Adbi wp781

ADBI Working Paper Series
MEASURING THE DEGREE

OF CORPORATE INNOVATION
David M. Reeb
No. 781
September 2017
Asian Development Bank Institute

David M. Reeb is a professor in the Department of Finance at National University of
Singapore Business School. He also serves as the director of research for the Centre for
Asset Management Research & Investments.
The views expressed in this paper are the views of the author and do not necessarily
reflect the views or policies of ADBI, ADB, its Board of Directors, or the governments
they represent. ADBI does not guarantee the accuracy of the data included in this paper
and accepts no responsibility for any consequences of their use. Terminology used may
not necessarily be consistent with ADB official terms.
Working papers are subject to formal revision and correction before they are finalized
and considered published.
The Working Paper series is a continuation of the formerly named Discussion Paper series;
the numbering of the papers continued without interruption or change. ADBI’s working
papers reflect initial ideas on a topic and are posted online for discussion. ADBI encourages
readers to post their comments on the main page for each working paper (given in the
citation below). Some working papers may develop into other forms of publication.
ADB recognizes “China” as the People’s Republic of China; “Hong Kong” as Hong Kong,
China; and “Korea” as the Republic of Korea.
In this publication, “$” refers to US dollars.
Suggested citation:
Reeb, D. M. 2017. Measuring the Degree of Corporate Innovation. ADBI Working Paper 781.
Tokyo: Asian Development Bank Institute. Available:
https://www.adb.org/publications/measuring-degree-corporate-innovation
Please contact the authors for information about this paper.
Email: dmreeb@nus.edu.sg
Asian Development Bank Institute

Kasumigaseki Building, 8th Floor
3-2-5 Kasumigaseki, Chiyoda-ku
Tokyo 100-6008, Japan
Tel: +81-3-3593-5500
Fax: +81-3-3593-5571
URL: www.adbi.org
E-mail: info@adbi.org
© 2017 Asian Development Bank Institute

ADBI Working Paper 781 D. M. Reeb
Abstract
Corporate innovation propels both company performance and economic growth. Yet,
measuring corporate innovation proves to be challenging, leading researchers to rely on a
variety of different signals, such as reported R&D expenditures, patent citations and new
product announcements. I posit that each of these signs of corporate innovation provides a
noisy, biased signal of a firm’s technological progress and capacity. Moreover, relying on
a single indicator of an activity eliminates useful information, suggesting that all of the
observable signals about corporate innovation should be included in measuring it. Using the
annual survey of senior executives by BCG/BusinessWeek to identify the most innovate
companies, I create two composite measures of corporate innovation. Finally, I evaluate how
a common use of these individual, noisy signals of innovation to capture R&D productivity
(patents scaled by R&D) influences studies on innovative efficiency. Simulation analysis
shows that scaling one noisy, biased signal of innovation by another (e.g., R&D productivity)
magnifies the noisy signal problem and leads to biased inferences. Arguably, the composite
measures, based on multiple signals of corporate innovation, provide more reliable
assessments of corporate innovation than any single indicator. Finally, I discuss the use of
composite measures of innovation in empirical research on technological innovation and the
implications for policy makers.
Keywords: innovation, R&D, patents, technological progress
JEL Classification: O30, O31, G31

Contents
1. INTRODUCTION ......................................................................................................... 1
2. CORPORATE INNOVATION AND GROWTH ............................................................ 4
2.1 Why Innovation? .............................................................................................. 4

2.2 The Economics of Innovation .......................................................................... 4
2.3 Management Research on Innovation............................................................. 5
2.4 Financing Innovation ....................................................................................... 5
3. INNOVATION DATA.................................................................................................... 6
3.1 R&D Expenditures ........................................................................................... 6

3.2 Patents ............................................................................................................ 8
3.3 New Product Announcements ....................................................................... 10
4. EVALUATING MULTIPLE SIGNALS......................................................................... 11
4.1 Composite Measures .................................................................................... 11

4.2 R&D Productivity ........................................................................................... 16
5. SUMMARY AND CONCLUSION............................................................................... 18
REFERENCES ..................................................................................................................... 21
1. INTRODUCTION
Researchers commonly view corporate innovation as a key foundation of economic
growth. Scherer (1986) describes how technological progress disrupts industries and
increases the per capita income in industrialized nations. Grossman and Helpman
(1990) observe that successful research improves a firm’s market power and profits.
Audretsch and Feldman (1996) discuss the prevalence and magnitude of R&D
spillovers, suggesting that corporate innovation exerts both direct and indirect influence
on economic growth. More generally, researchers find that corporate innovation
accelerates sales growth and increases a firm’s profitability (Franco 1989). Others
highlight, however, that these benefits and spillovers vary across ownership structures
(Anderson et al. 2017). Koh, Reeb, and Zhao (2017) report that firms with cautious
CEOs tend to engage in more innovation, while simultaneously failing to disclose their
company’s innovation activity. As a result, it is difficult to assess the determinants and
impact of innovation because of the difficulty in measuring it. Research on corporate
innovation often relies on firms’ disclosure about their R&D activities. However,
managers face a trade-off in disclosing their spending on innovation activities,
garnering a benefit from sharing it with investors and customers but incurring a cost
from informing competitors (Koh and Reeb 2015).
Academic research on innovation spans various business disciplines and also arises
in economics and sociology. While the concept of innovation remains clear, measuring
corporate innovation proves to be challenging and diverse. Accounting research
regularly focuses on R&D expenditures, while economists and management scholars
typically rely on patents and their citations. Others focus on new product
announcements to capture the technological progress of a firm (Reeb and Zhao 2017).
Unfortunately, none of these signals of corporate innovation fully capture the nature
of innovation and knowledge creation in the firm. Unsurprisingly, these individual
measures, such as R&D spending or patent citations, provide a noisy, imperfect
signal of the technological progress and capacity of the firm. In addition, each of these
signals of innovation depends on managerial disclosures about the company’s
research activities. Material R&D expenditures are a mandatory disclosure, while
patents and new product announcements represent voluntary disclosures to investors
that reinforce property rights and attract customers. Consequently, these noisy signals
of corporate innovation also suffer from a series of selection processes within the firm
and across markets.
These varying signals of innovation provide different snapshots or information about
innovation in Asian markets relative to western economies. Figure 1 shows the
percentage of innovative firms in selected stock exchanges, based on the proportion
that report R&D. Using reported R&D, the stock exchange of Taipei,China (TPEx)
appears to be one of the most innovative markets in the set, while the stock exchange
of the Republic of Korea (KRX) lags behind the markets in Mumbai (NSE) and
Hong Kong, China (HKEX). This metric indicates that over 70% of the firms in the stock
exchange of Taipei,China are engaged in innovative activities. Nevertheless, relying on
patents to signal corporate innovation provides a different view of innovative firms in
these markets. Figure 2 again ranks markets by the proportion of companies engaged
in innovation, using patents to capture or signal corporate innovation. Using this metric,
the stock exchanges of Frankfurt; New York; Taipei,China; Tokyo; and the Republic of
Korea all appear to exhibit similar magnitudes of corporate innovation, with roughly
15% to 25% of firms engaged in innovation. Figure 3 highlights the difference for each
market resulting from measuring corporate innovation, using R&D relative to patents
to signal corporate innovation activity. These separate signals provide the greatest
1
distortions in evaluating innovation activity on the stock exchanges of Taipei,China and

Tokyo, arguably because the choice to disclose innovation via reported R&D spending,
patenting activities, or product announcements differs across markets. For instance,
firms in the Taipei,China market appear to be more willing to report their R&D spending
relative to those in the stock exchange of the Republic of Korea. Relying on a single
signal of corporate innovation potentially fosters complacency about innovation among
policy makers in markets with fewer companies that fail to report R&D activities.
Figure 1: Ranking By Innovation: Proportion of Firms Reporting R&D*

(%)
* Derived from Koh et al. (2015).
Figure 2: Ranking By Innovation: Proportion of Firms Seeking Patents*

(%)
2
Figure 3: Difference in Market Ranking: Using R&D versus Patents*

(%)
To formally assess the measurement of corporate innovation, I begin with an analysis

of the importance of innovation and the types of questions that interest academics and
policy makers. In section 3, I discuss the nature of the noise in the three signals of
innovation most commonly used by empirical researchers in the social sciences. Next,
I focus on the selection biases in these measures and the nature of the managerial
choice to provide this information. The impact of signal noise and selection bias when
using a single signal of innovation then forms the motivation in section 4 to assess the
ability of these indicators of technological progress to capture or measure corporate
innovation. Relying on the annual survey of senior executives by BCG/BusinessWeek
to identify the most innovative companies, I then evaluate how well these signals
identify corporate innovation. I contend that employing multiple signals provides a
broader measure of corporate innovation. Arguably, the resulting factor-analysis-based
measure of corporate innovation and the weighted-index of corporate innovation each
provide a more reliable assessment of technological capacity than using a single signal
of innovation (Kline 2015).
In further analysis, I explore a common dual-signal approach to measuring innovation,
patents scaled by reported R&D (R&D productivity). I develop a simple simulation
analysis to gauge the effect of scaling one noisy, biased signal innovation by another
noisy, biased signal of innovation. I simulate both R&D spending and innovation
success and then create noisy signals of the true values (reported R&D and patents).
Analyzing both true R&D productivity and noisy R&D productivity reveals that the mean
and variance of reported R&D productivity are downward biased. Thus, even without
bias, scaling one noisy signal by another creates biased estimates of true R&D
productivity. Using noisy R&D productivity in growth regressions, I find that it provides
biased coefficient estimates and standard errors. Unsurprisingly, these problems are
even greater once biased signals are introduced into the analysis. In short, scaling
patents by reported R&D appears to intensify the problems in using these noisy, biased
signals of corporate innovation. Finally, in the conclusion section, I discuss how to
apply and use a multiple signal measure of corporate innovation in empirical research
on technological innovation and how this affects policy makers in Asia.
3
2. CORPORATE INNOVATION AND GROWTH

2.1 Why Innovation?
Economic research highlights the role of Schumpeter in defining or describing
the impact of innovation on economic growth and development (Ruttan 1959). This
literature builds on the notion that corporate innovation creates new products and
processes that antiquate existing products and firms. Tufano (1989) describes how
firms exploit new products to capture a greater market share. Research on
pharmaceutical firms, for instance, reveals that private-sector drug development can
add substantial value to a firm (DiMasi, Hansen, and Grabowski 2003). Other studies
emphasize that corporate innovation improves market competitiveness and leads to
persistent economic growth (Raymond et al. 2010). Aghion and Tirole (1994) show how
the allocation of property rights affects the frequency and magnitude of corporate
innovation. There is a strong consensus among both academics and practitioners that
corporate innovation drives firm performance and ultimately market capabilities and
success (Cantwell 2014).
Operationally, corporate innovation captures firms’ attempts to develop new products
or new methods to reduce the costs of goods sold or to improve the quality of existing
products (Grossman and Helpman 1990). Academics thus define innovation as
inventing a process or object, using it to develop a product or service for the market
and working to improve it in the future. Thus, innovation encompasses a variety of
different tasks. How should we then identify innovative firms? Presumably, innovative
firms are those that develop breakthrough products, innovative processes, unique
customer experiences and/or new business models (Yoon and Deeken 2013). By
definition, innovation is therefore a concept that covers multiple aspects or types
of corporate activities. Perhaps the easiest innovation to identify is a new product,
while more difficult innovations range from reductions in manufacturing costs to the
development of new applications for an existing problem. Part of the issue of identifying
innovation depends on the context of the analysis.
2.2 The Economics of Innovation

Research on the economics of innovation spans developmental economics, public
economics and industrial organization (Hall and Lerner 2010). Two common themes in
this body of research focus on determining how to encourage corporate innovation
and how to gauge the impact of this innovation on economic growth. Some of the
earliest research, such as that by Marshall (1890), emphasizes that knowledge growth
occurs in clusters and requires specialized labor pools. Hicks (1932) argues that high
labor costs spur the development of labor-saving inventions. Consistent with this
notion, Acemoglu (2010) finds that labor scarcity encourages innovation in labor-saving
devices. Building on this literature, Asheim and Coenen (2006) argue that spatial
proximity and concentration are key issues in knowledge production. Duranton and
Puga (2001) posit that innovation arises in diversified metropolitan areas and that, once
perfected, mass production moves to lower-cost, specialized cities.
Png (2017) reports that legal protection of a firm’s intellectual property spurs greater
investment in innovation. Aghion, Van Reenen, and Zingales (2005) describe how
product market competition increases the incremental profits from corporate innovation,
especially among more competitive firms. Thus, the literature on the determinants
of innovation focuses on the entirety of the innovation process. Gauging the degree
of corporate innovation to investigate its determinants requires a broad measure
4
that captures the various activities among firms. The spillover literature provides a good
example. The literature discussing the impact of corporate innovation on growth
focuses on both firm-specific and spillover effects from corporate R&D. Griffith,
Redding, and Van Reenen (2004) discover that a firm’s ability to absorb technology
transfers from others is a key determinant of success. Okamuro (2007) reports that
successful innovation depends on the coordination costs of the research project.
Nicholas (2008) finds that the valuation of knowledge capital depends on the quality
of the underlying technological inventions. Thus, the literature on innovation and
growth requires a comprehensive measure of innovation that includes both input and
output measures.
2.3 Management Research on Innovation

Building on industrial organization research, management scholars study how
managers set their innovation strategy, the types of organizational structures that
are the most effective and the way in which their innovation influences their rivals.
Cohen and Levinthal (1990) argue that innovation activity enhances a firm’s ability to
assimilate and exploit information from external sources. Moving beyond firms, others
suggest that research and development alliances represent a critical component in
developing successful innovation in industries with complex and diverse sources of
expertise (Powell, Koput, and Smith-Doerr 1996; Phelps 2010). Another stream of this
innovation literature centers on the ability of firms to fast-track their innovation process.
Eisenhardt and Tabrizi (1995) observe that firms can accelerate their innovation either
by compressing the sequential steps or by following an experiential model that relies
on improvisation and flexibility. Boudreau, Lacetera, and Lakhani (2011) find that
increasing the number of researchers reduces the incentives to exert effort but
increases the odds of finding an extreme-value solution. Koh, Reeb, and Zhao (2017)
document that cautious CEOs seek to speed up their innovation activities and
aggressively seek to keep this information from their competitors. Williamson and Yin
(2014) describe how firms in the People’s Republic of China fast-track innovation by
breaking the process into small components with multiple teams. These studies often
focus on innovation outcomes as a function of strategic decisions by top management.
Thus, the management innovation literature needs a measure of innovation that
incorporates both inputs and outputs of technological progress.
2.4 Financing Innovation

Accounting and finance research focuses on motivating and financing corporate
innovation. Dutta and Fan (2012) show that delegated R&D investment decision
making improves innovation output. Manso (2011) highlights that managerial
compensation contracts that allow for early failure and reward long-term success
facilitate the innovation process. Cheng (2004) describes how boards design
compensation contracts to minimize opportunistic reductions in R&D spending by
corporate managers. Others focus on financing’s potential to influence the creation of
new technological capabilities and products. Lerner, Sorensen, and Stromberg (2011)
report that LBO transactions do not appear to influence the innovation activities within
the firm. Fang, Tian, and Tice (2014) argue that greater stock market liquidity causes
a reduction in future innovation. Without the threat of a takeover, managers arguably
reduce their investments in research and development (Atanassov 2013). In contrast,
Cornaggia et al. (2015) argue that banking competition reduces corporate innovation
by public companies. In sum, the finance innovation literature appears to require a
measure of innovation that captures inputs and observable outputs to investors.
5
3. INNOVATION DATA
3.1 R&D Expenditures
R&D spending is arguably the most widely used measure of corporate innovation
activity. This mandatory disclosure provides a quarterly update on how much the firm
invested in R&D over the prior 3 months. Prior to 1974 firms often treated R&D
much like capital expenditures, placing this investment on the balance sheet (Lev and
Sougiannis 1996). However, accounting policy makers argued that R&D, while focusing
on generating future cash flows, is intangible and difficult to verify. Consequently,
corporate R&D began to be expensed under Generally Accepted Accounting
Principles. These disclosure regulations, SFAS 2 Accounting for Research and
Development Costs, require managers to report material R&D expenditures, providing
clear guidelines for managerial compliance. Specifically, managers must report all
relevant information and exclude trivial information to limit concerns about shrouding
(Koh, Reeb, and Zhao 2016). The Supreme Court provided guidance in 1976, deciding
that materiality occurs when a “reasonable investor” views the information as price
sensitive (Sauer 2007).
Firms spend a considerable amount on R&D, with the largest 1,000 firms spending
over $638 billion in 2016. Bushee (1998) argues that investments in research and
development constitute an essential factor of a firm’s market value. Kothari, Laguerre,
and Leone (2002) emphasize that both capital expenditures and R&D spending deliver
measurable market value benefits to the firm. R&D spending provides a signal about
the inputs into the innovation process. For questions or concerns about motivating
corporate innovation, this is an especially intuitive signal. Research in industrial
organization often stresses the importance of analyzing inputs in a variety of settings.
Athey and Bagwell (2008) stress that observing the costs of competitors can facilitate
collusion within an industry. Similarly, the competitive strategy emphasizes that
monitoring input prices allows investors and firms to assess the threats that current
and future rivals pose (Miller and Waller 2003). While competitive intelligence may
provide a range of R&D estimates about competitors, reported R&D provides a concise
point estimate.
One of the primary advantages of using R&D spending as a signal of innovation activity
is that a substantial number of firms provide this mandatory disclosure. For instance,
almost two-thirds of S&P 1500 firms report R&D. However, this signal of innovation
contains two sources of noise. First, managers must determine which activities
constitute research and development. An engineer who expends part of her effort on
innovation and part on quality control requires some system or decision about the
relative allocation of her time. Thus, the decision about classifying specific spending
as R&D outside of subunits that specialize in R&D is a discretionary choice of
the manager (Horwitz and Kolodny 1980). Sougiannis (1994) describes the second
concern for investors: does R&D spending today reflect benefits from prior R&D
investments? In other words, does the past success of R&D translate into future
success of today’s R&D spending? Any signal about innovation inputs provides a noisy
signal about future payoffs from this investment. Thus, R&D spending contains two
well-known sources of noise, one stemming from managerial discretion in categorizing
R&D and the other from the potential disconnection between innovation inputs and
innovation outputs.
6
The greatest challenge in using R&D to capture corporate innovation, however, arises
from the large number of firms that do not report any information about R&D. R&D
frequently requires specialized machinery, buildings, employees, computers and
assorted other items, suggesting that the classification decision for these expenses can
be subjective. Koh and Reeb (2015) argue that managers exploit this discretion not to
report R&D, resulting in a large number of firms with blank or missing R&D data.
Conceivably, managers consider the strategic response of their competitors to their
reported R&D and therefore choose not to report it to gain a competitive advantage
over them (Scotchmer 1991). This potential failure to report R&D spending creates a
selection bias in using R&D expenditures as a measure of innovation activity.
To assess whether firms without reported R&D actually do engage in innovation
activities, Koh and Reeb (2015) undertake a series of tests. Their first set of tests
compares patent activity between the firms that do not report R&D and those that
report zero R&D, finding that non-reporting R&D firms file 14 times more patents than
their zero R&D counterparts. Koh and Reeb (2015) also document significant
differences in patent characteristics between zero and non-reporting R&D firms. Using
patent citation data, they find that non-reporting R&D firms, relative to zero R&D firms,
receive more influential patents that take longer for competitors to discover. Perhaps a
more informative comparison is to compare patents in firms without reported R&D with
patents in firms that do report R&D. Koh and Reeb (2015) discover that the top 5% of
positive R&D firms receive substantially more patents than firms with missing R&D, yet
they also find that missing R&D firms with patents correspond to the bottom 90th
percentile of the patenting profiles of positive R&D firms. To investigate whether this
reporting choice is deliberate, Koh and Reeb (2015) exploit the forced change in
auditors following the collapse of Arthur Anderson. They find that this exogenous
change in auditors led firms to begin reporting R&D, resulting in these firms disclosing
R&D at the around the 26th percentile of all positive R&D firms.
Extending this line of research, Koh, Reeb, and Zhao (2016) assess whether this
missing R&D is material. Focusing on firms that switch from missing to positive R&D,
they find that these firms often report comparative figures for prior years (years that
were previously blank). Koh, Reeb, and Zhao (2016) document that the unreported
R&D numbers correspond to about the 55th percentile of their positive R&D peers.
Using stock returns, they also find that missing R&D firms co-move nearly 1,300%
more with positive R&D firms than with zero R&D firms. Turning to financial analysts,
Koh, Reeb, and Zhao (2016) report that analysts following missing R&D firms are over
five times more likely to cover a positive R&D firm than zero R&D firms. Moreover, they
report that an exogenous loss in analysts’ coverage leads missing R&D firms to begin
reporting R&D. Perhaps more importantly, Koh, Reeb, and Zhao (2016) document
that opportunistic insider trading in missing R&D firms allows managers to obtain
substantial rents relative to those found in positive R&D firms. Thus, withholding R&D
expenditures allows insiders of missing R&D firms to obtain private benefits. In short,
firms without reported R&D often engage in substantial R&D activities with the opacity
from not reporting, benefiting the managers of the firm.
Empirical innovation research typically handles the missing R&D problem in one of
two ways, either classifying these firms as zero R&D or discarding these observations.
If these missing observations simply result from noise, then either of these approaches
has potential merit (Koh et al. 2015). In contrast, if missing R&D represents a selection
bias or disclosure choice of the managers, it could influence the results of tests that
use it as a measure of innovation. Koh et al. (2015) investigate how these treatments
potentially influence research on corporate innovation. They focus on the research in
finance and management that explores how managerial overconfidence influences
7
corporate innovation. Specifically, Galasso and Simcoe (2011) and Hirshleifer et al.
(2012) document that overconfident managers, relative to their cautious peers, lead
their firms to participate in greater corporate innovation. Koh, Reeb, and Zhao (2017)
show that cautious CEOs are substantially less likely to report their R&D or seek
patents for their corporate innovation due to their concerns that their competitors will
learn from this information. Since cautious CEOs show a lower tendency to report their
R&D expenditures or file patents, the standard approach of excluding missing R&D
firms will lead to biased results. After adjusting for the propensity to disclose corporate
innovation, Koh et al.’s (2017) results suggest that cautious CEO firms are the ones
that engage in more corporate innovation.
R&D spending is an important, mandatory signal about corporate innovation. However,
it suffers from concerns about the noise in the signal and about managers’ potential
selection bias. The noise in this signal arises because it only captures inputs (rather
than outputs) and because of the managers’ discretion in classifying R&D activities.
The bias in this signal stems from the managers’ choice to report or not to report the
R&D, depending on their own assessment of materiality. Recent literature shows that
non-reporting R&D firms often have substantial R&D expenditures. Consequently,
strictly relying on R&D spending to measure or capture the innovation activities of a
firm can lead to biased and misleading results.
3.2 Patents
Recognizing these problems with R&D spending data, empirical research on corporate
innovation often relies on patent data. Griliches (1981), focusing on market value,
argues that patents provide a good measure of the intangible capital or innovation
capabilities of a firm. Jaffe (1986) reports that patents capture the technological
position of the firm. Predicated on the notion that patents provide an indicator of
successful innovation, Pakes and Griliches (1984) investigate the productivity of
innovation across firms (typically researchers compute productivity as patents per
R&D). More generally, researchers consider firms with large numbers of patents,
across a wide spectrum of technological classes, to contain the greatest innovation.
Patents, like academic articles, receive citations, which are also used to capture
corporate innovation. Trajtenberg (1990) advocates the use of patents, weighted by
their citations, to capture the value of a firm’s innovations. Exploiting the citation data
further, two other common measures of innovative capacity based on citations are the
originality and generality of the patent (Koh and Reeb 2015). These measures capture
the heterogeneity of the references in the patent and the width of the future citations
across technological spaces (Hall, Jaffe, and Trajtenberg 2001). Conceptually, these
measures build on the notion that patents and their citations represent an output
measure of successful, corporate innovation.
Developing new technologies and knowledge capacity inherently involves spillovers
and mimicking activity. One major stream of literature on innovation highlights the role
of a firm’s ability to protect its knowledge generation. Several firms can use knowledge
simultaneously, which provides benefits to the firms and improves the stock of general
knowledge capital in the region (Romer 1990). To limit these spillovers, firms rely on
both trade secrets and patents. Focusing on the mix of patents and trade secrets,
Mansfield (1986) reports that pharmaceutical firms view patents as an important tool,
while most other firms use them less than 20% of the time. Nevertheless, patents
provide an observable and measurable approach for firms to protect their innovation.
Moser (2005) finds that most inventions remain unpatented, but those that are patented
are some of the most successful. Early literature in the field of economics highlights
8
several benefits of using patents to measure innovation. For instance, patents and
their citations allow the researcher to investigate the number of successes, the depth of
the patent citations, the spread of the citations across other industries and the reliance
of the innovation on prior research in the same technological class (Jaffe 1986;
Trajtenberg 1990). Other studies emphasize how patents and their citations provide
an indicator of technological output and knowledge generation (Hall, Jaffe, and
Trajtenberg 2005).
One primary challenge in using patents or patent citations to measure innovation is
their relative scarcity. While R&D data may only be available for 58% of the Compustat
universe of firms, patents are even scarcer. Among the Compustat-listed firms, only
23% file patent applications (Koh et al. 2015), suggesting that the non-patenting R&D
firms are failed innovators (Reeb and Zhao 2017). Business research typically treats
these as zero patent firms, while economists simply tend to exclude them from the
analysis. Recent studies emphasize the common view that patents capture innovative
success, documenting that stock market liquidity, institutional ownership and
acquisition activity cause significant reductions in corporate innovation (Atanassov
2013; Fang, Tian, and Tice 2014; Bernstein 2015).
Firms potentially weigh the cost of filing a patent against the expected value of the
property rights protection (Evenson 1984). Of course patents also provide detailed
plans and schematics about the underlying technological innovation, which may benefit
competitors, yet the typical interpretation of this trade-off is that non-patenting firms
possess unimportant innovations. Moser (2012) provides empirical support for this
interpretation based on data from the world’s fairs from 1851 to 1915. She documents
that these inventors more frequently patented the inventions that ultimately proved to
be more valuable. Using a survey of Canadian firms, Hanel (2008) discovers that
patent-reliant firms exhibit higher profits than their non-patenting peers. Others indicate
that non-patenting firms, based on subsequent sales revenues, suffer from inefficient
R&D investments (Hussinger 2006). More generally, the cross-disciplinary literature
on patents and their citations depicts non-patenting R&D firms as failures. Building on
this notion, the previously noted approach to capturing R&D productivity relies on
scaling R&D by patents, which explicitly depicts non-patenting R&D firms as innovation
failures.
In contrast, other researchers highlight that patents help firms to secure their property
rights and that successful innovation is often kept as trade secrets. Cohen, Nelson, and
Walsh (2000) argue that firms appear to garner patents for strategic reasons instead
of simply seeking to secure their property rights. More specifically, these authors
suggest that firms seek patents to block competitors or as a way of advertising their
technological prowess. Firms have a choice between patenting their successful
innovations and keeping them as trade secrets (Horstmann, MacDonald, and Slivinski
1985). Survey evidence, across different time periods and locations, suggests that
firms use trade secrets more commonly than patents to protect successful innovation
(Levin et al. 1987; Cohen, Nelson, and Walsh 2000). A 2008 survey reveals that firms
choose to use trade secrets instead of seeking patent protection roughly two-thirds
of the time. Highlighting the value of trade secrets, Younge and Marx (2012) report
that non-compete agreement changes provide evidence of their widespread appeal
and value.
To investigate this issue, Reeb and Zhao (2017) report that both patenting and
non-patenting firms introduce valuable new products. Their analysis suggests that
innovation in patent-seeking firms focuses on product efficiency while non-patenting
R&D firms tend to concentrate on cost-efficiency improvements. Interestingly, Reeb
and Zhao (2017) document that executives in non-patenting R&D firms participate in
9
substantial opportunistic insider trading, which account for the findings in prior studies
of excess insider trading in innovative firms. They suggest that their results indicate
that findings that stock market liquidity, institutional ownership and acquisition activity
influence corporate innovation are misleading. Instead, Reeb and Zhao (2017)
document that intermediary coverage or firm characteristics influence the choice
between patenting and keeping an innovation a trade secret. Citations depend on the
firm filing a patent; therefore, they suffer from the same problems as patent counts
(Lerner and Seru 2015).
Patents and their citations are important signals about corporate innovation.
Nonetheless, they suffer from concerns about the noise in the signal and about
managers’ potential selection bias. The noise in these signals arises because patent
counts and their citations only provide a rough proxy for the innovation occurring in the
firm. Patents are a voluntary disclosure of the firm that potentially secures property
rights. However, these signals of corporate innovation, patents and citations, are
available for a much smaller segment of the population than R&D spending. The bias in
these signals stems from the managers’ choice to seek or not to seek to patent an
innovation, depending on their own assessment of the firm’s ability to keep the
information a trade secret. Recent literature shows that non-patenting firms often make
substantial innovations in cost efficiency relative to product quality. Consequently,
strictly relying on patents or their citations to measure the innovation activities of a firm
can lead to biased and misleading results. Turning to R&D productivity, this formulation
relies on scaling one noisy, biased measure of innovation by another noisy, biased
measure of innovation.
3.3 New Product Announcements

Another, less common, approach to measuring innovation centers on new product
announcements. New product announcements provide a potential signal about the
output of the firm’s R&D. Eddy and Saunders (1980) argue that new products capture
the output of technological innovation and investigate stock market reactions to new
product announcements. Consistent with this output perspective, Chaney, Divenny,
and Winer (1991) document substantive increases in firm value with new product
announcements for technological firms. Lee et al. (2000) find that new product
announcements are especially valuable for first movers but less so for slow movers.
Importantly, first movers suffer when late entrants imitate their products, suggesting
that the magnitude of the market reaction provides a potential gauge of innovation
success. New products tend to increase firm profits and size but are also associated
with decreases in company advertising (Bayus, Erickson, and Jacobson 2003).
Finance research suggests that increased taxes lead to fewer new product
announcements (Mukherjee, Singh, and Zaldokas 2017). More generally, researchers
consider new product announcements to capture the successful innovation of the
firm, with stock market reactions providing a gauge of the relative magnitude of
the innovation.
New product announcements span small and large firms, with roughly 18% of the
Compustat universe announcing new products (Reeb and Zhao 2017). Katila (2002)
emphasizes that new product announcements provide a multi-layered measure of the
innovativeness of the firm. However, new product announcements’ effects differ across
industries and firms (Chaney and Devinney 1992). Marketing research emphasizes that
the industry characteristics and the firm’s prelaunch activities explain some of the
heterogeneity in the market reactions to new product announcements (Wind and
Mahajan 1987). While new products capture the heart of what many consider to be the
10
outcome of corporate innovation activity (e.g. classic stories like Kodak and Xerox or
more recent ones like Tesla and Facebook), they only provide a signal regarding the
product side of corporate innovation. Thus, new product announcements provide a
noisy signal of corporate innovation, failing to capture cost-reducing innovations.
New product announcements are a voluntary disclosure of firms. Management
research indicates that the incentives to make new product announcements differ
depending on the types of goods and customers (Porter 1976). Bayus, Erickson, and
Jacobson (2003) suggest that firms trade off advertising and new product
announcements. Firms with multiple new products in a given period face a choice
between multiple new product announcements and greater paid advertising;
they depend on their assessment of the greatest impact on potential customers
(Hendricks and Singhal 1997; Rabino and Moore 1989). Thus, both firm and industry
characteristics potentially influence the choice between advertising and new product
announcements, suggesting selection issues in new product announcements.
New product announcements are an important, voluntary signal about corporate
innovation. However, they also suffer from concerns about the noise in the signal and
about managers’ potential selection bias. The noise in this signal arises because it only
captures product outputs (rather than cost outputs) and because of the difficulty in
assessing the value of such announcements. The bias in this signal stems from the
managers’ choice regarding whether or not to announce new products, depending on
their own assessment of paid advertising for this market or product. Consequently,
strictly relying on new product announcements to measure the innovation activities of a
firm can lead to biased and misleading results.
4. EVALUATING MULTIPLE SIGNALS

4.1 Composite Measures
Each of the various signals discussed above regarding corporate innovation provides a
noisy and potentially biased indication of firms’ innovation activities. Holmstrom (1989)
argues that omitting informative signals about an activity eliminates useful information,
making it difficult to assess performance. While his arguments are embedded in a
model of incentivizing agents, the basic intuition for creating a measure of innovation is
similar. The firm chooses an innovation strategy that has an impact on its performance,
which managers may not fully reveal for both strategic and agency reasons.
Shareholders and empirical researchers observe these noisy signals of the firm’s
innovation strategy. Business and economics research usually focuses on one of these
signals and forms an opinion about the innovative capabilities of the firm. Marketing
scholars suggest that all of the observable signals that contain information about a
construct should be included in the measure (e.g., Gerbing and Anderson 1988).
Psychometrics research has dealt with noisy measures about an underlying issue that
is coupled with selection bias for decades (Cronbach and Meehl 1955). The concern
arose from seeking to incorporate multiple responses into a questionnaire to form a
single measure (or construct as these researchers often call it) about a specific
underlying issue or viewpoint in a subject (Nunnally 1978). Researchers face two
problems in trying to compare individuals’ beliefs or values with those of the general
population. First, they usually use multiple questions to solicit signals about the
subjects’ underlying issue or viewpoint on a particular topic (Cautin and Lillenfeld
2015). Consequently, they have multiple, noisy signals that potentially capture an
11
underlying issue. Second, they also face a selection bias, because respondents often
leave some of the questions blank (Kline 2015).
To develop a single scale or measure, with noisy signals that are also influenced by
selection bias, clinical psychology research uses item‒total correlations or a factor
analysis approach to identify the set of signals to incorporate into the measure. Starting
with Nunnally (1978), the item‒total correlation approach investigates how well each of
the signals correlates with the others in assessing corporate innovation. Specifically,
this approach starts with several potential signals about corporate innovation and
identifies the set of signals that best correlates with the total score of the item pool
(Sullivan 1994). Although there is no complete solution to this problem, it is possible to
assess the relative merits of combining several noisy, biased signals in a single
measure of corporate innovation.
The sample for this analysis is based on the BCG/BusinessWeek senior executive
survey regarding the most innovative companies. Starting in 2005, the
BCG/BusinessWeek survey was conducted by the Boston Consulting Group, a
management-consulting firm with offices in 37 countries when it began this initiative.
The survey was conducted in conjunction with BusinessWeek, a weekly business
magazine that began operations in 1929 and was purchased by Bloomberg in 2009.
The first senior executive survey contained a total of 940 executives from 68 countries,
ranking Apple, 3M, GE, Microsoft and Sony as the five most innovative firms
(BCG Survey 2005). Overall, the survey names the most innovative companies each
year, naming 10 for a few years and expanding the number to 50 in 2007. The survey
respondents are primarily CEOs, presidents, strategy executives and brand managers.
By 2010 the survey included 1,590 senior executives, over half of whom were listed as
holding C-Suite positions (BGS Survey 2010). The top five firms in 2004 were Apple,
Google, Microsoft, IBM and Toyota. Both the business press and the mainstream
press generate substantial coverage of these annual reports, including Reuters, The
Telegraph in the UK and The Sydney Morning Herald.
Interestingly, several firms, considered as some of the most innovative, exhibit missing
or limited information in several of the commonly used signals of innovation. For
instance, the 2010 list includes Coca-Cola, a firm that does not report R&D in its
financial statements but has six R&D centers around the world. Wal-Mart is also listed
in the 2010 report and is often described as an innovative leader in supply chain
management because of its use of communication and computer technologies (Brunn
2006). However, Wal-Mart only averaged two patents per year from 2005 to 2009. Of
course, firms like Pfizer, Boeing and Honda also appear on the lists, and these firms
typically have substantial amounts of R&D, patents and new product announcements.
The respondents to the BCG/BusinessWeek survey report that they measure
innovation internally, focusing on profits, idea generation and revenue growth.
Consequently, in addition to the measures noted above (R&D, patents, patent citations
and new product announcements), I consider four firm characteristics in the analysis.
Specifically, I incorporate revenue, revenue growth, advertising and profitability as
potential signals of corporate innovation (BCG Survey 2005; Reeb and Zhao 2017).
I limit the sample to public firms listed in the years 2007‒2010 to capture patents and
their citations. This gives a sample of 63 highly innovative firms during the period 2007
to 2010. The sample comprises large firms from several different countries and
industries. I gather financial data from Global Compustat, patent data and citations
from Noah Stoffman’s website and new product announcements from Reeb and Zhao
(2017). Patents are a simple, annual count measure of patent applications, while
citations capture the number of citations that they receive in a given year. I combine
these items to capture citation-weighted patents, which I compute as the product
12
of patents plus one and citations plus one. New products are the number of
announcements about the firm’s new products. R&D spending and advertising are
simply the reported R&D and advertising, scaled by total assets. Sales and sales
growth are also based on annual income statement data. I use the log of sales.
My analysis starts with all seven potential signals (sales, sales growth, weighted
patents, new products, R&D, advertising and profits). Table 1 provides descriptive
statistics and correlations for these seven variables. Weighted patents, profits, R&D
and new product announcements are all highly correlated with each other. For
instance, the correlation between weighted patents and new product announcements is
.32. This preliminary analysis suggests that the three traditional signals of innovation
(R&D, weighted patents and new product announcements) and firm profitability provide
potential signals about corporate innovation.
Table 1: Descriptive Statistics

Products Growth Advertising Profits WPatents R&D Sales
Products 1
Growth .04 1
Advertising .00 .02 1
Profits .09 .34 .14 1
WPatents .32 .05 .06 .22 1
R&D .13 .15 –.01 .33 .59 1
Sales .20 .03 –.13 –.11 .18 –.03 1
Item‒total correlation analysis tests the reliability of each of the items against the set of
the other items. I proceed in stages to identify the appropriate set of signals to include
in a composite measure of corporate innovation. In the first stage, I apply item‒total
correlation analysis to all seven potential signals, which shows that these seven items
do not appear to capture the same concept. Consequently, I drop the variable with the
lowest item‒rest correlation. In the first iteration, I drop advertising, as it has the lowest
item‒rest correlation (score of .002). I repeat this analysis and drop growth (item‒rest
correlation = .044) and profits (item‒rest correlation = .099). The remaining signals
(R&D, weighted patents, new products and sales) provide the final set of variables for a
combined measure of corporate innovation.
Based on these results, I construct two composite measures of corporate innovation
using factor analysis and index-ranking analysis to reduce these four signals of
innovation into a single measure. Factor analysis requires the signals to be highly
correlated and load on a single factor (Duru and Reeb 2002). This dovetails nicely with
the requirement from item‒correlation analysis that the signals capture some common,
underlying construct. Table 2 provides the results of the factor analysis, showing that
these innovation signals load on a single factor that explains 75.56% of the cumulative
variance. Weighted patents and R&D spending have the highest factor loadings of
the four potential signals of innovation. Taken as a whole, these results suggest that
factor analysis with the four variables provides a potentially useful measure of
corporate innovation. Table 3 provides the average predicted factor score for each firm
in the analysis. 1
1
To create a factor score centered on 100, I scale each predicted value of factor 1 by the standard
deviation of the factor, multiply it by 10 and add 100. The Stata code for this analysis is available
on request.
13
Table 2: Factor Analysis and Measuring Innovation

Panel A: Total Variance Explained
Component Eigenvalue Proportion Cumulative
Factor 1 1.63598 .4090 .4090
Factor 2 0.16905 .0423 .4513
Factor 3 –0.17501 –.0438 .4075
Factor 4 –0.20571 –.0514 .3561
* Component extracted: only one component has an eigenvalue greater than 1.
Panel B: Factor Loadings

Variable Loading
R&D .6289
Citations .6846
Patents .3942
New Product Announcements .7851
Table 3: Composite Measures of Corporate Innovation

Factor Index Factor Index
Company Name Measure Measure Company Name Measure Measure
3M 108.7 91.8 Johnson and Johnson 105.4 114.8
Amazon 105.6 114.8 JP MorganChase 93.7 114.8
Amgen 105.7 80.3 LG 108.1 91.8
Apple 104.9 114.8 Marriott 90.1 57.4
ArcelorMittal 90.7 68.9 McDonalds 91.1 57.4
AT&T 104.4 149.2 Merck and Co 109.6 103.3
Banco Santander 88.45 34.42 Microsoft 125.3 160.7
Best Buy 91.0 80.3 Nestle 92.2 80.3
Boeing 111.9 149.2 Nike 99.4 91.8
BP 91.7 80.3 Nintendo 92.2 68.9
Caterpillar 104.0 114.75 Nokia 121.6 172.1
China Mobile 90.6 68.8 Nordstrom 89.6 45.9
Cisco 115.9 160.7 Oracle 110.8 137.7
Citigroup 91.3 103.3 Pfizer 104.5 126.2
Coca Cola 89.9 45.9 Proctor and Gamble 105.0 137.7
Daimler 100.6 103.3 Reliance Industries 89.4 34.4
Dell 104.4 114.8 Royal Dutch Shell 91.9 91.8
Duke Energy Co 89.8 34.4 Siemens 111.1 137.7
ExxonMobil 103.0 126.2 Sony 112.8 137.7
Facebook 93.2 57.4 Southwest Airlines 90.7 57.4
Fiat 92.9 80.3 Starwood Hotels 89.7 57.4
Fidelity 89.3 34.4 Target 90.1 91.8
Ford 108.9 172.1 Telefonica 91.2 91.8
GE 109.9 91.8 Toyota 97.7 126.2
continued on next page
14
Table 3 continued
Factor Index Factor Index
Company Name Measure Measure Company Name Measure Measure
Genentech 111.1 160.7 TwentyFirst Cent Fox 90.2 45.9
Goldman Sachs 95.4 80.3 Verizon 97.4 114.8
Google 106.6 103.3 Vodafone 92.7 114.8
HP 112.3 172.1 Volkswagen 94.2 103.3
Honda 113.3 126.3 Wal-Mart 94.6 126.2
HSBC 89.8 45.9 Walt Disney 91.3 68.9
Infosystems 92.6 68.8
Intel 119.5 149.2 Correlation .83
IBM 119.9 160.6
I also compute a composite measure of innovation based on an index-ranking method

and the same for four signals of corporate innovation. In contrast to factor analyses,
index ranking equally weights each of the four variables to create the composite
measure of corporate innovation. In this approach each firm in the sample is ranked
along each of the four signals of corporate innovation. The firms are ranked by quintile
along each of the four dimensions, with the firms in the bottom 20% ranked in the
lowest quintile (1) and those with the highest signals ranked in the top quintile (5). The
four scores are then added together to give the Innovation Index; the higher the score,
the more innovative the firm. Column 2 of Table 4 provides the Innovation Index score
for each firm. 2 The two composite measures of corporate innovation are correlated
at .83. Importantly, neither composite measure relies on a single signal of corporate
innovation.
Table 4: Simulation Analysis

Panel A: Mean Analysis
True Noisy Noisy–Biased Diff. 1–2 Diff. 1–3
1 2 3 4 5
Spend/R&D 10.05 10.08 12.08 –0.48 –28.69***
Success/Patents 15.28 15.33 27.41 –0.68 –114.03***
Productivity 1.58 0.70 2.45 40.26*** –36.95***
Panel B: Standard Errors

True Noisy Noisy–Biased
1 2 3
Spend/R&D 0.063 0.089 0.096
Success/Patents 0.095 0.112 0.127
Productivity 0.015 0.009 0.029
2
To create an index score centered on 100, I multiply the raw index scores by 11.475. In a larger sample,
I would use deciles or vigintiles.
15
Table 4 continued
Panel C: Multivariate Analysis

Dependent Variable = Growth
True Values Noisy Signals Noisy–Biased
Constant .704 3.940*** 1.225 4.466*** –1.914 4.912***
(0.30) (4.99) (0.80) (5.68) (–1.11) (6.31)
Firm Size .289 –0.374** 0.193 –0.370** 0.378 –0.375**
(0.84) (–2.20) (0.65) (–2.17) (1.16) (–2.23)
Spending or R&D –0.865 0.771 1.759***
(–0.51) (0.59) (10.83)
Success or Patents 4.116*** 2.512** 2.268***
(3.38) (2.43) (2.68)
Productivity 0.237*** –0.241 –0.241***
(2.72) (–1.61) (–5.63)
R-Square .02 .01 .01 .01 .11 .03
4.2 R&D Productivity

Factor analysis with different signals of innovation aims to uncover the underlying
or latent technological capacity of a firm. However, another approach to combining
innovation signals is popular in the academic literature, specifically scaling one signal
of innovation by another. Typically, this approach focuses on scaling patents by
corporate R&D expenditures, which are then labeled R&D productivity. To assess the
impact of scaling one noisy, biased signal by another such signal of innovation, I
undertake a simple simulation analysis. First, I simulate true R&D spending, true
innovation success and firm size, which are labeled spending, success and size. To
keep the simulation as simple as possible, I simulate all the values using the normal
distribution. I next create noisy signals of these individual signals, reported R&D and
patents, which have similar means to the true values but higher variances. The noise in
each signal is also normally distributed with a mean of zero.
Second, I add selection bias to each of these signals to create noisy–biased signals of
R&D and patents. I model disclosure bias as a function of firm size and a random
variable, assuming that disclosure bias decreases with firm size. The noisy–biased
signals of innovation have higher means and variances than true R&D spending and
innovation success. R&D productivity is computed using the true values, the noisy
signals and the noisy–biased signals of innovation.
Table 4 (Panel A) shows the mean R&D productivity for each of the three sets of
innovation measures. Even without selection bias, the noisy measures of R&D and
patents give a downward-biased estimate of true productivity. Panel B shows that the
variance of R&D productivity is also downward biased when computed using the noisy
signals. This finding arises across a variety of different assumptions about the nature of
the noise in the measures (changing the scope of the noise or the distribution). The
results in Table 4 stem from a 1,000-observation simulation. Figure 4 shows the mean
R&D productivity with the noisy signals based on 5,000 simulations. Without any bias in
the signals, the distribution of R&D productivity is well below the true R&D productivity.
16
Figure 4: R&D Productivity with Noisy Signals
* The vertical line at 1.58 is the true R&D productivity of the population.
Panels A and B of Table 4 also show the results of R&D productivity using the noisy–
biased signals of innovation. These estimates are consistently different from true R&D
productivity. While the nature of the selection bias influences the direction of these
results, the upward bias in R&D productivity occurs across a wide spectrum of potential
modeling choices for selection bias.
Finally, I investigate how these three versions of R&D productivity perform in a
multivariate setting. I simulate firm growth and make two assumptions about it. First, I
assume that smaller firms exhibit higher growth rates than larger firms. Second,
I assume that successful innovation leads to firm growth. Consistent with these
assumptions, the Column 1 results in Panel C of Table 4 show that innovation success
is correlated with firm growth. Column 2 shows that true R&D productivity is also
associated with firm growth.
Columns 3–4 repeat these regressions using the noisy signals of reported R&D and
patents to compute R&D productivity. The noisy signals of R&D and patents give
similar inferences to the true measures in the base regression (comparing columns 1
and 3). However, using R&D productivity with these noisy signals gives a downward-
biased coefficient estimate. R&D productivity does not appear to be a reliable measure
of corporate innovation using the noisy signals of reported R&D and patents. Columns
5–6 repeat these regressions with the noisy–biased signals of innovation. In this setting
the coefficient estimates and standard errors of R&D are biased upwards. Moreover,
the coefficient estimates and standard errors of R&D productivity are also biased.
Figure 5 shows the results of repeating this analysis 5,000 times. Rather than
mitigating the problem of noisy–biased signals of innovation, the use of R&D
productivity appears to magnify or intensify the problem.
17
Figure 5: Plot of the Beta Coefficient for R&D Productivity (Noise-Biased)
* The vertical line at 0.238 is the true beta coefficient for R&D productivity.
5. SUMMARY AND CONCLUSION

Innovative firms often exhibit stock market premiums, and both the academic and
the business press suggest that they foster economic growth. Corporate innovation
consists of several different tasks, including the development of new products and
new production methods and the improvement of existing products (Grossman and
Helpman 1990). Thus, innovation focuses on making something new, making it
cheaper or making it better. Research on corporate innovation extends across several
social science research streams, focusing on the encouragement and impact of
corporate innovation. Studies on the determinants of innovation concentrate on the
entire innovation process. However, measuring corporate innovation remains difficult,
because it covers multiple tasks and types of corporate activities.
R&D spending, patent counts, patent citations and new product announcements are
the most commonly used measures of corporate innovation. One challenge of these
measures is the large number of firms without reported R&D, patents or new product
announcements. While R&D spending is a mandatory disclosure, patents and new
product announcements are voluntary disclosures with additional benefits (acquiring
property rights and customers). R&D spending is only available for 58% of the
Compustat universe; patents are available for 23% of the firms, while new product
announcements only cover 18% of the firms (Reeb and Zhao 2017).
Firms potentially weigh the costs and benefits of disclosing innovation along each
of these lines by trading off the potential costs and benefits. Each of these signals
conveys information about corporate innovation, yet they all suffer from concerns about
the noise in the signal and about managers’ potential selection bias. The noise in these
signals arises because they only capture a single component of firms’ innovation
process, such as inputs, intermediate steps or outputs. The bias in these signals stems
from the managerial disclosure choices. Consequently, strictly relying on new product
announcements to measure the innovation activities of a firm can lead to biased and
misleading results.
18
Excluding informative signals about innovation discards useful information, making it

difficult to measure the activity (Holmstrom 1989). Firms decide on their investments in
innovation and, for both strategic and agency reasons, may choose to keep this
information private. The econometrician observes several noisy and biased signals
of the firm’s innovation strategy. The ability to measure innovation, across a wide
spectrum of activities and firm types, is a salient concern for research that seeks to
understand and evaluate corporate innovation. My results imply that research on
corporate innovation that relies on a single signal of innovation potentially leads to
misleading conclusions. Instead, researchers should rely on multiple signals of
corporate innovation to measure it. Using item–total analysis to identify common
signals of innovation, I find four signals that are potentially important. I then develop
both factor analysis and index-based composite measures of corporate innovation,
relying on R&D, patent citations, new product announcements and profitability. Both
approaches provide a composite measure of corporate innovation that incorporates
multiple signals of innovation activity.
Using a composite measure of corporate innovation does not completely mitigate the
concerns about measure noise. However, using a composite measure potentially
reduces some of the concerns in using a single noisy, biased signal of corporate
innovation. The factor analysis approach to measuring innovation in this study attempts
to limit the impact of noise and selection bias by concentrating on the shared
components of each signal that only captures one dimension of the firm’s technological
capacity. In contrast, the common practice of scaling one noisy and biased signal of
innovation (such as patents) by another noisy, biased signal of corporate innovation
(R&D) exacerbates the potential problems with these signals. Specifically, scaling one
noisy signal by another noisy signal is likely to magnify the noise-to-signal ratio and
lead to biased empirical results. Once selection bias is introduced into the signals, R&D
productivity performs even more poorly than before.
The measurement of corporate innovation also affects policy makers. Both the
academic and the popular press commonly reference corporate innovation as an
important driver of national prosperity. Politicians often focus on inputs such as
research and development expenditures, while organizations like the World Bank often
discuss R&D productivity as they seek to capture returns to dollars spent on innovation.
Others use this type of data to argue for policy changes, noting that companies in one
market spend more on R&D than companies in another market. However, these
comparisons are difficult to assess. The noise of these signals differs across markets,
suggesting that such comparisons capture both differences in signal noise and
differences in true innovation capacity. The nature of the noise in measures of
innovation often stems from the type of innovation undertaken by the firms in that
economy. Companies involved in process innovation are more willing to report their
R&D expenditures, while companies engaged in new product development often seek
to minimize information about their research and development outputs. Consequently,
policy makers that rely on a single signal of corporate innovation to evaluate corporate
innovation run the risk of making decisions based on biased inputs, suggesting
that they should also seek to incorporate multiple signals of corporate innovation into
their analyses.
19
Focusing strictly on patents to measure innovation in a cross-market setting is

especially problematic, as firms differ in their use of trade secrets and patents to
protect successful innovation. The ownership structure of the firm, the legal
environment, the distance from competitors and a host of market characteristics
potentially influence firms’ decisions to patent their corporate activities. For instance, on
the stock exchanges of New York and Taipei,China, roughly 19% of firms with R&D
expenditures seek patents while the other 81% appear to rely more on trade secrets. In
contrast, fewer than 4% of positive R&D firms in the stock exchange of the Republic of
Korea seek patents. In fact, in the stock exchange of the Republic of Korea, firms
without reported R&D often received more patents in a given year than firms with R&D
expenditures. Similarly, the decision to report R&D expenditures appears to differ
across markets in Asia. Relying on a single, noisy signal of corporate innovation to
capture innovative capacity in a firm or market can therefore provide a distorted view of
progress. These distortions can be especially pronounced when attempting to compare
companies located in different economies. Assessing innovation in Asia and comparing
it with innovation in western economies will be strongly influenced by the metric chosen
to capture innovation. Ideally, we should use multiple signals of innovation to measure
the degree of innovation in companies and markets across Asia.
20
REFERENCES
Acemoglu, D. 2010. “When Does Labor Scarcity Encourage Innovation?” Journal of
Political Economy 118: 1037–1078.
Aghion, P., and J. Tirole. 1994. “The Management of Innovation.” Quarterly Journal of
Economics 109: 1185–1209.
Aghion, P., J. Van Reenen, and L. Zingales. 2013. “Innovation and Institutional
Ownership.” American Economic Review 103: 277–304.
Anderson, R., D. Reeb, and W. Zhao. 2017. “Family firms and Innovation Diversity.”
Working Paper, National University of Singapore.
Asheim, B., and L. Coenen. 2006. “Contextualizing Regional Innovation Systems in a
Global izing Learning Economy: On Knowledge Bases and Institutional
Frameworks.” Journal of Technology Transfer 31: 163–173.
Atanassov, J. 2013. “Do Hostile Takeovers Stifle Innovation?” Journal of Finance 68:
1097–1131.
Athey, S., and K. Bagwell. 2008. “Collusion with Persistent Cost Shocks.”
Econometrica 76: 493–540.
Audretsch, D., and M. Feldman. 1996. “R&D Spillovers and the Geography of
Innovation and Production.” American Economic Review 86: 630–640.
Bayus, B., G. Erickson, and R. Jacobson. 2003. “The Financial Rewards of New
Product Introductions in the Personal Computer Industry.” Management
Science 49: 197–210.
BCG Surveys. 2005-2017. “The Most Innovative Companies.” Boston Consulting
Group. Available https://www.bcgperspectives.com/content/interactive/
innovation_growth_most_innovative_companies_interactive_guide/
Bernstein, Shai. 2015. “Does Going Public Affect Innovation?” Journal of Finance 70,
No. 4: 1365–1403.
Boudreau, K., N. Lacetera, and K. Lakhani. 2011. “Incentives and Problem Uncertainty
in Innovation Contests: An Empirical Analysis.” Management Science 57:
843–863.
Brunn, S. 2006. Wal-Mart the World: The World’s Biggest Corporation in the Global
Economy. New York: Routledge.
Bushee, B. 1998. “The Influence of Institutional Investors on Myopic R&D Investment
Behavior.” Accounting Review 73: 305–333.
Cantwell, J. 2014. “Revisiting International Business Theory: A Capabilities-Based
Theory of the MNE.” Journal of International Business Studies 45: 1–7.
Cautin, R., and S. Lillenfeld. 2015. Encyclopedia of Clinical Psychology Set, Volume 1.
Malden, MA: John Wiley & Sons.
Chaney, P., and T. Devinney. 1992. “New Product Innovations and Stock Price
Performance.” Journal of Business Finance and Accounting 19: 677–695.
Chaney, P., T. Devinney, and R. Winer. 1991. “The Impact of New Product
Introductions on the Market Value of Firms.” Journal of Business 64: 573–610.
Cheng, S. 2004. “R&D Expenditures and CEO Compensation.” Accounting Review 79:
305–328.
21
Cohen, W., and D. Levinthal. 1990. “Absorptive Capacity: A New Perspective on

Learning and Innovation.” Administrative Science Quarterly 35: 128–152.
Cohen, W., R. Nelson, and J. Walsh. 2000. “Protecting their Intellectual Assets:
Appropriability Conditions and Why US Manufacturing Firms Patent (or Not).”
NBER Working Paper 78552.
Cornaggia, J., Y. Mao, X. Tian, and B. Wolfe. 2015. “Does Banking Affect Innovation?”
Journal of Financial Economics 115: 189–209.
Cronbach, L., and P. Meehl. 1955. “Construct Validity in Psychological Tests.”
Psychological Bulletin 52: 281–302.
DiMasi, J., R. Hansen, and H. Grabowski. 2003. “The Price of Innovation: New
Estimate of Drug Development Costs.” Journal of Health Economics 22:
151–185.
Duranton, G., and D. Puga. 2001. “Nursery Cities: Urban Diversity, Process Innovation,
and the Life Cycle of Products.” American Economic Review 91: 1454–1477.
Duru, A., and D. Reeb. 2002. “International Diversification and Analyst Forecast
Accuracy and Bias.” Accounting Review 77: 415–433.
Dutta, S., and Q. Fan. 2012. “Incentives for Innovation and Centralized versus
Delegated Capital Budgeting.” Journal of Accounting and Economics 53:
592–611.
Eddy, A., and G. Saunders. 1980. “New Product Announcements and Stock Prices.”
Decision Sciences 11: 90–97.
Eisenhardt, K., and B. Tabrizi. 1995. “Accelerating Adaptive Processes: Product
Innovation in the Global Computer Industry.” Administrative Science Quarterly
40: 84–110.
Evenson, Robert E. 1984. “International Invention: Implications for Technology Market
Analysis.” In R&D, Patents, and Productivity, edited by Z. Griliches, 89–123.
Chicago: University of Chicago Press.
Fang, Vivian W., Xuan Tian, and Sheri Tice. 2014. “Does Stock Liquidity Enhance or
Impede Firm Innovation?” Journal of Finance 69: 2085–2125.
Franco, L. 1989. “Global Corporate Competition: Who’s Winning, Who’s Losing, and
the R&D Factor Seems as One Reason Why.” Strategic Management Journal
10: 449–474.
Galasso, A., and T. Simcoe. 2011. “CEO Overconfidence and Innovation.”
Management Science 57: 1469–1484.
Gerbing, D., and J. Anderson. 1988. “An Updated Paradigm for Scale Development
Incorporating Unidimensionality and its Assessment.” Journal of Marketing
Research 25: 186–192.
Grossman, G., and E. Helpman. 1990. “Trade, Innovation, and Growth.” American
Economic Review 80: 86–91.
Griffith, R., S. Redding, and J. Van Reenen. 2004. “Mapping the Two Faces of R&D:
Productivity Growth in a Panel of OECD Industries.” Review of Economics and
Statistics 86: 883–895.
Griliches, Z. 1981. “Market Value, R&D, and Patents.” Economic Letters 7: 183–187.
22
Hall, B., A. Jaffe, and M. Trajtenberg. 2001. “The NBER Patent Citation Data File:
Lessons, Insights, and Methodological Tools.” NBER Working Paper.
Hall, B., A. Jaffe, and M. Trajtenberg. 2005. “Market Value and Patent Citations: A First
Look.” Rand Journal of Economics 36: 16–38.
Hall, B., and J. Lerner. 2010. “The Financing of R&D and Innovation.” Handbook of the
Economics of Innovation 1: 609–639.
Hanel, P. 2008. “The Use of Intellectual Property Rights and Innovation by
Manufacturing Firms in Canada.” Economics of Innovation and New Technology
17: 285–309.
Hendricks, K., and V. Singhal. 1997. “Delays in New Product Introductions and the
Market Value of the Firm: The Consequences of Being Late to Market.”
Management Science 43: 422–436.
Hicks, J. 1932. The Theory of Wages. London: Macmillan.
Hirshleifer, D., A. Low, and S. H. Teoh. 2012. “Are Overconfident CEOs Better
Innovators?” Journal of Finance 67: 1457–1498.
Holmstrom, B. 1989. “Agency Costs and Innovation.” Journal of Economic Behavior
and Organization 12: 305–327.
Horstmann, Ignatius, Glenn M. MacDonald, and Alan Slivinski. 1985. “Patents as
Information Transfer Mechanisms: To Patent or (Maybe) Not To Patent.”
Journal of Political Economy 93: 837–858.
Horwitz, B., and R. Kolodny. 1980. “The Economic Effects of Involuntary Uniformity in
the Financial Reporting of R&D Expenditures.” Journal of Accounting Research
18: 38–74.
Hussinger, K. 2006. “Is Silence Golden? Patents versus Secrecy at the Firm Level.”
Economics of Innovation and New Technology 15: 735–752.
Jaffe, Adam B. 1986. “Technological Opportunity and Spillovers of R&D: Evidence from
Firms’ Patents, Profits, and Market Values.” American Economic Review 76:
984–1001.
Katila, R. 2002. “New Product Search over Time: Past Ideas in their Prime?” Academy
of Management Journal 45: 995–1010.
Kline, P. 2015. A Handbook of Test Construction: Introduction to Psychometric Design.
New York: Routledge.
Koh, P., and D. Reeb. 2015. “Missing R&D.” Journal of Accounting & Economics 60:
73–94.
Koh, P.-S., D. Reeb, E. Sojli, and W. Tham. 2015. “Measuring Innovation around the
World.” Working Paper, National University of Singapore.
Koh, P., D. Reeb, and W. Zhao. 2016. “Is Missing R&D Material?” Working Paper,
National University of Singapore.
———. 2017. “CEO Confidence and Unreported R&D.” Management Science.
Forthcoming.
Kothari, S., Laguerre, T., and Leone, A. 2002. “Capitalization versus Expensing:
Evidence on the Uncertainty of Future Earnings from Capital Expenditures and
R&D Outlays.” Review of Accounting Studies 7: 355–382.
23
Lee, H., K. Smith, C. Grimm, and A. Schomburg. 2000. “Timing, Order and Durability of
New Product Advantages with Imitation.” Strategic Management Journal 21:
23–30.
Lerner, Josh, and Amit Seru. 2015. “The Use and Abuse of Patent Data: Issues for
Corporate Finance and Beyond.” Working Paper.
Lerner, J., M. Sorensen, and P. Stromberg. 2011. “Private Equity and Long Run
Performance: The Case of Innovation.” Journal of Finance 66: 445–477.
Lev, B., and T. Sougiannis. 1996. “The Capitalization, Amortization, and Value-
Relevance of R&D.” Journal of Accounting and Economics 21: 107–138.
Levin, R., A. Klevorick, R. Nelson, S. Winter, R. Gilbert, and Z. Griliches. 1987.
“Appropriating the Returns from Industrial Research and Development.”
Brookings Papers on Economic Activity 1987: 783–831.
Mansfield, E. 1986. “Patents and Innovation: An Empirical Study.” Management
Science 32: 173–181.
Manso, G. 2011. “Motivating Innovation.” Journal of Finance 66: 1823–1860.
Marshall, A. 1890. Principles of Economics: An Introductory Volume. New York:
MacMillan.
Miller, K., and H. Waller. 2003. “Scenarios, Real Options, and Integrated Risk
Management.” Long Range Planning 36: 93–107.
Moser, P. 2005. “How Patents Influence Innovation? Evidence from Nineteenth-
Century World’s Fairs.” American Economic Review 95: 1214–1236.
Moser, Petra. 2012. “Innovation without Patents: Evidence from World’s Fairs.” Journal
of Law & Economics 55: 43–74.
Mukherjee, A., M. Singh, and A. Zaldokas. 2017. “Do Corporate Taxes Hinder
Innovation?” Journal of Financial Economics 124: 195–221.
Nicholas, T. 2008. “Does Innovation Cause Stock Market Runups? Evidence from the
Great Crash.” American Economic Review 98: 1370–1396.
Nunnally, J. 1978. Psychometric Methods. McGraw Hill Publishing, New York, NY.
Okamuro, H. 2007. “Determinants of Successful R&D Cooperation in Japanese Small
Businesses: The Impact of Organizational and Contractual Characteristics.”
Research Policy 36: 1529–1544.
Pakes, A., and Z. Griliches. 1984. “Patents and R&D at the Firm Level: A First Look.”
R&D, Patents, and Productivity: 55–72.
Phelps, C. 2010. “A Longitudinal Study of the Influence of Alliance Network Structure
and Composition on Firm Exploratory Innovation.” Academy of Management
Journal 53: 890–913.
Png, I. 2017. Law and Innovation: Evidence from the State Trade Secrets Laws.”
Review of Economics and Statistics 99: 167–179.
Porter, M. 1976. Interbrand Choice, Strategy and Bilateral Market Power. Boston MA:
Harvard Press.
Powell, W., K. Koput, and L. Smith-Doerr. 1996. “Internorganizational Collaboration
and the Locus of Innovation: Networks of Learning in Biotechnology.”
Administrative Science Quarterly 41: 116–145.
24
Rabino, S., and T. Moore. 1989. “Managing New-Product Announcements in the

Computer Industry.” Industrial Marketing Management 18: 35–43.
Raymond, W., P. Mohnen, F. Palm, and S. Schim. 2010. “Persistence of Innovation in
Dutch Manufacturing: Is it Spurious?” Review of Economics and Statistics 92:
495–504.
Reeb, D., and W. Zhao. 2017. “Stealth or Failed Innovation?” Working Paper, National
University of Singapore.
Romer, P. 1990. “Endogenous Technological Change.” Journal of Political Economy
98: 71–102.
Ruttan, V. 1959. “Usher and Schumpeter on Invention, Innovation, and Technological
Change.” Quarterly Journal of Economics 73: 596–606.
Sauer, R. 2007. “The Erosion of the Materiality Standard in the Enforcement of the
Federal Securities Laws.” Business Lawyer 62: 317–357.
Scherer, F. M. 1986. Innovation and Growth: Schumpeterian Perspectives. Boston:
MIT Press.
Scotchmer, S. 1991. “Standing on the Shoulders of Giants: Cumulative Research and
the Patent Law.” Journal of Economic Perspectives 5, No. 1: 29–41.
Sougiannis, T. 1994. “The Accounting Based Valuation of Corporate R&D.” Accounting
Review 69: 44–68.
Sullivan, D. 1994. “Measuring the Degree of Internationalization of a Firm.” Journal of
International Business Studies 25: 325–342.
Trajtenberg, M. 1990. “A Penny for Your Quotes: Patent Citations and the Value of
Innovations.” RAND Journal of Economics 21: 172–187.
Tufano, P. 1989. “Financial Innovation and First-Mover Advantages.” Journal of
Financial Economics 25: 213–240.
Williamson, P., and E. Yin. 2014. “Accelerated Innovation: The New Challenge from
China.” MIT Sloan Management Review 55: 27–34.
Wind, J., and V. Mahajan. 1987. “Marketing Hype: A New Perspective for New Product
Research and Introduction.” Journal of Product Innovation Management 4:
43–49.
Yoon, E., and L. Deeken. 2013. “Why it Pays To Be a Category Creator.” Harvard
Business Review 91: 21–23.
Younge, K., and M. Marx. 2012. “The Market Value of Knowledge Protection: Evidence
from a Natural Experiment.” MIT Working Paper.
25

Adbi wp781

Uploaded by

Copyright:

Available Formats

Adbi wp781

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

Adbi wp781

Uploaded by

Copyright:

Available Formats

ADBI Working Paper Series

MEASURING THE DEGREE

Asian Development Bank Institute

In this publication, “$” refers to US dollars.

Please contact the authors for information about this paper.

Asian Development Bank Institute

© 2017 Asian Development Bank Institute

Keywords: innovation, R&D, patents, technological progress

JEL Classification: O30, O31, G31

2. CORPORATE INNOVATION AND GROWTH ............................................................ 4

2.1 Why Innovation? .............................................................................................. 4

3.1 R&D Expenditures ........................................................................................... 6

4. EVALUATING MULTIPLE SIGNALS......................................................................... 11

4.1 Composite Measures .................................................................................... 11

5. SUMMARY AND CONCLUSION............................................................................... 18

distortions in evaluating innovation activity on the stock exchanges of Taipei,China and

Figure 1: Ranking By Innovation: Proportion of Firms Reporting R&D*

* Derived from Koh et al. (2015).

Figure 2: Ranking By Innovation: Proportion of Firms Seeking Patents*

* Derived from Koh et al. (2015).

Figure 3: Difference in Market Ranking: Using R&D versus Patents*

* Derived from Koh et al. (2015).

To formally assess the measurement of corporate innovation, I begin with an analysis

2. CORPORATE INNOVATION AND GROWTH

2.2 The Economics of Innovation

2.3 Management Research on Innovation

2.4 Financing Innovation

3.3 New Product Announcements

4. EVALUATING MULTIPLE SIGNALS

Table 1: Descriptive Statistics

Table 2: Factor Analysis and Measuring Innovation

Panel B: Factor Loadings

Table 3: Composite Measures of Corporate Innovation

I also compute a composite measure of innovation based on an index-ranking method

Table 4: Simulation Analysis

Panel B: Standard Errors

Panel C: Multivariate Analysis

4.2 R&D Productivity

Figure 4: R&D Productivity with Noisy Signals

Figure 5: Plot of the Beta Coefficient for R&D Productivity (Noise-Biased)

5. SUMMARY AND CONCLUSION

Excluding informative signals about innovation discards useful information, making it

Focusing strictly on patents to measure innovation in a cross-market setting is

Cohen, W., and D. Levinthal. 1990. “Absorptive Capacity: A New Perspective on

Rabino, S., and T. Moore. 1989. “Managing New-Product Announcements in the

You might also like