Is R&D risky?
DOI | http://doi.org/10.1002/smj.2520 |
Published date | 01 April 2017 |
Author | Philip Bromiley,Yu Zhang,Devaki Rau |
Date | 01 April 2017 |
Strategic Management Journal
Strat. Mgmt. J.,38: 876–891 (2017)
Published online EarlyView 2 May 2016 in WileyOnline Library (wileyonlinelibrary.com) DOI: 10.1002/smj.2520
Received 16 June 2014;Final revisionreceived 29 September 2015
IS R&D RISKY?
PHILIP BROMILEY,1*DEVAKI RAU,2and YU ZHANG3
1Merage School of Business, University of California, Irvine, Irvine, California, U.S.A.
2Department of Management, College of Business, Northern Illinois University, De
Kalb, Illinois, U.S.A.
3China Europe International Business School, Shanghai, China
Research summary: Many studies use research and development (R&D) intensity or R&D
spending as a proxy for risk taking, but we have little evidence that either associates positively
with rm risk. We analyze the relations between R&D intensity (R&D spending to sales) and
R&D spending on the one hand and 11 different indicators of rm risk on the other, using data
from 1,907 to 3,908 rms in various industries over 13years. The analysis nds a general lack of
consistent positive association between R&D and rm risk, making the use of R&D as an indicator
of risk taking questionable. Furthermore, R&D intensity and spending do not correlatepositively,
suggesting they measuredifferent constructs. We discuss potential reasonsfor these nonsignicant
results. Our study demonstrates that researchers should avoid casual use of R&D as a proxy for
risk taking without explicitly providing a clear denition and measurement model for risk.
Managerial summary: Risk is a key construct in strategic management research. Many studies
in this area measure risk taking by research and development (R&D) intensity (the ratio of R&D
spending to sales) or R&D spending. However, since R&D intensity and spending have also been
used to measure various other things such as information processing demands, this raises the
question of whether R&D intensity and spending are valid indicators of rm risk. We examine
this issue by considering the associations of R&D intensity and R&D spending with conventional
measures of rm risk. We nd a general lack of consistent positive association between R&D
and rm risk, making the use of R&D as an indicator of risk taking questionable. Furthermore,
R&D intensity and spending do not correlate positively, suggestingthey measure different things.
Copyright © 2016 John Wiley & Sons, Ltd.
INTRODUCTION
Since Bowman (1980), a substantial research tra-
dition in strategy has addressed risk. Most of the
literature attempts either to explain rm risk taking
or to estimate the inuence of risk taking on rm
performance (Bromiley and Rau, 2010).
Keywords: risk; R&D intensity; R&D spending; construct
measurement; validity
*Correspondence to: P. Bromiley, Merage School of Business,
University of California, Irvine, Irvine CA, 92697-3125. E-mail:
bromiley@uci.edu
[Correction added on 30 May 2016, after rst online publication:
The corresponding author’s address has been corrected.]
Copyright © 2016 John Wiley & Sons, Ltd.
Strategic management research has adopted
a number of measures of risk and risk taking
(Bromiley and Rau, 2010). Many studies measure
risk taking by research and development (R&D)
intensity or R&D spending (Barker and Mueller,
2002; Chen and Miller, 2007; Devers etal., 2008;
Hoskisson, Hitt, and Hill, 1993; Miller and Bromi-
ley, 1990). However, researchers also use R&D
spending or intensity to measure other constructs,
including time horizon (Bushee, 1998; Lundstrom,
2002), resources as dened by the resource-based
view (Mahoney and Pandian, 1992), information
processing demands (Henderson and Fredrickson,
1998), and other constructs. Indeed, Ketchen,
Ireland, and Baker (2013) question interpreting
Is R&D Risky? 877
R&D as reecting any specic construct given that
scholars have claimed it reects so many different
constructs.
We adopt a differentperspective on this problem,
considering whether R&D intensity or R&D spend-
ing as an indicator of rm risk taking associates
positively with other measures of rm risk— its
nomological validity. We begin by considering the
conceptualization of the risk construct in strategic
management research. Strategy discussions some-
times use the terms risk and risk taking inter-
changeably, and mix a variety of concepts, includ-
ing preference for a desired level of risk, behav-
iors, or activities that increase risk; lack of ability to
predict performance; and variability in performance
outcomes. We also review some theories used in
strategic management research that portray R&D as
a means to reduce risk, and more particularly,poten-
tial variability in a rm’soutcomes and discuss their
implications for our analyses and expected ndings.
Following these discussions of R&D and risk in
strategic management scholarship, we present anal-
yses that test four potential relations between R&D
spending or intensity, and other measures (both ex
ante and ex post) of rm risk. These include models
in which R&D spending or intensity creates risk so
that R&D positively inuences contemporaneous
or subsequent rm risk, and models in which a
general risk propensity inuences R&D so rm risk
inuences contemporaneous or subsequent R&D
spending or intensity. While we think R&D spend-
ing or intensity inuencing current or subsequent
rm risk best ts arguments that R&D is risky, we
include the additional relations for completeness.
For robustness, we perform the analyses with 11
different risk metrics, including ones based on
stock price, variation in ROA, downside risk in
terms of ROA, and analyst forecasts.
This article contributes to strategic manage-
ment research by examining the validity of R&D
intensity and R&D spending as proxies for risk
taking. Evaluating measurement validity is critical
in generating credible research (Boyd, Gove, and
Hitt, 2005a, 2005b; Boyd et al., 2013; Ketchen
et al., 2013; Podsakoff, Shen, and Podsakoff, 2006;
Venkatraman and Grant, 1986). Despite the recog-
nition of the importance of rigorous construct mea-
surement in strategic management, however,“ …as
to date, relatively little emphasis has been placed on
measurement issues within strategic management”
(Boyd et al., 2013: 3). Our study thus represents a
step toward addressing an important measurement
issue in strategic management research.
RISK PREFERENCES, BEHAVIORS, AND
OUTCOMES
Many studies across strategy, accounting, and
nance have used R&D as a proxy for risk tak-
ing, or equivalently, used risk taking theoretical
arguments to develop hypotheses to explain R&D
(see, e.g., Barker and Mueller, 2002; Baysinger and
Hoskisson, 1989; Chen, 2008; Chrisman and Patel,
2012; Devers et al., 2008; Eberhart, Maxwell, and
Siddique, 2008; Gentry and Shen, 2013; Hill and
Snell, 1988; Hoskisson and Hitt, 1988; Hoskisson
et al., 1993; Kor, 2006; McAlister, Srinivasan, and
Kim, 2007; Wedig, 1990).
When we describe a rm activity as risky, we
implicitly claim that doing this activity increases
rm risk. That is, if a rm action is a legitimate
form of rm risk taking, it should positively inu-
ence rm risk. However, the term risk has sev-
eral different connotations in strategic management
research. Risk can refer to rm preferences, behav-
iors or actions, or outcomes.
Let us begin by examining the treatment of rm
risk preferences, that is, rms’ desired levels of
risk, in strategic management theories. The most
commonly used theories to generate hypotheses
regarding rm risk taking do not have risk prefer-
ence as a construct. For example, in expected utility
theory, decision-makers do not have an explicit
value or preference associated with risk; rather, risk
preference is a derived description that reects the
curvature of the utility function. In the behavioral
theory of the rm (Cyert and March, 1963), risk
does not appear as a construct. The behavioral the-
ory of the rm explicitly assumes that organizations
do not have consistent preferences, which rules out
their having consistent risk preferences. Prospect
theory, also often used to explain rm risk, is an
individual level theory that explains choice based
on how one values specic potential outcomes and
weights their probabilities. Again, risk preference
per se is not a construct in the theory, although
we can infer a risk preference from the pattern
of choices predicted by the theory (see Bromiley,
2010).
These theories without direct risk preferences
differ from both agency theory’s treatment of risk
and the prescriptive literature on risk. Agency
Copyright © 2016 John Wiley & Sons, Ltd. Strat. Mgmt. J.,38: 876–891 (2017)
DOI: 10.1002/smj
To continue reading
Request your trial