° Academy of Management Journal

2000, Vol. 43. No. 6, ]nB-n29.

IS CEO PAY IN HIGH-TECHNOLOGY FIRMS

RELATED TO INNOVATION?
DAVID B. BALKIN
University of Colorado al Boulder

GIDEON D. MARKMAN
Rensselaer Polyltechnic Institute

LUIS R. GOMEZ-MEJIA
Arizona State University

This study uses the resource-based view ofthe firm and agency theory to examine the
relationship between innovation and CEO pay in 90 high-technology firms. With firm
size, performance, and other factors controlled, CEO short-term compensation was
related to innovation as measured hy numher of patents and R&D spending. The data
also suggest a le.ss consistent temporal relationship between innovation and long-term
CEO compensation in the high-technology firms. In a control sample of 74 low-
technology firms, there was no relationship hetween innovation and either short- or
long-term CEO pay.

High-technology firms need to produce a steady innovation is less critical for success. We used a
stream of innovations in order to survive in hyper- comprehensive measure of innovation activity that
competitive technology markets (D'Aveni, 1994; includes both numbers of patents and investments
Hamel & Prahalad, 1994; Schilling & Hill. 1998). Yet in innovation and research and development
at present little is known ahout the relationship be- (R&D). Prior CEO pay studies including R&D have
tween chief executive officer (CEO) pay and innova- used it primarily as a surrogate for executive risk
tion (Comez-Mejia & Wiseman, 1997). This study taking across a broad spectrum of firms (e.g., Hosk-
makes two important contrihutions to this literature. isson, Hitt, & Hill, 1991, 1993; Hoskisson & John-
First, it combines the resoiu'ce-based view of tlie firm son, 1992).
and agency theory to suggest that top executives in
high-technology companies should he rewarded for
sustaining the firms' capability to innovate. Thus, it THEORETICAL FRAMEWORK AND
brings the resource-based view into the large and HYPOTHESES
ever-growing research on CEO pay, most of which has
an agency-theoretic orientation. Taking this com- According to the resource-based theory of the
bined perspective, we argue that in a very dynamic firm, firms are heterogeneous in terms of the re-
business environment, such as that of high-technol- sources they control. Organizational resources con-
ogy firms, the value and uniqueness of knowledge- sist of ali the assets, capabilities, attributes, and
intensive resources can be swiftly lost to competitors. knowledge a firm possesses that enable it to de-
Therefore, evidence of CEO innovation efforts in velop and implement strategies that improve its
high-technology firms needs to be rewarded using performance (Barney, 1991). A firm's resources can
short-tenn pay (to support continuous self-trans- be a source of competitive advantage in markets
formation) and long-term compensation (to nurture when it is difficult for the firm's rivals to obtain like
and develop the core competencies that result in en- resources. Scarce resources create entry barriers for
during uniqueness and value in order to outnm the firms that do not have them (Wemerfelt, 1984). For
competition). example, a pharmaceutical firm that develops a
new drug that is superior to other products for
The study also fills an important empirical gap in treating a serious disease will patent the drug and
the executive compensation literature. It directly take legal action against those who infringe on the
examines, in a comparative fashion, the differential patent, thus establishing entry harriers that make it
relationship hetween innovation efforts and short- more difficult to imitate the drug until the patent
and long-term CEO pay within high-technology expires. Both the patent to make the drug (which
firms, where innovation is a key source of compet- grants the patent owner legal rights to protect its
itive advantage, and low-technology firms, where intellectual property) and the capability to make

1118
2000 Balkin, Markman, and Gomez-Mejia 1119

other innovative drugs are examples of resources flict with innovation, such as additional advertis-
that may provide a competitive advantage to a firm. ing, plant expansion, or mergers and acquisitions,
A technological innovation improves the perfor- the result may be short-term profits achieved at the
mance of a product or service in terms of its quality, cost of having no new innovations in the pipeline
cost, speed, or number of features (Cohen & in the long term. An empty pipeline can be a fatal
Levinthal, 1990). The technological innovation problem in fast-changing high-technology markets
may be part of a continuous refinement of a prod- where product life cycles may average only 18 to 24
uct, or it can result in a discontinuous and radical months (Schilling & Hill, 1998). Innovation re-
change that makes other firms' technologies obso- quires making long-term investments in R&D
lete (Tushman & Anderson, 1986). projects that may have a negative impact on more
Resources that can provide competitive advan- immediate financial statements (Hoskisson et al.,
tage to a firm (1) are valuable, (2) are rare, (3) are 1993). Consequently, one would expect that high-
difficult to imitate, and (4) have qualities that make tech firms should link CEO pay to evidence of
them irreplaceable (Barney, 1991). In a high-tech- innovation efforts.
nology firm, the capability to be a technology inno- Furthermore, since executives are the leaders of
vator has the potential to be a resource that is flrms, the form and structure of executive pay
valuable, rare, hard to imitate, and difficult to sub- should mirror the pay strategies applied to scien-
stitute for (Grant, 1991). It is a major challenge to tists and engineers and other key employees
obtain talented scientists and engineers and to de- (Werner & Tosi, 1995). There is evidence that the
velop protocols that lead to successful innovations pay of scientists and engineers in high-technology
that can be taken to the market. firms is often linked to meeting innovation goals
In high-technology markets, firms that have the such as achieving significant technological project
capability to innovate can be expected to generate milestones (Balkin & Bannister, 1993; Riggs, 1983).
greater profits than those that are noninnovators. R&D-intensive firms operate in what Hambrick
To generate profits, a firm must continuously make and Finkelstein (1987) defined as a high-discretion
large expenditures in R&D in order to sustain its context. This context has two key implications for
capability to innovate at the cutting edge of tech- CEO compensation. First, in this context managers
nology (Jelinek & Schoonhoven, 1993). By sustain- enjoy much autonomy and face a wide latitude of
ing R&D spending, an organization increases its choices when making strategic decisions. As re-
ahsorptive capacity to learn and take advantage of sources devoted to innovation increase, the poten-
technological knowledge that is available by scan- tial impact of the CEO on a firm's relative success
ning the external environment (Cohen & Levinthal, or failure also increases. Thus, CEO compensation
1990). The capability to take advantage of new should increase accordingly because "implicitly or
technological information with a first mover prod- explicitly, boards of directors will tend to recognize
uct or a fast-follower product (one that comes to the host of decisions that need to be made in such
market soon after the pioneering firm's product) is situations, and will draw an inference that there
a critical performance factor in high-technology in- can be a substantial performance difference that
dustries. By sustaining R&D spending, a CEO sup- can emanate from a 'good' manager and a 'bad'
ports a firm's need for absorptive capacity enabling manager in an attempt to obtain the former"
learning and competition in high-velocity envi- (Hambrick & Finkelstein, 1987: 396). Second,
ronments (Eisenhardt, 1989). As Schilling and investments in innovation are inherently risky
Hill noted, "R&D projects are the precursor to com- (Baysinger, Kosnick, & Turk, 1991; Graves &
mercial development projects and are necessary Langowitz, 1993). In other words, emphasis on in-
to develop cutting edge strategic technologies" novation implies a greater variability of outcomes
(1998: 70). and a greater probability of failure. In the words of
Top executives reside at the strategic apexes of Finkelstein and Boyd, "When multiple courses of
their firms, and they make, influence, or are ulti- action are possible, uncertainty and complexity go
mately responsible for critical resource allocations up, and it becomes more difficult to predict firm
pertaining to investments in new products and performance with much accuracy" (1998: 181). As
technologies. These decisions are in turn likely to a result, CEOs in firms devoting more resources to
be influenced by how executives are rewarded. For innovation should receive higher pay in order to
a firm competing in high-technology markets, the compensate them for bearing this greater risk
incentive system should induce executives to make (Finkelstein & Hambrick, 1996). In particular, CEOs
resource allocation decisions that will assist the face a greater probability of termination and a
lirm in sustaining its innovation capability. If ex- higher likelihood of a negative reputation in the
ecutives deploy firm resources for goals that con- market if expensive investments in innovation do
1120 Academy of Management Journal December

not produce the desired ex post results, because it in R&D-intensive high-technology firms is inher-
is difficult for outsiders to distinguish unfortunate ently higher because decisions might fail in spite of
circumstances from poor decisions (Wiseman & the best efforts of managers.
Gomez-Mejia, 1998). Alternatively, principals in high-technology
At the same time that CEOs should be more firms may reward executives for indicators of inno-
highly compensated as investments in innovation vation activity (such as R&D investments and pat-
expand, it seems logical that, as the productivity of ents) that are more clearly influenced by manage-
such investments increases, there should also be rial decisions and that are believed to have a
concomitant increases in their pay. Because com- beneficial effect on firm performance, even though
plexity makes the relationship between senior the nexus between those decisions and perfor-
managers' actions and firm performance very un- mance outcomes is rather ambiguous. Such a sys-
certain in high-technology markets, executive com- tem reduces managerial risk bearing, increasing the
pensation packages may be loosely linked to ob- probahility that agents in high-technology firms
served performance results. If one assumes that the will channel firm resources into risky R&D projects
ability to innovate gives a firm a competitive ad- with uncertain payoffs.
vantage and that an executive has more power to Baysinger, Kosnik, and Turk (1991) found that
influence innovation efforts than performance per the presence of insiders on a firm's board and the
se, compensation should be more tightly linked to existence of institutional investors with significant
the former than to the latter. In other words, high- equity positions in the firm led to higher R&D
technology firms are expected to provide financial spending, inferring that these parties were more
rewards for executives who can sustain the organi- capable of evaluating executives for the appropri-
zational capability to innovate apart from observed ateness of their decisions rather than mechanisti-
firm performance results. Therefore, the greater the cally relying on observed performance results
technological intensity of a firm, the more execu- (which would increase managerial risk bearing and
tive pay should he aligned with innovation. thus discourage R&D investments]. However,
Tbe literature on agency theory also supports the Baysinger and colleagues' study, like the rest ofthe
argument that executives in R&D-intensive firms literature reviewed earlier, did not examine the
should be rewarded for evidence of innovation- extent to which executives are actually rewarded
related activities within the firms, independently for evidence of innovation-related activities when
of observed financial performance outcomes. these activities are most necessary for a sustainable
"Agency theory [is characterized] hy its emphasis competitive advantage. The present study empiri-
on the risk attitudes of principals and agents" (Bar- cally examines this important research question,
ney & Hesterly, 1996: 124). Agents are assumed to comparing findings for high- and low-technology
be risk-averse because their employment security firms.
and income are tied to one firm. Principals, on the Executive compensation consists of short-term
other hand, are assumed to be risk-neutral because and long-term pay (Westphal & Zajac, 1994). Short-
they can diversify their shareholdings over multi- term pay for executives consists of the base salary
ple Firms. This risk differential (Beatty & Zajac, and a short-term bonus tied to performance objec-
1994) creates a conflict of interest between risk- tives of one year or less and is paid in the form of
averse managers, who prefer to be conservative in cash. Long-term pay for executives consists primar-
their decisions to reduce their risk exposure, and ily of stock options and other forms of equity-based
risk-neutral principals, who prefer tliat agents max- compensation tied to achieving objectives over pe-
imize firm returns (Garen, 1994). "The challenge of riods ranging from three to five years.
corporate governance is to set np incentive align- High-technology firms can take a short- or a long-
ment mechanisms that alter the risk orientation of term approach to innovation (Leonard-Barton,
agents to align them with the interests of princi- 1995). For example, semiconductor manufacturing
pals" (Wiseman & Gomez-Mejia, 1998: 133). firms snch as Intel have developed several succes-
For incentive alignment to be effective in making sive generations of integrated circuit chips simul-
an agent's risk orientation more consistent with taneously to ensure that increasingly powerful, up-
that of principals, it is necessary to use "payoff graded chips reach the market in a steady stream,
criteria" that agents can influence. To the extent before those of competitors. An innovation (pro-
that the uncertainty of outcomes is high, the use of tected by a patent) applied to a soon-to-be-released
incentives tied to those outcomes would mean that product may also henefit products that are cur-
agents bear more risk because they "feel that they rently under development and are expected to
are not responsible for their own pay" (Eisenhardt, reach the market in the next three to five years.
1988: 490). As noted earlier, outcome uncertainty Because technological innovations involve high
2000 Balkin, Markman. and Gomez-Mejia 1121

levels of tacit knowledge, incremental, cumulative, to the high-technology firms in the analysis. The
and radical innovations are generally intertwined high-technology sample consisted of 90 electronics
in a yet-to-be-specified, iterating system (Saviotti & and health science firms (see the Appendix).
Mani, 1998). Thus, it may be difficult to disentan- The high-technology firms were listed in the
gle the short- and long-term implications of a Forbes 1994-95 special issues on CEO pay. The
patent or other type of innovation. The achieve- firms selected in the Forbes data set are the 800
ment of a patent at one time may pave the way for largest U.S. companies in terms of sales. Their av-
an innovative product later. erage R&D intensity (R&D spending as a percentage
With proper incentives for both short- and long- of annual sales) in 1993 was 7.19 percent, which is
term innovations, a CEO should be motivated to above the 5 percent benchmark for R&D intensity
manage the flow of products through a firm's inno- that several earlier studies have used to identify
vation pipeline into the market. In a high-technol- high-technology companies (Balkin & Gomez-
ogy firm, an uninterrupted flow of innovations is a Mejia, 1987; Koberg, Rosse, & Bergh, 1994). Data
critical activity that sustains its ability to compete. representing CEO pay and firm performance wtire
The importance of aligning CEO pay with innova- collected for a three-year period (1992-94). Each
tions in high-technology firms from different time CEO in the data set (for both the high-technology
perspectives is consistent with the resotirce-based and control samples) had a minimum CEO tenure
and agency theory explanations made earlier in the of three years during this period (that is, all CEOs
paper (Barney, 1991; Jensen & Murphy, 1990). Over remained in their posts for the entire period the
the short term, innovations can be valued on their study was conducted). The average annual sales of
technical merits, as units of intellectual property the flrms in the sample was $7.2 billion in 1993 and
(such as patents), and, over longer time periods, S6.B billion in 1992.
innovations will be valued in the market, which The control sample consisted of 74 firms, also
will react to the stock of the firm that produced the listed in the Forbes 1994-95 special issues, that
innovation. The extent to which a firm is capable of were in non-high-technology industries (durable
doing innovation better than competitors should be and nondurable manufacturing) yet comparable to
favorably reflected in its stock price and its relative the high-technology sample in terms of size and
appreciation, which in turn will benefit the CEO as performance (see the Appendix). The average an-
a shareholder. Thus, the CEO should be partially nual sales ofthe firms in the control group was $6.2
rewarded for innovation as a technical benchmark billion in 1993 and $5.9 billion in 1992. The aver-
with cash, a short-term reward, and partially re- age R&D spending (as a percentage of annual sales
warded for innovation as added financial value in for 1993) was 1.9 percent, which is about a fourth of
the form of stock appreciation, a long-term reward. the high-technology sample's average and is below
This formulation suggests that innovations should the 5 percent henchmark scholars have used to
be related to both short- and long-term CEO pay in identify high-technology firms. All CEOs remained
high-technology firms. This discussion leads to the with their firms for the entire period under inves-
two hypotheses to be tested in this study: tigation. As in the high-technology sample, this
characteristic in effect controls for the potentially
Hypothesis 1. tn high-technology firms, inno- positive impact of CEO transitions on CEO pay
vation should have a positive effect on short- (Gomez-Mejia, Tosi, & Hinkin, 1987).
term executive pay, over and ahove what could
be expected on the basis of financial perfor-
mance results. Operational Measures
Hypothesis 2. In high-technology firms, inno- CEO pay. We used the two primary measures of
vation should have a positive effect on long- CEO pay: (1) short-term compensation, consisting
term executive pay, over and above what could of annual salary and bonus, and (2) long-term pay.
be expected on the basis of financial perfor- The annual salary and bonus represent the total
mance results. cash compensation received during a specific year.
Annual salary and bonus for 1993 and 1994 (in
thousands of dollars) were taken from the Forbes
METHODS data set.
Long-term pay represents the equity-based com-
Sample and Data Collection Procedures pensation of a CEO. The estimation procedure we
Two samples of firms were selected for this used to value long-term compensation was based
study. The control sample consisted of a set of on the approach used hy Lambert, Larcker, and
non-high-technology firms that could be compared Weigeh (1993) and by Finkelstein and Boyd (1998);
1122 Academy of Management Journal December

it is to value stock options at 25 percent of their for 1992 and 1993, as reported in Compact Disclo-
exercise price. The data for the long-term pay came sure. A control variable was used to control for
from proxy statements for 1993 and 1994 as re- industry effects in both samples (Powell, 1996). In
ported in the Lexis-Nexis data research software the high-technology sample, the industry variable
(version 4.1). Long-term pay was thus the number was coded 1 for electronics firms and 0 for nonelec-
of shares multiplied by 25 percent of the exercise tronics firms, which were primarily in the health
price. sciences. In the control sample, the industry vari-
Innovation. Innovation was a composite measure able was coded 1 for durable goods companies and
computed by adding standardized values of R&D 0 for others. We did not use a finer industry hreak-
spending and number of patents. We computed down in order to maintain an adequate ratio of
separate composite innovation measures for 1992 cases to variables. Tosi and Gomez-Mejia (1994), an
and 1993. earlier study of CEO pay, used this industry classi-
In this study, R&D spending was used as a proxy fication.
for innovation, including hoth projects under de- A set of four control variables representing gov-
velopment and innovations not publicly disclosed. ernance structure was also used in the analysis.
The other measure of innovation used in this study, These control variahles consisted of (1) CEO tenure,
patents, represents only a subset, because a firm a measure often used as an indicator of CEO en-
can choose other legal forms, such as copyright and trenchment (Hill & Phan, 1991), (2) inside direc-
trade secrets, to protect its intellectual property tors' ratio (Baysinger et aL, 1991; Westphal & Zajac,
from competitors (Tabak & Barr, 1998). Combining 1994), or the ratio of inside hoard members (current
patents aud R&D spending captures organizational firm member) to outside board members (others),
innovation efforts more completely, with R&D and (3) the presence of individual and/or corporate
spending measuring investments in innovation and owners with 5 percent or more of the company's
number of patents indicating innovation outputs. stock (O'Reilly, Main, & Crystal, 1988; Tosi &
Another reason patents and R&D spending were Gomez-Mejia, 1989), which are hoth indicators of
combined is that, as individual variables, they are owner control of an organization. Individual owner
highly correlated with each other. We report R&D control and corporate owner control consisted of
spending as a company's annual R&D expendi- two dummy variables coded 1 for the presence of
tures, as did Graves and Langowitz (1993) and De- one or more individual or corporate owners with 5
chow and Sloan (1991). The annual R&D expendi- percent or more ownership of the company stock.
tures (in millions of dollars) during 1992 and 1993 Firms without such owners were coded 0. The gov-
were gathered from Compact Disclosure. Number ernance structure data for the inside directors' ratio
of patents has been used by other scholars as a were collected from the Standard & Poor's (S&P)
measure of innovation productivity (Acs & Au- Register of Corporations, Directors, and Executives
dretsch, 1988; Griliches, 1990; Holthausen, Larker, for 1993 and from proxy statements for 1993 pro-
& Sloan, 1995). Not all patents are equally valuable, vided by Compact Disclosure for the individual
but patent counts have been shown to be related to and corporate ownership variables. The CEO ten-
changes in firm value, profitability, and sales ure variahle was provided by the Forbes 1994 spe-
growth (Griiiches, 1990). The annual number of cial issue on CEO pay that reports CEO experience
company patents during 1992 and 1993 was gath- through the year 1993.
ered from the U.S. Patents Office. A control vari-
able representing firm size was used to control for
scale effects on R&D spending and number of pat- Analysis
ents (Cohen & Levinthal, 1990).
Hierarchical regressions were performed on the
Control variables. Several measures were used two measures of CEO pay for the years 1994 and
as control variables in the study. These included 1993: short-term pay (salary and bonus) and long-
firm size and firm performance, the two most com- term pay. As did Finkelstein and Boyd (1998), we
mon predictors of executive pay (as per an in-press gathered the independent (innovation) and control
meta-analysis by Tosi, Werner, Katz, and Gomez- variables entered in each regression analysis for the
Mejia). Firm size was annual sales for 1992 and year prior to the year for which the dependent
1993, in millions of dollars, from Compact Disclo- measures of GEO pay were reported. The lagged
sure data. Firm performance was measured as re- hierarchical regressions were performed separately
turn on assets (ROA), a ratio widely used in other on the high-technology and control samples. First,
studies of executive compensation (Finkelstein & the seven control variables were entered as a block
Boyd, 1998; Sanders & Carpenter, 1998; Westphal & in step 1. Next, we entered the innovation variable
Zajac, 1994). Firm performance consisted of ROA (the standardized measure of R&D spending and
 ts in ts (s
O (D o:
ro q to eo
CD •a- Ol lO
CO O O O O CO I s d CS r - od d d d d CO

o CD CD a". I - H'^
C-J Ol o ^ r-1 CO CO cs^ ro CO 01
r-1 O rH O •* t s d eo d rH
d d d d CO
d 6 CO cs cD """
• * I-l
fM
eo

! I I I I

CD CO m o
a o o o
r rr

l l l i I I

r r

I [

13
d
o
. CO CO CO
rH rH rH cs rH ta
1 1 1 I I
28
20
35
19

f-J r^ O) rH
06
50

20

24

o
1 I I I

1 I I I

o o o o
O jj
M "3.
u c
(^ CD l/J rH N TH
p ro cs es i ; cs
d C O d IS d d O O O Ol

IS
a
O

Q S .22
•a J2 "3

= D V V
E 5 Q.a
 1124 Academy of Management Joumal December

number of patents) in step 2 to examine the corre- and for the control sample, it was $1.13 million in
sponding level of statistical significance and the 1994 and $1.00 million in 1993. The magnitude of
amount of unique variance in CEO pay (A/?^) ac- the observed differences between the high-technol-
counted for by innovation. Finally, each of the re- ogy and control samples on these variables did not
gressions was tested for the presence of heterosce- reach statistical significance.
dasticity by examining the distributions ofthe error
Table 2 summarizes the lagged hierarchical re-
residuals.
gression analysis on short-term pay (salary and bo-
nus) for 1994 and 1993. In the regression analysis
RESULTS for the high-technology sample, the control vari-
ables were entered first and explained 39 and 36
Table 1 summarizes the descriptive statistics and percent of the variance in short-term pay for 1994
correlation matrix of the variables used in the high- and 1993, respectively. The innovation variable,
technology and control samples. The mean size entered in step 2, accounted for 29 and 17 percent
(annual sales revenues) of a company in the high- of additional variance in cash compensation for
technology sample was $7.2 billion in 1993 and 1994 and 1993. respectively (p < .01). The beta
S6.9 billion in 1992. and the control sample mean
coefficients of innovation are significant in both the
was $6.2 billion in 1993 and S5.9 billion in 1992.
1994 and 1993 regressions (p < .01). These find-
The mean ROA in the high-technology sample was
.09 in 1993 and .07 in 1992. and in the control ings strongly support Hypothesis 1, which predicts
sample it was .05 in 1993 and .04 in 1992. Thus, the a positive relationship between innovation and
average size of the companies was comparable in short-term CEO pay in high-technology firms.
the two samples, while the ROA was somewhat The lagged regression analysis in the control
higher in the high-technology sample. The ROA of sample showed that control variables accounted for
the control sample is comparable to that ofthe S&P 20 and 41 percent of short-term pay variance for
500 industrials, which was .03 in 1993 and .03 in 1994 and 1993. When innovation was added in step
1992 {Standard & Poor's, 1998). The average CEO 2 for the years 1994 and 1993, it accounted for a
salary and bonus for tho high-technology sample miniscule amount (1 percent) of additional ex-
was $1.59 million in 1994 and $1.33 million in plained variance. The beta coefficient of the inno-
1993, and for the control sample, it was $1.43 mil- vation variable was nonsignificant in the control
lion in 1994 and $1.23 milUon in 1993. The long- sample regressions for both years. The tests for
term CEO pay for the high-technology sample was differences in the regression coefficients (Maddala.
$1.14 million in 1994 and $0.72 million in 1993. Rao. & Vinod. 1993) for the innovation variable in

TABLE 2
Resulls of Hierarchical Regression Analysis on Salary and Bonus'
Higb-Technology Sample Conlrol Sample

1994 1993 1994 1993

Variable Step 1 Step 2 Slep 1 Slep 2 Step I Step 2 Slep ] Step 2

ROA .16 .22* .17 .25* .15 .17 .22 .23
Size .53** .21 .36* .08 .35" .28 .53** .56"
CEO tenure -.05 .18 .19 -.13 -.11 -.02 -.03
Inside director ratio .06 M -.03 .01 -.05 .07 -.09 -.08
Individual owner control -.07 -.22*' -.24 -.40** -.20 -.22 -.27 -.26
Corporate owner control -.16 .12 -.28 -.29' .07 .03 -.18 -.16
Industry .12 .10 .21 .24 .14 .13 -.03 -.03
Innovations .66** .52** .16 .06

B- .68 .36 .53 .20 v21 .41 .42

.29 .17 .01 .01
F 3.61"* 10.30** 3.71*" 6.35'* 1.32 1.25 3.74** 3.22**

For the high-tech samplp. n = 90; for the control sample, n = 74. Values are standardized regression coefficients. ROA, size, and
innovations were gathered from the year prior tn the year reported for CEO pay in each regression.
* p < .05
** p < .01
2000 Balkin. Markman, and Gomez-Mejia 1125

the high-technology and control samples were sta- control firms and that the results are more consis-
tistically significant at p ^ .05. tent for short- than for long-term pay.
Another set of lagged hierarchical regressions
was also performed on long-term pay. The results,
summarized in Table 3, indicate that for the high- DISCUSSION
technology sample the innovation variahle entered This study extended the application of the
in step 2 accounted for 0 and 30 percent of addi- resource-based theory of the firm to the domain of
tional variance for 1994 and 1993 long-term pay, executive compensation. With a sample of firms
respectively. Ofthe two years, only the 1993 model selected from high-technology industries, it was
was significant (p ^ »01). The regressions on 1994 shown that a firm's capahility to innovate, as mea-
and 1993 long-term pay for the control sample were sured by its numher of patents and R&D investment
not significant. The difference in the regression level, is related to CEO pay apart from the ohserved
coefficients for the innovation variahle hetween financial performance results. Earlier research that
hoth samples [with long-term pay as the dependent has examined high-technology reward systems has
variable) was statistically significant (p < .05) for shown that technology managers, scientists, engi-
1993. The data provide mixed support for Hypoth- neers, and other R&D employees are also likely to
esis 2, which predicts a positive relationship be- have a portion of their pay contingent on the
tween innovation and long-term pay in high-tech achievement of technology milestones (e.g.. Balkin
firms. & Comez-Mejia, 1987; Gomez-Mejia & Balkin,
In short, the data suggest that an empirical rela- 1989).
tionship exists between innovation and CEO short- The finding that among high-technology firms
term pay in the high-technology sample. There is innovation explains an average 23 and 15 percent
also some mixed evidence to suggest an empirical of the variance in CEOs' short- and long-term pay,
relationship exists between innovation and long- respectively, (vis-a-vis negligihle amounts among
term pay in the high-technology sample. The re- their low-technology counterparts) is an important
gressions in the control samples did not reveal ev- contrihution to the executive pay literature. It
idence that innovation and CEO pay (short or long shows that CEO incentives for innovation are used
term) were empirically related. Thus, our empirical extensively where they are needed the most—that
findings indicate that the hypothesized relation- is, among knowledge-intensive firms.
ship hetween CEO pay and innovation existed in The lack of a relationship between firm perfor-
the high-technology firms we studied but not in the mance (as measured by ROA) and CEO pay in the

TABLE 3
Results of Hierarchical Regression Analysis on Long-Term Pay"
High-Technology Sample Control Sample

1994 1993 1994 1993

Variable Step 1 Step 2 Step 1 Step 2 Step 1 Step 2 Step 1 Step 2

ROA -.02 -.02 -.32" -.19 .10 .12 .33 .33

Size .27 .24 .42** .08 .15 .08 .23 .20
CEO tenure -.01 -.01 .06 .08 .11 .13 .22 .22
Inside director ratio .13 .14 -.07 -.03 -.25 -.30 -.23 -.26
Individual owner control -.17 -.18 .20 .04 -.15 -.15 -.29 -.28
Corporate owner control -.11 -.11 ,01 .02 -.09 -.12 -.09 -.10
Industry .11 .11 -.19 -.16 -.17 -.18 -.08 -.08
Innovations .05 .68** .17 .07

R^ .13 .13 .28 .58 .11 .13 .16 .17

^B^ .00 .30 .02 .01
F 1.12 0.97 3.00** 8.98** 0.57 0.58 li.Iifi 0.75,.

I''or the high-tech sample, n = 90: for the control sample, n = 74. Values are standardized regression coefficients. ROA, ize, and
innovations were gathered from the year prior to the year reported for CEO pay in each regression.
* p< .05
**p<.01
1126 Academy of Management JoamaJ December

high-technology sample is consistent with other the questions examined in this study and apply a
research showing that in a turbulent environment combined resource-based view of the firm and
(such as that faced by high-technology firms), there agency theory to different contexts to explain CEO
is greater "decoupling" between CEO pay and ob- pay under conditions of uncertainty requiring man-
jective performance indicators (Garen. 1994). This agerial discretion. A key resource other than inno-
decoupling may be due to the fact that the variance vation, such as brand reputation or marketing and
of these indicators and their noise increases with distribution channels, might also be studied under
environmental uncertainty, so that performance at- conditions of environmental uncertainty. It would
tributions to the CEO become more tenuous. be useful to test whether a CEO is rewarded for
Our findings also support the "positivist" agency maintaining these resources,.assuming satisfactory
argument that when outcome uncertainty is high, measures could be obtained. The results of this line
principals will base agents' rewards on criteria they of research would provide additional evidence to
can control and that are believed to have a salutary suggest tbat in conditions requiring significant
effect on firm performance, even though the cause- managerial discretion, CEO pay criteria should cap-
effect relationship in terms of rewarded actions- ture management decisions that favorably affect
performance outcomes is difficult to demonstrate. firm performance while being neither performance-
Specifically, in high-technology industries execu- outcome-based nor behavioraliy assessed in a strict
tives may be rewarded for innovation-related activ- sense.
ities such as R&D projects and patents rather than
for financial outcomes. This incentive would be
likely to induce executives of these firms to take
higher risks (as evidenced by investments in inno- REFERENCES
vation), not only because they are paid to do so, but Acs, A. J., & Audretsch, D. B. 1988. Innovation in large
also because they do not bear the financial risk and small firms: An empirical analysis. American
associated with the uncertain performance conse- Economic Review, 78: 678-690.
quences of those actions. Balkin, D. B., & Bannister, B. 1993. Explaining pay forms
Consistent with expectations, in the control sam- for strategic employee groups in organizations: A
ple of low-technology firms, executives were not resource dependence perspective. Joumal of Occu-
rewarded for innovation-related activities. Interest- pational and Organizational Psychology, 66: 139-
ingly enough, executive pay is not related to per- 151.
formance in this group of firms either. This may Balkin, D, B., & Gomez-Mejia, L. R. 1987. Toward a con-
perhaps be attributed to contingency variables not tingency theory of compensation strategy. Strategic
included in the model, such as structural complex- Management Joumal, 8: 169-182.
ity and labor market constraints (cf. Gomez-Mejia, Barney, J. 1991. Firm resources and sustained competi-
1994]. In any case, these results are consistent with tive advantage. Journal of Management, 17: 9 9 -
much of the statistical evidence yielded by hun- 120.
dreds of studies conducted during the past 70 Barney. J., & Hesterly, W. 1996. Organizational econom-
years showing a weak to nonexistent relationship ics: Understanding the link between organizations
between executive pay and firm performance and economic analysis. In S. R. Clegg, C. Hardy, &
(Gomez-Mejia & Wiseman, 1997; Tosi et al., 2000). W. R. Nord (Eds.). Handbook of organization stud-
The results of this study of CEO pay and innova- ies. 115-147. London: Sage.
tion contribute to the growing body of research that Baysinger, B. D., Kosnik, R. D., & Turk, T. A. 1991. Effects
examines management under uncertain conditions of board and ownership structure on corporate R&D
ranging from turbulent to placid. Research on cor- strategy. Academy of Management Journal, 34:
porate governance has found that CEO duality (hav- 205-214.
ing the same person serve as CEO and board chair- Beatty. R. P., & Zajac, E. J. 1994. Managerial incentives,
person) is related to highly uncertain situations, monitoring, and risk bearing: A study of executive
where there is a need for significant managerial compensation, ownership, and board structure in
discretion (Boyd, 1995: Finkelstein & D'Aveni, initial public offerings. Administrative Science
1994). The present study contributes to this re- Quarterly, 39: 313-335.
search by showing that in the context of high dis- Boyd, B. K. 1994. Board control and CEO compensation.
cretion due to uncertainty, a characteristic of high- Strategic Management foumal, 15: 335-344.
technology industries, CEOs are rewarded for Boyd, B. K. 1995. CEO duality and firm performance: A
taking risks, sucb as making investments in inno- contingency model. Strategic Management Jour-
vation. nal, 16: 301-312.
Tbe next steps for research would be to extend Cohen, W. M., & Levinthai, D. A. 1990. Absorptive ca-
2000 Balkin, Markman. and Gomez-Mejia 1127

pacity: A new perspective on learning and innova- tors: A survey. Jaumal of Economic Literature, 28:
tion. Administrative Science Quarterly, 35: 128- 1661-1707.
152. Hamhrick, D. C, & Finkelstein, S. 1987. Managerial dis-
D'Aveni, R. A. 1994. Hypercompetition: Managing the cretion: A bridge between polar views of organiza-
dynamics af strategic maneuvering. New York: tional outcomes. In L. L. Cummings & B. M. Staw
Free Press. (Eds,), Research in organizational behavior, vol. 9:
369-406. Greenwich, CT: lAI Press.
Dechow, P. M., & Sloan, R. C. 1991. Executive incentives
and the horizon problem. Joumal of Accounting Hambrick, D. C, & Finkelstein. S. 1996. The effects of
and Economics, 14: 51-89. ownership structure on conditions at the top: The
case of CEO pay raises. Strategic Management Jour-
Eisenhardt, K. M. 1988. Agency and institutional theory nal, 16: 175-193.
explanations: The case of retail sales compensation.
Academy of Management Journal, 31: 488-511. Hamel, G., & Prahalad, C, K. 1994. Competing for the
future. Harvard Business Review, 72(4): 122-128.
Eisenhardt, K. M. 1989. Making fast strategic decisions in
Hill, C. W., & Phan, P. 1991. CEO tenure as a determinant
high-velocity environments. Academy of Manage-
of CEO pay. Academy of Management Joumal, 34:
ment Journal, 32: 543-576.
707-717.
Finkelstein, S.. & Boyd, B. K. 1998. How much does the
Holthausen, R, W., Larcker, D. F., & Sloan, R. G. 1995,
CEO matter? The role of managerial discretion in the
Business unit innovation and the .structure of exec-
setting of CEO compensation. Academy of Manage-
utive compensation. Journal of Accounting and
ment Journal, 41: 179-199.
Economics, 19: 279-313.
Finkdstein, S.. & D'Aveni, R. A. 1994. CEO duality as a Hoskisson, R, E., Hitt, M. A,, & Hill, C. W. 1991. Mana-
doubled-edged sword: How boards of directors bal- gerial risk taking in diversified firms: An evolution-
ance entrenchment avoidance and unity of com- ary perspective. Organization Science, 2: 296-313.
mand. Academy of Management Joumal, 37:
1079-1108. Hoskisson, R. E., Hitt, M. A.. & Hill, C. W. 1993. Mana-
gerial incentives and investment iu R&D in large
Finkelstein, S., & Hamhrick, D. C. 1996. Top executives multiproduct firms. Organization Science, 4: 325-
and their effects an organizations. Minneapolis- 341.
St. Paul: West.
Hoskisson, R. E,, & Johnson, R. A. 1992. Corporate re-
Garen, J. E. 1994. Executive compensation and principal- structuring aud strategic chauge: The effect on diver-
agent theory. Jaumal of Political Economy, 102: sification strategy and R&D intensity. Strategic
1175-1199. Management Joumal, 13: R25-634.
Gomez-Mejia, L. R. 1994. Executive compensation: A Jelinek, M., & Schoonhoven, C. B. 1993. The innovation
reassessment and a future research agenda, In G. marathon: Lessons from high technology firms.
Ferris (Ed.), Research in personnel and human re- San Francisco; Jossey-Bass.
sources management, vol. 12:161-222. Greenwich,
Jensen, M. C, & Murphy, K. J. 1990. Performance and top
CT: JAI Press. management incentives. Journal of Political Econ-
Comez-Mejia, L. R., & Balkin, D. B. 1989. Effectiveness of omy, 98: 225-264.
individual and aggregate compensation strategies. Koberg, C. S., Rosse, J., & Bergh, D. 1994. Toward a
Industrial Relations, 28: 431-445. definition and typology of high technology firms. In
Gomez-Mejia, L. R., Tosi, H., & Hinkin, T. 1987. Mana- L, R. Comez-Mejia & M. W. Lawless (Eds,). Manage-
gerial control, performance, and executive compen- ment of competitive advantage in high technology:
sation. Academy of Management Journal, 30: 3-26. Greenwich, CT: JAI.
51-70. Lambert, R, A., Larcker, D. F., & Weigelt, K. 1993. The
Comez-Mejia, L. R., & Wiseman, R. 1997. Reframing ex- structure of organizational incentives. Administra-
ecutive compensation: An assessment and outlook. tive Science Quarterly, 38: 438-461.
Joumal of Management, 23: 291-374. Leonard-Barton. D. 1995, We/Zsprings of knowledge:
Grant, R. M. 1991. The resource-based theory of compet- Building and sustaining the sources of innovation.
itive advantage: Implications for strategy formula- Boston: Harvard Business School Press.
tion. California Management Review, 33(3): 115- Maddala, G. S,, Rao, C. R,, & Vinod, H. D. 1993. Econo-
135. metrics. Amsterdam, NY: North-Holland.
Graves, S. B., & Langowitz, N. S. 1993. Innovative pro- O'Reilly, L., Main, B., & Crystal, G. 1988. CEO compen-
ductivity and returns to scale in the pharmaceutical sation as tournament aud social comparison: A tale
industry. Strategic Management Journal, 14: 593- of two theories. Administrative Science Quarterly,
605, 33: 257-274.
Criliches, Z. 1990. Patent statistics as economic indica- Powell, T. C. 1996. How much does industry matter? An
1128 Academy of Management Journal December

alternative empirical test. Strategic Management APPENDIX

Journal, 17: 323-334. List of Companies Used in the Study
Riggs. H. E. 1983. Managing high-technology compa-
nies. Belmont, CA: Lifetime Learning. Higb-Tecbnology Sample Control Sample
(n = 90) {n = 74)
Sanders, W. G., & Carpenter, M. A. 1998. International-
ization and firm governance: The roles of CEO com- Electronics Durable
pensation, top team compensation, and board struc- E-Systems Caterpillar
tnre. Academy of Management Journal, 41: 158- GenCorp Cooper Industries
178. Harsco Deere & Company
Raytheon Emerson Electric
Saviotti, P. P.. & Mani. G. S. 1998. Technological evolu- Rockwell Hubbell
tion, self-organization, and knowledge. Journal af Sequa ., Illinois Too! Works
High Technology Management Research, 9: 255- SimdstTdnd Stanley Works
271. Teledyne Black & Decker
Textron Brunswick
Schilling. M. A., & Hill, C. W. L. 1998. Managing the new Advanced Micro Dana
product development process. Academy of Man- Alltel Genuine Parts
agement Executive, 12(3): 67-81. AT&T : Johnson Controls
AMP Paccar
Standard & Poor's. 1998. Analyst's handbook. New
Apple Computers Premark International
York: Standard & Poor's. AST Research Springs Industries
Tabak, F., & Barr. S. H. 1998. Innovation attributes and Avnet TRW
category membership: Explaining intention to adopt Cabletron Systems Whirlpool
tecbnological innovations in strategic decision mak- Cisco Systems Avon Products
ing contexts. Journal of High Technoiogy Manage- Compaq Computers Colgale-Palmolive
Computer Associates Fruit of ihe Loom
ment Research, 9: 17-34.
Conner Peripherals Gillette
Tosi, H. L., & Gomez-Mejia, L. R. 1989. Tbe decoupling of Coming James River Corporation VA
CEO pay and performance: An agency theory perspec- Dell Computer Mattel
Digital Equipment Nike
tive. Administrative Science Quarterly, 34: 169-189.
Harris Corp Polaroid
Tosi, H. L., & Gomez-Mejia. L. R. 1994. Compensation Hewlett-Packard Protiter & Ciamble
monitoring and firm performance. Academy of IBM Reehok International
Managemeni fournal, 37: 1002-1016. Intel
Service Corporation International
LDDS Communications Shaw Industries
Tosi, H, L. Werner. S.. Katz, J., & Gomez-Mejia. L. R.
Lotus Development Nondiu-able
2000. How much does performance matter: A meta- MCI Communications Amerada Hess
analysis of CEO compensation studies, fournal of Merisel Amoco
Management, 26: 301-339. Microsoft Ashland
Molex Baker Hughes
Tusbman, M. L., & Anderson, P. 1986. Technological
Motorola Burlington Resources
discontinuities and organizational environments.
National Semiconductor Ghe\Ton
Administrative Science Quarterly, 31: 439-465. Novell Gonsol Natural Gas
Werner, S., & Tosi. H. L. 1995. Other people's money: Oracle Systems Diamond Shamrock
Tbe effects of ownersbip on compensation strategy Quantum Dresser Industries
and managerial pay. Academy of Management Raychem Enron
Journal, 38: 1672-1691. Seagate Technology Halliburton
Silicon Graphics Kerr-McGee
Wernerfelt. B. 1984. A resource-based view of the firm. Southern New England Occ:idental Petroleum
Strategic Matiugement Journal, 5: 171-180. Telecommunications Pennzoi.1
Sprint USX Marathan
Westpbal. |. D.. & Zajac, E. J. 1994. Substance and sym- Sun Microsystems Anheuser-Busch Company
bolism in CEOs' long-term incentive plans. Admin- Tandem Computers Archer Daniels
istrative Science Quarterly, 39: 367-390. Tektronix Campbell Soup
Wiseman, R., & Gomez-Mejia, L. R. 1998. A bebavioral Texas Instruments Gbiquita Brands International
Unisys Coca-Cola
agency model nf managerial risk taking. Academy of
Abbott Lahoratories GPC International
Management Review, 23: 133-153.
Health Science Dean Foods
Allargan General Mills
ALZA HJ Heinz
Americem Home International Multifoods
Products PepsiCo
Continued on next page
2000 Balkin, Markman, and Gomez-Mejia 1129

David B. Balkin is a professor of management at tbe

APPENDIX (continued) University of Colorado at Boulder College of Business
and Administration, He received his Pb.D. from the Uni-
Higb-Teclinology Sample Control Sample
versity of Minnesota. His research interests include com-
American Medical Pioneer Hi-Bred International pensation strategy and the management of innovation.
Amgen Quaker Oats
Gideon D. Markman (Ph,D., University of Colorado at
Bausch & Lomb Sara Lee
Baxter International Tyson Foods Boulder) is an assistant professor of entrepreneurship al
Becton Dickinson Universal the Lally Scbool of Management and Tecbnology, Rens-
Bergen Brunswig Wm Wrigley Jr. selaer Polytechnic Institute. His research interests in-
Beverly Enterprises Champion International clude innovation management and technical entrepre-
Bindley Western Crown Cork & Seal neurship.
Biomet Louisiana-Pacific
Bristol-Myers Sonoco Products Luis R. Gomez-Mejia is the Dean's Council of 100 Dis-
Cardinal Distribution Stone Container tinguished Scholar and Professor at Arizona State Uni-
Chiron Temple-Inland versity College of Business. He received bis Pb.D. from
Forest Labs Westvaco the University of Minnesota. His research interests are
Genelech Williamette Industries macro compensation issues, including executive com-
HnalthTrust Engelhard
Hillenbrand Industries Freeport-McMoRan
pensation and compensation strategy.
Humana LTV
IVAX Nucor
fohnson & [obnson Phelps Dodge
Manor Care
Marion Merrell Dow
McKesson
Medtronic
Merck
National Intergroup
National Medical
PacifiCare Health
Pall
Pfizer
Rhone-Poulenc Rorer
Schering-Plough
Stryker
United Healthcare
US Healthcare
tlS Surgical
Warner- Lambert
VVcillPdinl Heallh