploratory firms experience a higher failure-to-success ratio than exploitative firms,
which is a main antecedent of stock price crashes. We obtain forward citation information for every patent obtained by our sample
firms. The citations that a patent receives from subsequent patents are called forward citations. Prior studies have shown that forward
citations are positively and significantly related to the value of the patent (e.g., Carpenter et al., 1981; Trajtenberg, 1990). Therefore
we define patents with no forward citations as failure. The rationale is that these patents, although filed to protect the firm's in-
tellectual property and legally granted by the USPTO, have no commercial value and thus do not enhance value for the firm, which is
effectively a failure from business perspective. In contrast, patents with at least one subsequent citation are considered as successes.
We calculate the failure-to-success ratio for each firm in each year and examine its relation with innovation strategy. As reported
in Table 5, exploratory innovation entails a significantly higher failure-to-success ratio than exploitative innovation. Such funda-
mental nature of exploratory innovation makes exploratory firms more prone to crash risk.
5.3. Innovation strategy and interim disclosure of negative news
In this section we provide evidence that exploratory firms disclose less interim bad news about their innovation projects than
exploitative firms, which makes them more prone to future crash risk. To operationalize the inquiry, we search the LexisNexis for
disclosure of innovation activities made by our sample firms (using company name and the following keywords: “research,” “de-
velopment,” “product,” “service,” “innovation,” and “invention”).
Gu and Li (2003) classified disclosure into three categories. We used the same categories: (1) information about progress of
164
Table 5
Comparison of failure-to-success ratio between exploratory and exploitative firms.
Panel A: Exploratory intensity
Dep Var = Failure-to-Success Ratio t+1
(1) (2)
EXPLOREt 0.035*** 0.036***

(0.007) (0.008)
RETt −0.000
(0.009)
SIZEt 0.004***
(0.001)
MBt −0.000
(0.000)
LEVt −0.010
(0.013)
ROAt 0.021
(0.023)
R&Dt 0.031
(0.028)
Constant −0.013* −0.046***
(0.007) (0.012)
Year and Industry Fixed Effects Included Included
R2 0.36 0.38
Observations 13,310 13,310
Panel B: Exploitative intensity
Dep Var = Failure-to-Success Ratio t+1
(1) (2)
EXPLOITt −0.067*** −0.084***

(0.017) (0.018)
RETt 0.000
(0.010)
SIZEt 0.004***
(0.001)
MBt −0.000
(0.000)
LEVt −0.008
(0.013)
ROAt 0.023
(0.023)
R&Dt 0.037
(0.029)
Constant 0.019*** −0.011
(0.004) (0.011)
R2 0.36 0.39
This table reports OLS regression estimates of failure-to-success ratio on corporate innovation strategy. The
key independent variable in Panel A and Panel B is exploration intensity EXPLORE and exploitation intensity
EXPLOIT, respectively. The dependent variables are failure-to-success ratio, where failure is measured by
number of patents with zero future citations and success is measured by number of patents with at least one
future citation, respectively. Definitions of variables are provided in Appendix 1. Year and industry fixed
effects are included. Robust standard errors clustered by firm and year are displayed in parentheses. ***, **,
and * denote significance at the 1, 5, and 10% levels.
165
Table 6
Corporate innovation strategy and disclosure of interim negative innovation news.
Panel A: Management Disclosure of Innovation Activities per Firm-year
Disclosure type Mean Median Sd.Dev
Information about progress of innovation (e.g., major milestone of R&D; details of pipeline projects or new products under 0.49 0 1.38
development; turnover of key scientists and details of research teams; indication of whether R&D projects are on schedule)
Negative News 0.10 0 0.34
Panel B: Innovation Strategy and Disclosure of Interim negative news
Dep Var= NegativeNewst+1
(1) (2)
EXPLOREt −0.040** –
(0.018) –
EXPLOITt – 0.032**
– (0.016)
Constant and Controls Included Included
Year & Industry Fixed Effects Included Included
Pseudo R2 0.12 0.12
Observations 4330 4330
This table reports OLS regression estimates of corporate innovation strategy and percentage of negative interim innovation-related news. Panel A
presents the distribution of management disclosure of innovation progress and negative news per firm-year. Panel B reports regression estimate of
innovation strategy on disclosure of negative interim news. Definitions of other variables are provided in Appendix 1. Robust standard errors
clustered by firm are displayed in parentheses. ***, **, and * denote significance at the 1, 5, and 10% levels, respectively.
innovation (e.g., major milestones of R&D, details of pipeline projects, details of research teams, whether R&D projects are on
schedule); (2) information about completion/commercialization of innovation (e.g., new product launch, licensing and royalty,
transfer or sale of technology); and (3) information about strategy of innovation (e.g., goal, objective, or plan of innovation; relation
with current innovation; time frame; acquisition of other firms for new technology or other innovation capabilities). Untabulated
statistics suggest that there is more disclosure about progress of innovation, followed by completion/commercialization of innovation
and information about strategy of innovation, respectively. We focus on disclosure about the progress of innovation projects (Type I
disclosure) in order to identify interim bad news.
To provide an example of interim negative news related to innovation projects, in March 2010 A.P. Pharma, Inc. (Nasdaq Ticker:
APPA), a specialty pharmaceutical company, announced that it had received a Complete Response Letter from the US Food and Drug
Administration regarding its New Drug Application (NDA) for APF530, a treatment for the prevention of acute and delayed-onset
chemotherapy-induced nausea and vomiting. The news release notes that “questions remain that preclude the approval of the NDA in
its current form…A.P. Pharma is in the process of preparing a resubmission responsive to the deficiencies listed in the Complete
Response Letter.”7
Table 6 Panel A presents the overall distribution of disclosure per firm-year about the progress of innovation and the disclosure of
negative news. The mean negative news is relatively small, suggesting that the disclosure of negative interim news is not a high-
frequency event. In Panel B, we examine the impact of corporate innovation strategy on disclosure of interim negative news. The
dependent variable is the number of interim negative news items as a percentage of all news items released by a firm in a year about
the progress of its innovation projects. Following the disclosure literature (e.g., Bamber and Cheon, 1998; Ali et al., 2014), we control
for a vector of firm and industry characteristics that may affect management disclosure practice.8 The coefficient estimate on EX-
PLORE is significantly negative, while the coefficient estimate on EXPLOIT is significantly positive. Taken together, the evidence
supports our conjecture that exploratory firms disclose less interim bad news about their innovation projects than exploitative firms.
5.4. Cross-sectional analyses
In this section, we examine how the relation between innovation strategy and crash risk varies with the severity of agency
problems (Hypothesis 2) and governance quality (Hypothesis 3).
7
See news release “FDA Denies Approval of A.P. Pharma, Inc. Nausea Drug” at https://www.biospace.com/article/releases/fda-denies-approval-
of-a-p-pharma-inc-nausea-drug-/
8
Including firm size, institutional ownership, market-to-book ratio, return-on-assets ratio, negative earnings, leverage, asset tangibility, stock
return volatility over the prior year, as well as capital expenditure scaled by total assets. We suppress the coefficients of control variables for brevity.
166
Table 7
Relation between innovation strategy and crash risk, and the moderating role of agency-related factors.
Panel A: CEO age
Dep Var = NCSKEWt+1 DUVOLt+1 Failure-to-Success Ratio t+1 NegativeNewst+1
(1) (2) (3) (4)
EXPLOREt 0.350** 0.097* 0.131*** −0.047**

(0.164) (0.058) (0.050) (0.023)
EXPLOREt╳CEOAGEt −0.005** −0.004*** −0.002** 0.010*
(0.002) (0.002) (0.001) (0.006)
CEOAGEt 0.003 0.001 0.001 0.008
(0.002) (0.001) (0.001) (0.007)
Constant and Control Included Included Included Included
R2 0.09 0.09 0.39 0.14
Observations 11,564 11,564 11,564 4010
Panel B: CEO equity-based incentives
ep Var = NCSKEWt+1 DUVOLt+1 Failure-to-Success Ratio t+1 NegativeNewst+1
(1) (2) (3) (4)
EXPLOREt 0.047** 0.020* 0.037*** −0.032*

(0.024) (0.012) (0.009) (0.017)
EXPLOREt╳EQUITYt 0.002** 0.001* −0.000 −0.002**
(0.000) (0.000) (0.000) (0.001)
EQUITYt −0.001 −0.001 0.000 −0.002
(0.001) (0.001) (0.000) (0.003)
R2 0.09 0.09 0.39 0.14
Observations 12,177 12,177 12,177 4128
This table reports OLS regression estimates of two outcome variables (NCSKEW and DUVOL) and two mediating variables (Failure-to-Success Ratio
and NegativeNews) on exploratory intensity, and the moderating role of agency related factors. The agency-related factor is CEO age in Panel A, and
CEO equity-based incentives in Panel B. Definitions of variables are provided in Appendix 1. Year and industry fixed effects are included. Robust
standard errors clustered by firm and year are displayed in parentheses. ***, **, and * denote significance at the 1, 5, and 10% levels, respectively.
5.4.1. Agency problems

We use two proxies to measure the severity of agency problems (i.e., CEO's self-interest). The first proxy is CEO age. Prior studies
highlight the relevance of CEO age for agency problems. Serfling (2014) and Li et al. (2017) find that because of career concerns,
younger CEOs undertake bolder investment projects. Huang et al. (2012) find that older CEOs are associated with higher-quality
financial reporting.9
The second proxy is CEO equity-based incentives. The wealth of managers who receive stock-based compensation is sensitive to
their firm's stock prices (e.g., Cheng and Farber, 2008; Cheng and Warfield, 2005). While this sensitivity can motivate managers to
make value-increasing decisions, it can also induce managerial short-termism and thus create agency problems.10
We report the results in Table 7. While the relation between exploration innovation and crash risk (i.e., NCSKEW and DUVOL)
continue to be significantly positive, the interaction term between them and CEO age is significantly negative in Panel A, and the
interaction term between them and CEO equity-based incentives is significantly positive in Panel B. Results on the two mediating
variables (Failure-to-Success Ratio and NegativeNews) are largely consistent. Together, these findings support Hypothesis 2 that the
relation between exploration and stock price crash risk is stronger for firms with managers who are more self-interested.
5.4.2. Governance quality

We use three proxies to measure governance quality. The first proxy is board busyness. Fich and Shivdasani (2006) and Core et al.
(1999), among others, provide evidence suggesting that busy boards, with directors who take on multiple directorships, may not
effectively monitor management and thus indicate poor governance quality. The second proxy is a comprehensive governance quality
9
The mean and median of CEO age in our sample are 55 and 56, respectively.
10
The mean and median of CEO equity-based incentives in our sample, measured as the natural logarithm of CEO's total portfolio vega (computed
as the change in the value of the CEO's equity holdings for a 1% change in the standard deviation of stock returns in a given year), are 4.83 and 4.29,
respectively.
167
Table 8
Relation between innovation strategy and crash risk, and the moderating role of governance quality.
Panel A: Director busyness
(1) (2) (3) (4)
EXPLOREt 0.034* 0.018* 0.035*** −0.032*

(0.019) (0.010) (0.013) (0.018)
EXPLOREt╳BUSYt 0.352** 0.125* 0.073** −0.027**
(0.172) (0.075) (0.036) (0.012)
BUSYt −0.311 −0.122 −0.064 0.014
(0.250) (0.125) (0.051) (0.037)
R2 0.08 0.08 0.02 0.11
Observations 7931 7931 7931 2236
Panel B: G-index
(1) (2) (3) (4)
EXPLOREt 0.045** 0.016* 0.072** −0.053**

(0.021) (0.009) (0.035) (0.022)
EXPLOREt╳GINDEXt 0.047** 0.017* 0.010* −0.030*
(0.022) (0.010) (0.006) (0.017)
GINDEXt −0.034 −0.014 −0.006 0.003
(0.026) (0.010) (0.005) (0.008)
R2 0.10 0.10 0.36 0.14
Observations 5279 5279 5279 1973
Panel C: Analyst coverage
(1) (2) (3) (4)
EXPLOREt −0.017 −0.006 0.042* 0.004

(0.006) (0.031) (0.023) (0.012)
EXPLOREt╳ANALYSTt 0.029 0.012 0.004 0.002
(0.025) (0.012) (0.009) (0.006)
ANALYSTt 0.047* 0.025** 0.006 0.014*
(0.024) (0.012) (0.009) (0.008)
R2 0.09 0.09 0.40 0.12
Observations 11,104 11,104 11,104 3988
This table reports OLS regression estimates of two outcome variables (NCSKEW and DUVOL) and two mediating variables (Failure-to-Success Ratio
and NegativeNews) on exploratory intensity, and the moderating role of governance quality related factors. The governance-related factor is director
busyness in Panel A, and G-index in Panel B, and analyst coverage in Panel C. Definitions of variables are provided in Appendix 1. Year and industry
fixed effects are included. Robust standard errors clustered by firm and year are displayed in parentheses. ***, **, and * denote significance at the 1,
5, and 10% levels.
168
Table 9
Comparison between firms with and without patents during the sample period.
Panel A: Univariate comparison
Firms with Patents Firms without Patents Difference
No. of Obs. Mean (1) S.D. Median No. of Obs. Mean (2) S.D. Median t-test (1)–(2)
Crash Risk
NCSKEW 29,696 −0.016 0.799 −0.050 172,031 −0.131 0.865 −0.135 0.115***
DUVOL 29,696 −0.025 0.377 −0.034 172,031 −0.073 0.407 −0.083 0.048***
Firm Characteristics
Sales 28,003 6.831 1.922 6.800 124,701 4.861 2.381 4.812 1.970***
ROA 27,886 0.122 0.096 0.136 125,557 0.054 0.141 0.093 0.068***
LEV 27,918 0.194 0.179 0.172 127,288 0.225 0.229 0.172 −0.031***
Capex 27,719 0.053 0.048 0.040 125,843 0.064 0.090 0.038 −0.011***
PPE/Assets 28,022 0.241 0.182 0.193 127,661 0.281 0.249 0.196 −0.040***
TobinQ 27,962 2.304 2.504 1.706 126,622 2.709 63.461 1.446 −0.405
MB 27,927 3.384 4.300 2.459 125,274 2.971 4.975 1.875 0.413***
ACCM 27,994 0.145 0.247 0.088 127,956 0.224 0.361 0.123 −0.079***
RET 29,641 −0.175 0.246 −0.096 174,209 −0.378 0.536 −0.185 0.203***
EQUITY 21,451 19.250 45.491 0.840 17,095 10.496 30.982 0.000 8.754***
HHI 28,056 0.245 0.198 0.185 128,376 0.221 0.180 0.163 0.245***
Panel B: Multivariate analysis
Dep Var =
NCSKEWt+1 DUVOLt+1
PatentFirmst 0.006 0.001

(0.007) (0.003)
Constant and Controls Included Included
R2 0.16 0.16
This table compares characteristics of firms that filed at least one patent during the sample period of this study and those of firms with no
patents during the sample period of this study. Definitions of variables are provided in Appendix 1. ***, **, and * denote significance at the 1, 5,
and 10% levels, respectively.
measure, G-index developed by Gompers et al. (2003), which counts a firm's anti-takeover provisions (ATPs).11 A high G-index value
represents strong managerial power and thus lower governance quality. The third proxy is analyst coverage. Firms with lower analyst
coverage are subject to less external monitoring (e.g., He and Tian, 2013).12
We report the results in Table 8. While the relation between exploration innovation and crash risk continue to be significantly
positive, the interaction term between them and busy board (in Panel A) and G-Index (in Panel B) are also significantly positive. We
do not find significant results for analyst coverage. Together, these results largely support Hypothesis 3 that the relation between
exploration and stock price crash risk is stronger for firms with poor governance quality.
5.5. Comparison with no-patent firms
Our sample is limited to firms that file and obtain at least one patent during the sample period. However, not all firms file patents.
To alleviate the concern of sample selection bias, we compare an array of characteristics and crash risk between firms that filed at
least one patent during our sample period (“patent firms”) and firms in Compustat that do not own any patent during our sample
period (“no-patent firms”).
We report the results in Table 9. In terms of crash risk, the univariate comparison in Panel A shows that patent firms have higher
11
RiskMetrics provides data on the G-index up to 2006. Our sample period encompasses four publications (2000, 2002, 2004, and 2006).
Following Gompers et al. (2003), we assume that during the years between publications, a firm has the same ATPs it had in the previous publication
year. We therefore replace the missing values of the G-index with the values reported in the previous publication year.
12
The mean and median of busy board in our sample, defined as percentage of independent directors holding three or more directorships in listed
companies, are 28.1% and 0%, respectively. The mean and median of G-index in our sample are 9 and 9.23, respectively. The mean and median of
analyst coverage, defined as the natural logarithm of one plus the number of analysts that issue earnings forecasts for the firm, are 2.24 and 2.30,
respectively.
169
Table 10
Innovation strategy and access to external financing.
Panel A: Access to debt financing
Dep. Var = Debt Issuet+1 Net Debt Issue t+1
(1) (2) (3) (4)
EXPLOREt −0.011* −0.002

(0.006) (0.003)
EXPLOITt 0.013* 0.001
(0.007) (0.005)
SIGMAt 0.395*** 0.396*** −0.000 0.001
(0.154) (0.153) (0.101) (0.101)
RETt 0.027* 0.027* 0.004 0.004
(0.015) (0.015) (0.012) (0.011)
SIZEt 0.008*** 0.008*** −0.001 −0.001
(0.002) (0.002) (0.001) (0.001)
MBt −0.000 −0.000 0.001** 0.002**
(0.001) (0.001) (0.001) (0.000)
ROAt 0.001 0.001 −0.009 −0.010
(0.021) (0.021) (0.014) (0.014)
ACCMt −0.004 −0.003 0.005 0.005
(0.006) (0.006) (0.006) (0.006)
Log(PatNum)t −0.010*** −0.010*** 0.001 0.001
(0.001) (0.001) (0.000) (0.000)
Constant 0.045*** 0.034*** 0.008 0.009
(0.015) (0.013) (0.009) (0.009)
R2 0.03 0.03 0.03 0.02
Observations 12,214 12,214 12,867 12,867
Panel B: Access to corporate venture capital
Dep. Var = Prob(CVC = 1)
(1) (2)
EXPLOREt 0.058*
(0.032)
EXPLOITt −0.102*
(0.057)
SIZEt 0.493*** 0.457***
(0.052) (0.049)
ROAt 0.494*** −0.114
(0.052) (0.553)
LEVt −0.865** −0.904***
(0.353) (0.353)
ACCMt 0.159 0.261**
(0.108) (0.114)
Constant −5.510*** −4.883***
(0.814) (0.781)
Psudo-R2 0.23 0.23
Observations 1737 1737
This table reports OLS and Probit regression estimates of access to external financing on corporate innovation strategy. In
Panel A, the dependent variable DebtIssue is long-term debt issuance over total book assets. NetDebtIssue is calculated as
long-term debt issuance net of long-term debt reduction over total book assets. In Panel B, the dependent variable CVC is a
dummy variable that equals 1 if a firm has corporate venture capital backing as of the IPO year. Definitions of variables are
provided in Appendix 1. Robust standard errors clustered by firm and year are displayed in parentheses. ***, **, and *
denote significance at the 1, 5, and 10% levels, respectively.
crash risk than no-patent firms. However, because the univariate comparison does not take into account other factors that could affect
firm crash risk, in Panel B we conduct a regression analysis with the same set of control variables as in our baseline models.
PatentFirms is an indicator variable that equals 1 for firms that filed at least one patent during our sample period. As shown in Panel B,
the coefficient on PatentFirms is insignificant across both columns. There does not appear to be systematic differences in stock price
crash risk between firms with or without patents.
Also, compared to no-patent firms, patent firms are larger and more profitable and have lower leverage, lower Capex and PPE
intensity, higher market-to-book ratio, lower accruals management, and higher return and market concentration. Also interestingly,
170
patent firms on average have greater CEO vega, which is consistent with the argument that CEOs are incentivized to engage in risky
activities (i.e., innovation).
5.6. Other sources of external financing
We also explore whether exploratory and exploitative firms differ in terms of other sources of external financing. In particular, we
examine access to debt financing and corporate venture capital. We measure access to debt financing by using two measures based on
Denis and Sibilkov (2010). The first measure DebtIssue is long-term debt issuance over total book assets. The second measure Net-
DebtIssue is calculated as long-term debt issuance net of long-term debt reduction over total book assets. The results are reported in
Panel A of Table 10. We find some evidence that exploratory (exploitative) innovation is negatively (positively) associated with debt
issue. One plausible explanation is that exploratory innovation is associated with higher risk and the associated future payoff is highly
uncertain, which is not desirable from a creditor's perspective. Therefore, exploratory firms likely experience greater external fi-
nancing friction.13
Many innovation-intensive firms are currently backed by corporate venture capital (CVC) who invests strategically in startups
whose main businesses align closely with a corporation's own business directives. Chemmanur et al. (2014) find that CVC helps their
portfolio firms achieve a higher degree of innovation productivity. Therefore, we examine whether exploratory and exploitative firms
have differential access to CVC as of the IPO year.14 We report the probit model results in Panel B. We find some evidence that
exploratory firms are associated with a higher likelihood of CVC financing, which is a major type of equity financing for companies
before IPO.
5.7. Robustness tests
We conduct several tests to ensure the robustness of our findings. First, in the main analyses we use 60% as a threshold for the
classification of exploration and exploitation. We also use 80% as an alternative threshold, and the results remain qualitatively
unchanged. The coefficient estimate on EXPLOREt is 0.050 (p-value = .025) when the dependent variable is NCSKEWt+1 and 0.023
(p-value = .034) when the dependent variable is DUVOLt+1. The coefficient estimate on EXPLOITt is −0.077 (p-value = .019) when
the dependent variable is NCSKEWt+1 and − 0.034 (p-value = .070) when the dependent variable is DUVOLt+1.
Second, in the main analyses we relate innovation strategy measures in year t to stock price crash risk measures in year t + 1. For
a robustness check, we examine the relation between innovation strategy in year t and two-year ahead crash risk (i.e., year t + 2). The
coefficient estimate on EXPLOREt is 0.059 (p-value = .008) when the dependent variable is NCSKEWt+2 and 0.026 (p-value = .013)
when the dependent variable is DUVOLt+2. The coefficient estimate on EXPLOITt is −0.073 (p-value = .024) when the dependent
variable is NCSKEWt+2 and − 0.027 (p-value = .076) when the dependent variable is DUVOLt+2.
Third, in the main analyses, we exclude financial firms (SIC 6000–6999) because these firms are regulated and their financial
characteristics are systematically different from firms in other industries. For a robustness check, we include those firms and our
findings remain qualitatively the same. The coefficient estimate on EXPLOREt is 0.049 (p-value = .022) when the dependent variable
is NCSKEWt+1 and 0.022 (p-value = .050) when the dependent variable is DUVOLt+1. The coefficient estimate on EXPLOITt is
−0.052 (p-value = .045) when the dependent variable is NCSKEWt+1 and − 0.023 (p-value = .088) when the dependent variable is
DUVOLt+1.
6. Concluding remarks
In this paper, we examine the impact of corporate innovation strategy on firm-specific stock price crash risk. Using a sample of
innovation-intensive US firms during the period of 1992–2012, we find that firms with high exploration intensity are associated with
higher future stock price crash risk. In contrast, firms with high exploitation intensity are associated with lower future crash risk. We
use firm fixed effects and a change specification to address the endogeneity concern and our results remain robust.
An examination of underlying mechanisms suggests that exploratory firms incur higher failure-to-success ratio than exploitative
firms, and they are less likely to disclose interim negative news about their innovation projects. We also find that the documented
relationships are stronger for firms with more severe agency problems and poorer governance quality.
Findings of this paper advance our understanding of how a public firm's choice of innovation strategy affects its performance in
the capital markets. In particular, our findings help managers of public companies understand more fully the capital market con-
sequences that they may face when attempting to develop a competitive edge based on certain innovation strategy. Findings of this
paper also enhance our understanding of the determinants of crash risk. We highlight corporate innovation strategy as an important
predictor of stock price crash risk, which warrants investors' attention when making investments in knowledge-intensive firms.
13
A stream of theoretical and empirical research (Froot et al., 1993; Bolton et al., 2011) suggests that external finance frictions give rise to a
precautionary or hedging motive to hold cash. A demand for internal funds arises because financial frictions limit the firm's ability or increase the
firm's cost to raise external finance. Consistent with this argument, Jia (2017) finds that exploratory firms incur a higher level of cash holdings,
greater internal financing, and lower dividend ratio than exploitative firms.
14
We focus on the IPO year because CVC, as a type of VC, is typically invested before the company goes public and may exit several years after the
IPO.
171
Appendix A. Variable definition
Variable Definition
NCSKEWt+1 The negative skewness of firm-specific weekly returns over year t + 1, calculated by taking the negative of
the third moment of firm-specific weekly returns for the year and normalizing it by the standard deviation
of firm-specific weekly returns raised to the third power;
DUVOLt+1 The down-to-up volatility. For any stock in year t + 1, we separate all the weeks with firm-specific weekly
returns below the annual mean (down weeks) from those with firm-specific weekly returns above the
period mean (up weeks) and compute the standard deviation for each of these subsamples separately. The
down-to-up volatility is measured as the natural logarithm of the ratio of the standard deviation in the
down weeks to the standard deviation in the up weeks.
Failure-to-Success Ratio between failed patents (i.e., patents with zero future citations) and successful patents (i.e., patents
Ratiot+1 with at least one future citations) for a firm in year t + 1;
NegativeNewst+1 Ratio of negative news to all news issued by a firm about the progress of its innovation projects in year
t + 1;
EXPLOREt Number of exploratory patents filed in year t divided by the number of all patents filed by the firm in the
same year; a patent is classified as exploratory if at least 60% of its citations are based on new knowledge;
EXPLOITt Number of exploitative patents filed in year t divided by the number of all patents filed by the firm in the
same year; a patent is classified as exploitative if at least 60% of its citations are based on existing
knowledge;
DTURNt The detrended average monthly stock turnover in year t, calculated as the average monthly share turnover
in year t minus the average monthly share turnover in t-1, where monthly share turnover is calculated as
the monthly share trading volume divided by the number of shares outstanding over the month;
SIGMAt The standard deviation of firm-specific weekly returns over year t;
RETt The average of firm-specific weekly returns over year t;
SIZEt The natural logarithm of the book value of total assets in year t;
MBt Market-to-book ratio, calculated by the market value of equity divided by the book value of equity in year t;
LEVt Leverage ratio, defined as book value of debt divided by book value of total assets measured at the end of
year t;
ROAt Return on assets ratio, defined as operating income before depreciation divided by total assets, measured at
the end of year t;
ACCMt The absolute value of the estimated residuals from the Modified Jones model in year t;
R&Dt Research and development expenditure divided by book value of total assets measured at the end of year t;
CEOAGEt Age of a firm's CEO in year t;
EQUITYt Logarithm of change in the value of the CEO's equity holdings for a 1% change in the standard deviation of
stock returns in a given year t;
ANALYSTt Natural logarithm of one plus the number of analysts that issue earnings forecasts for the firm in year t;
INDt Percentage of independent directors in a firm in year t;
BUSYt Percentage of independent directors holding three or more directorships in listed companies in year t;
GINDEXt How many anti-takeover provisions (ATPs) a firm has in year t;
Salest The natural logarithm of revenue in year t;
CAPEXt Capital expenditure scaled by book value of total assets measured at the end of fiscal year t;
PPE/Assetst Property, Plant & Equipment divided by book value of total assets measured at the end of fiscal year t;
TobinQt Market-to-book ratio during fiscal year t, calculated as [market value of equity plus book value of assets
minus book value of equity minus balance sheet deferred taxes (set to 0 if missing)] divided by book value
of assets;
HHIt Herfindahl index of 4-digit SIC industry where a firm belongs, measured at the end of fiscal year t;
PatentFirmst An indicator variable that equals 1 for firms that own at least one patent during the sample period, 0
otherwise;
DebtIssuet+1 long-term debt issuance over total book assets in year t + 1;
NetDebtIssuet+1 long-term debt issuance net of long-term debt reduction over total book assets in year t + 1;
CVC A dummy variable that equals 1 if a firm has corporate venture capital backing as of the IPO year;
Log(PatNum)t Natural logarithm of one plus firm's total number of patents filed (and eventually granted) in year t;
Log(CitePat)t Natural logarithm of one plus citations per patent for a firm in year t;
172
References
Ali, A., Kalsa, S., Yeung, E., 2014. Industry concentration and corporate disclosure policy. J. Account. Econ. 58, 240–264.
Andreou, P.C., Antoniou, C., Horton, J., Louca, C., 2016. Corporate governance and firm-specific stock price crashes. Eur. Financ. Manag. 22, 916–956.
Balsmeier, B., Fleming, L., Manso, G., 2017. Independent boards and innovation. J. Financ. Econ. 123, 536–557.
Bamber, L., Cheon, Y.S., 1998. Discretionary management earnings forecast disclosures: antecedents and outcomes associated with forecast venue and forecast
specificity choices. J. Account. Res. 36, 167–190.
Beasley, M.S., 1996. An empirical analysis of the relation between the board of director composition and financial statement fraud. Account. Rev. 71, 443–465.
Benner, M.J., 2010. Securities analysts and incumbent response to radical technological change: evidence from digital photography and internet telephony. Organ. Sci.
21, 42–62.
Benner, M.J., Tushman, M., 2002. Process management and technological innovation: a longitudinal study of the photography and paint industries. Adm. Sci. Q. 47,
676–706.
Benner, M.J., Tushman, M., 2003. Exploitation, exploration, and process management: the productivity dilemma revisited. Acad. Manag. Rev. 28, 238–256.
Bolton, P., Chen, H., Wang, N., 2011. A unified theory of Tobin's q, corporate investment, financing, and risk management. J. Financ. 66, 1545–1578.
Callen, J.L., Fang, X., 2015a. Short interest and stock price crash risk. J. Bank. Financ. 60, 181–194.
Callen, J.L., Fang, X., 2015b. Religion and stock price crash risk. J. Financ. Quant. Anal. 50, 169–195.
Cao, C., Xia, C., Chan, K.C., 2016. Social trust and stock price crash risk: evidence from China. Int. Rev. Econ. Financ. 46, 148–165.
Carpenter, M., Narin, F., Woolf, P., 1981. Citation rates to technologically important patents. World Patent Inf. 3, 160–163.
Chemmanur, T.J., Loutskina, E., Tian, X., 2014. Corporate venture capital, value creation, and innovation. Rev. Financ. Stud. 27, 2434–2473.
Chen, J., Hong, H., Stein, J.C., 2001. Forecasting crashes: trading volume, past returns, and conditional skewness in stock prices. J. Financ. Econ. 61, 345–381.
Chen, C.L., Kim, J.B., Yao, L., 2017. Earnings smoothing: does it exacerbate or constrain stock price crash risk? J. Corp. Finan. 42, 36–54.
Cheng, Q., Farber, D.B., 2008. Earnings restatements, changes in CEO compensation, and firm performance. Account. Rev. 83, 1217–1250.
Cheng, Q., Warfield, T.D., 2005. Equity incentives and earnings management. Account. Rev. 80, 441–476.
Cohen, W.M., Levinthal, D.A., 1994. Fortune favors the prepared firm. Manag. Sci. 40, 227–251.
Core, J., Holthausen, R., Larcker, D., 1999. Corporate governance, chief executive officer compensation, and firm performance. J. Financ. Econ. 51, 371–406.
Crossan, M.M., Lane, H.W., White, R.E., 1999. An organizational learning framework: from intuition to institution. Acad. Manag. Rev. 24, 522–537.
Custódio, C., Ferreira, M.A., Matos, P., 2017. Do General Managerial Skills Spur Innovation? Management Science, Forthcoming.
Dechow, P., Sloan, R., Sweeney, A., 1995. Detecting earnings management. Account. Rev. 70, 193–225.
Denis, D., Sibilkov, V., 2010. Financial constraints, investment, and the value of cash holdings. Rev. Financ. Stud. 23, 247–269.
Dimson, E., 1979. Risk measurement when shares are subject to infrequent trading. J. Financ. Econ. 7, 197–226.
Fich, E.M., Shivdasani, A., 2006. Are busy boards effective monitors? J. Financ. 61, 689–724.
Froot, K.A., Scharfstein, D.S., Stein, J.C., 1993. Risk management: Coordinating corporate investment and financing policies. J. Financ. 48, 1629–1658.
Ghemawat, P., Ricart i Costa, J.E., 1993. The organizational tension between static and dynamic efficiency. Strateg. Manag. J. 14, 59–73.
Gompers, P., Ishii, J., Metrick, A., 2003. Corporate governance and equity prices. Q. J. Econ. 118, 107–156.
Gu, F., Li, J.Q., 2003. Disclosure of innovation activities by high-technology firms. Asia Pacific J. Account. Econ. 10, 143–172.
Habib, A., Hasan, M.M., 2017. Business strategy, overvalued equities, and stock price crash risk. Res. Int. Bus. Financ. 39, 389–405.
Habib, A., Hasan, M.M., Jiang, H., 2017. Stock price crash risk: Review of the empirical literature. Accounting and Finance, Forthcoming. Harvey and Siddique, 2000.
Harvey, C.R., Siddique, A., 2000. Conditional skewness in asset pricing tests. J. Financ. 55, 1263–1295.
He, G., 2015. The effect of CEO inside debt holdings on financial reporting quality. Rev. Acc. Stud. 20, 501–536.
He, J., Tian, X., 2013. The dark side of analyst coverage: the case of innovation. J. Financ. Econ. 109, 856–878.
He, Z., Wong, P., 2004. Exploration vs. exploitation: an empirical test of the ambidexterity hypothesis. Organ. Sci. 15, 481–494.
Holmqvist, M., 2004. Experiential learning processes of exploration and exploitation within and between organizations: an empirical study of product development.
Organ. Sci. 15, 70–81.
Huang, H.W., Ena, R., Lee, C.C., 2012. CEO age and financial reporting quality. Account. Horiz. 2012, 725–740.
Hutton, A.P., Marcus, A.J., Tehranian, H., 2009. Opaque financial reports, R2, and crash risk. J. Financ. Econ. 94, 67–86.
Jia, N., 2017. Corporate innovation strategy, analyst forecasting activities and the economic consequences. J. Bus. Financ. Acc. 44, 812–853 (Forthcoming).
Jia, N., Tian, X., 2018. Accessibility and materialization of firm innovation. J. Corp. Finan. 48, 515–541.
Jin, L., Myers, S.C., 2006. R2 around the world: new theory and new tests. J. Financ. Econ. 79, 257–292.
Kaplan, S., Tripsas, M., 2008. Thinking about technology: applying a cognitive lens to technological change. Res. Policy 37, 790–805.
Katila, R., Ahuja, G., 2002. Something old, something new: a longitudinal study of search behavior and new product introduction. Acad. Manag. J. 45, 1183–1194.
Kim, J.B., Zhang, L., 2016. Accounting conservatism and stock price crash risk: firm-level evidence. Contemp. Account. Res. 33, 412–441.
Kim, J.B., Li, Y., Zhang, L., 2011a. Corporate tax avoidance and stock price crash risk: firm-level analysis. J. Financ. Econ. 100, 639–662.
Kim, J.B., Li, Y., Zhang, L., 2011b. CFO vs. CEO: equity incentives and crashes. J. Financ. Econ. 101, 713–730.
Kim, Y., Li, H.D., Li, S.Q., 2014. Corporate social responsibility and stock price crash risk. J. Bank. Financ. 43, 1–13.
Levinthal, D.A., March, J.G., 1993. The myopia of learning. Strateg. Manag. J. 14, 95–112.
Li, X., Low, A., Makhija, A.K., 2017. Career concerns and the busy life of the young CEO. J. Corp. Finan. 47, 88–109.
Litov, L.P., Moreton, P., Zenger, T.R., 2012. Corporate strategy, analyst coverage, and the uniqueness paradox. Manag. Sci. 58, 1797–1815.
Manso, G., 2011. Motivating innovation. J. Financ. 66, 1823–1860.
March, J.G., 1991. Exploration and exploitation in organizational learning. Organ. Sci. 2, 71–87.
Marra, A., Mazzola, P., Prencipe, A., 2011. Board monitoring and earnings management pre- and post-IFRS. International Journal of Accounting 46, 205–230.
McGrath, R.G., 2001. Exploratory learning, innovative capacity, and managerial oversight. Acad. Manag. J. 44, 118–131.
O'Reilly, C.A., Tushman, M.L., 2003. Organizational ambidexterity: past, present and future. Acad. Manag. Perspect. 27, 324–338.
Peasnell, K.V., Pope, P.F., Young, S., 2005. Board monitoring and earnings management: do outside directors influence abnormal accruals? J. Bus. Financ. Acc. 32,
1311–1346.
Rindova, V.P., Petkova, A.P., 2007. When is a new thing a good thing? Technological change, product form design, and perceptions of value for product innovations.
Organ. Sci. 18, 217–232.
Schmidt, J., Calantone, R., 1998. Are really new product development projects harder to shut down? J. Prod. Innov. Manag. 15, 111–123.
Serfling, M.A., 2014. CEO age and the riskiness of corporate policies. J. Corp. Finan. 25, 251–273.
Sorescu, A., Spanjol, J., 2008. Innovation's effect on firm value and risk: Insights from consumer packaged goods. J. Mark. 72, 114–132.
Trajtenberg, M., 1990. A penny for your quotes: Patent citations and the value of innovations. RAND J. Econ. 21, 172–187.
Uotila, J., Maula, M., Thomas, K., Zahra, S.A., 2009. Exploration, exploitation, and financial performance: Analysis of S&P 500 corporations. Strateg. Manag. J. 30,
221–231.
Winter, S.G., Szulanski, G., 2001. Replication as strategy. Organ. Sci. 12, 730–743.
Yuan, R., Sun, J., Cao, F., 2016. Directors' and officers' liability insurance and stock price crash risk. J. Corp. Finan. 37, 173–192.
173

Corporate Innovation Strategy and Stock Price Crash Risk

Uploaded by

Copyright:

Available Formats

Corporate Innovation Strategy and Stock Price Crash Risk

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Corporate Innovation Strategy and Stock Price Crash Risk

Uploaded by

Copyright:

Available Formats

Journal of Corporate Finance 53 (2018) 155–173

Contents lists available at ScienceDirect

Journal of Corporate Finance

Corporate innovation strategy and stock price crash risk☆

ARTICLE INFO ABSTRACT

2. Literature review and hypothesis development

2.1. Literature on corporate innovation strategy

2.2. Hypothesis development

3. Sample selection and summary statistics

3.1. Sample selection

3.2. Variable measurement

3.2.1. Measuring stock price crash risk

DUVOLj, t = log (nu 1) Wj2, /(nd 1) Wj2,

3.2.2. Measuring innovation strategy

3.2.3. Measuring control variables

3.3. Sample description and summary statistics

4. Baseline empirical results

4.1. Baseline results

SIC Code Industry Number of Obs. Percentage

36 Electrical and Electronic Equipment 2425 18.22%

Panel B: Summary statistics

Variable 25% Median Mean 75% S.D. N

NCSKEWt+1 −0.416 −0.029 0.020 0.386 0.748 13,310

Panel C: Correlation matrix

NCSKEW / DUVOLi, t + 1 = + EXPLOITi, t + Controli, t + Yeart + Industryj + i, t (5)

Dep Var = NCSKEWt+1 DUVOLt+1

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

EXPLOREt 0.061*** 0.062*** 0.027** 0.030**

Panel B: Exploitative intensity

Dep Var = NCSKEWt+1 DUVOLt+1

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

EXPLOITt −0.062** −0.055** −0.033** −0.028*

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

EXPLOREt 0.059** 0.028** – –

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

ΔEXPLOREt 0.082** 0.038**

5.2. Innovation strategy and failure-to-success ratio

5.3. Innovation strategy and interim disclosure of negative news

Dep Var = Failure-to-Success Ratio t+1

EXPLOREt 0.035*** 0.036***

Panel B: Exploitative intensity

Dep Var = Failure-to-Success Ratio t+1

EXPLOITt −0.067*** −0.084***

Disclosure type Mean Median Sd.Dev

Panel B: Innovation Strategy and Disclosure of Interim negative news

Dep Var= NegativeNewst+1

5.4. Cross-sectional analyses

Dep Var = NCSKEWt+1 DUVOLt+1 Failure-to-Success Ratio t+1 NegativeNewst+1

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

EXPLOREt 0.350** 0.097* 0.131*** −0.047**

Panel B: CEO equity-based incentives

ep Var = NCSKEWt+1 DUVOLt+1 Failure-to-Success Ratio t+1 NegativeNewst+1

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

EXPLOREt 0.047** 0.020* 0.037*** −0.032*

5.4.1. Agency problems

5.4.2. Governance quality

Dep Var = NCSKEWt+1 DUVOLt+1 Failure-to-Success Ratio t+1 NegativeNewst+1

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

EXPLOREt 0.034* 0.018* 0.035*** −0.032*

EXPLOREt 0.061* 0.062* 0.027 0.030

EXPLOITt −0.062 −0.055 −0.033** −0.028*

EXPLOREt 0.059 0.028 – –

ΔEXPLOREt 0.082 0.038

EXPLOREt 0.035* 0.036*

EXPLOITt −0.067* −0.084*

EXPLOREt 0.350** 0.097* 0.131* −0.047

EXPLOREt 0.045** 0.016* 0.072 −0.053