You are on page 1of 21

Wealth creation and managerial pay: MVA and EVA

as determinants of executive compensation


Ali Fatemi
a
, Anand S. Desai
b
, Jeffrey P. Katz
b,
*
a
DePaul University, Chicago, IL, USA
b
Kansas State University, Manhattan, KS, USA
Received 1 June 2001; received in revised form 1 October 2002; accepted 1 October 2002
Abstract
Designing effective compensation contracts has become increasingly complex due to the
globalization of the executive work force and the multitude of incentive schemes. We examine the
relationships between managerial pay and firm performance among domestic and global firms using
economic value added (EVA) and market value added (MVA) to assess wealth creation. Our work
suggests that top managers in domestic- and globally focused firms are not only incented to increase
EVA, but also rewarded for past additions to MVA. The results of our research suggest that managers
of highly globalized firms tend to be paid at higher levels, reflecting the increased complexity of
managing global firms.
D 2003 Elsevier Science Inc. All rights reserved.
JEL classification: J33; G3
Keywords: Compensation; Pay for performance; Economic value added; Market value added
1. Introduction
Previous studies have proposed that optimal executive compensation contracts perfectly
align the interests of the executives with those of the firms shareholders (Grossman &
Hart, 1983; Harris & Raviv, 1979). In theory, such contracts act as incentive mechanisms
for executives to engage in behaviors that maximize the firms value and reward
executives for such behavior (Fama, 1980; Jensen & Meckling, 1976). Whether executive
compensation contracts meet this test of optimality, ex ante or ex post, is an empirical
question subject to ongoing investigation (Tosi, Werner, Katz, & Gomez-Mejia, 2000).
1044-0283/03/$ - see front matter D 2003 Elsevier Science Inc. All rights reserved.
doi:10.1016/S1044-0283(03)00010-3
* Corresponding author. Tel.: +1-785-532-7451; fax: +1-785-532-7024.
E-mail address: jkatz@ksu.edu (J.P. Katz).
Global Finance Journal 14 (2003) 159179
Several studies have examined the relationships between measures of firm performance
and top manager pay. For example, Murphy (1985) found a statistically significant
relationship between the level of pay and performance, while Mehran (1995) found firm
performance is positively related to managements ownership stake and to the percentage
of its equity-based compensation. However, Jensen and Murphy (1990) did not find a
significant relationship between changes in firm value and changes in executive compen-
sation. Miller (1995) showed no support for a linear relationship between pay and
performance, but found strong support for a convex relationship. Hadlock and Lumer
(1997) found that payperformance sensitivities have significantly increased over time for
small firms, but not for large firms.
More recently, in a study examining the role of boards in setting managerial pay, Porac,
Wade, and Pollock (1999) found evidence that boards make comparisons within and
between industries in which the firm competes to support their top management
compensation decisions. The authors conclude that boards of directors tend to anchor
their comparability judgments by examining other firms performance. This suggests that
top manager performance is assessed based on relative measures and with an eye toward
the industry environment affecting the firm.
Unfortunately, most of the studies exploring the nature of the relationship between
managerial pay and performance have used accounting-based measures of performance
(such as return on equity [ROE] or return on assets [ROA]). Such measures may bear little
resemblance with the economic return earned by the firm since accounting-based measures
do not account for the risk incurred by the firms managers in their search for growth and
profitability (Shiely, 1996). For example, earnings growth which may follow a decision to
increase the size of the firm does not automatically lead to a per-share growth in firm value
because the former may be achieved at excessive capital costs (Copeland, Koller, &
Murrin, 1995). In addition, even studies using measures of performance based on market
returns fail to adjust returns for the level of risk exposure (Harris & Raviv, 1979). Thus,
the exact relationship between pay and performance can be somewhat different than what
the empirical results suggest because the impact of risk is not adequately accounted for in
commonly employed measures of performance (Lehn & Makhija, 1996; Stewart, 1991).
Our study is designed to further clarify the nature of the payperformance relationship
by adding risk to the equation. Specifically, we seek to investigate the relationship between
top management compensation and two measures of risk-adjusted firm performance:
economic value added (EVA) and market value added (MVA). EVA and MVA are
measures developed and trademarked by the Stern Stewart and Co. First suggested by
Stewart (1991), EVA can be thought of as a proxy for the measurement of economic
returns. It is the firms residual profitability in excess of capital costs. A firms EVA is
positive when after-tax operating profits exceed the dollar cost of capital (COC). MVA is a
closely related measure in that it is the present value of all expected future EVA and can be
thought of as the net present value of the firm.
Variations of these measures have been proposed, and used, by others (Copeland et al.,
1995; Rappaport, 1986). However, EVA and MVA have received wider attention both in
the corporate world and in scholarly research (see for example, Hodak, 1994; Lehn &
Makhija, 1996; Spinner, 1995; Tully, 1993; Uyemura, Kantor, & Pettit, 1996). Included
among these are studies that have attempted to document the presence (or lack thereof) of
A. Fatemi et al. / Global Finance Journal 14 (2003) 159179 160
a relationship between EVA and measures of stock price performance. Dodd and Chen
(1996), for example, report that EVA explains only slightly more than one fifth (20.2%) of
the variation in stock returns for a sample of 566 firms, comparing unfavorably with
ROA, which explains almost a fourth (24.5%) of the variation. Further, Dodd and Johns
(1999) found some differences in performance between the adopters and nonadopters of
EVA. However, as documented by Weaver (2001), there are significant differences in how
firms measure EVA. Examining a set of 29 EVA adopters who participated in his survey,
he found that none of the respondents measured the EVA the same way. Directly relevant
to our purpose is Weavers finding that the adopters top reasons for implementation of
EVA are to enhance financial management and to enhance compensation metrics.
Also of direct interest to this work is Kramer and Peters (2001) finding that, as a proxy
for MVA, EVA does not suffer from any industry-specific bias. However, they also
conclude that EVA is consistently outperformed by the net operating profit after tax
measure.
We believe EVA and MVA are reasonable proxies for the measurement of owner wealth
maximization while taking into account the relative risk-based costs of doing so (Hodak,
1994; Shiely, 1996). However, the relationships between executive compensation and
these measures of firm performance have not yet been explored empirically. In this study,
we seek to examine these relationships. Under the pay-for-performance hypothesis, we
expect a positive relationship between executive compensation and firm performance. In
this study, firm performance is measured in the context of value creation for the owners of
the firm using MVA and EVA. We also examine the contribution of these measures in
explaining the cross-sectional variation in executive compensation relative to the account-
ing-based measure of ROA.
Executive compensation generally consists of several components such as salary,
bonus, stock options, and long-term incentive payments. It is plausible that certain
components, such as bonuses, are used as a reward for past performance, while other
components related to firm value are designed to provide the correct incentive for future
performance (Murphy, 1985). The complex design of the total compensation package
requires that we separately examine the relationship between firm performance and each of
these components.
Recently, there has been an increase in the body of research pointing to the global
managerial labor market as the basis for better understanding differences in levels of pay,
as well as the mix of incentive plan components. That is, there is increasing evidence that
top managers in highly global firms have higher proportions of performance-based pay in
their total compensation contract (Carpenter, Sanders, & Gregersen, 2001). In addition,
Richard (2000) suggests that pay policies that include equity participation practices
pioneered in the United States are becoming common throughout the global business
community. We hope to assess whether there are differences in the relationships between
pay and performance based on the level of international impact of the firm.
We also seek to examine the causal order of the relationshipthat is, whether
compensation serves as a reward for past performance, or as an incentive for enhanced
future performance. Given shareholder wealth maximization as the goal of the firm, is the
executive compensation scheme used by the firm incentive-compatible? To answer this
question, we use leading and lagged values of firm performance. If compensation is an
A. Fatemi et al. / Global Finance Journal 14 (2003) 159179 161
incentive for top managers to perform in certain ways, then we would expect top manager
pay to be unrelated to past performance, but instead would observe a positive relationship
between current compensation and future performance. On the other hand, if compensa-
tion is a reward for superior performance, we would expect top manager pay to be
positively related to past performance, but not to future performance (Gomez-Mejia, Tosi,
& Hinkin, 1987; Tosi et al., 2000).
The rest of our paper is organized as follows. In the next section, we describe our data.
We present and discuss our results in Section 3 and our conclusions and implications for
further research are presented in the last section.
2. Data
Executive compensation and firm performance measures used in this study are obtained
from three sources: the Standard and Poors ExecuComp database, Stern Stewart and Co.s
Performance 1000 database, and Standard and Poors Compustat database.
We define top managers as individuals with the title of Chairman, CEO, President and
senior-level Vice President. Compensation data are obtained from the ExecuComp data-
base. In our study, we use three measures of compensation: salary, bonus, and total direct
compensation (TDC). TDC is defined as the sum of salary, bonus, value of restricted stock
granted, value of stock options granted, and other annual items, which include perquisites,
payments to cover executives taxes, preferential earnings payable but deferred at
executives election, and preferential discounts of stock purchases. Data requirements
for the incentive/reward hypothesis require us to use annual compensation in the 4-year
period from 1992 to 1995.
The number of executives varies across the firms in our sample, ranging from 1 to 12.
Larger firms tend to have a greater number of executives. Since we are interested in the
total executive compensation package of the firm, for each firm (and for each year) in the
sample period, we aggregate the compensation component for all executives of this firm
listed in the ExecuComp database. Implicit in this aggregation is the assumption that the
firms compensation policy applies uniformly to top executives of different ranks. Prior
research has shown that this hierarchical assumption is reasonable (Demski & Sappington,
1989; Werner & Tosi, 1995).
Firm performance measures are obtained from two sources. EVA and MVA are
obtained from the Performance 1000 database. ROE and ROA are obtained from the
Compustat database.
Economic theory, human capital theory, and agency theory suggest that top manager
pay will be positively related to firm size (Agarwal, 1981; Becker, 1964; Deckop, 1988;
Jensen & Meckling, 1976). That is, economic theory of marginal revenue products
predicts greater pay in firms of greater size (Gomez-Mejia et al., 1987). Additionally, if
an effective CEO can create greater profits for large firms than for small ones, then it is
likely that the marginal productivity of the CEO would vary directly with size. Further,
human capital theory predicts greater pay in firms of greater size (Becker, 1964). The
prediction follows from the observation that the top position in a larger firm requires
greater human capital because a larger firm is more complex, more difficult to manage,
A. Fatemi et al. / Global Finance Journal 14 (2003) 159179 162
and demands more responsibility than in a smaller firm. To control for firm-size effects,
we standardize our variables using the total capital employed by the firm, as explained
below. Total capital (CAPITAL) is obtained from the Performance 1000 database.
Jensen and Meckling (1976) propose that executive compensation is related to the
firms risk. The higher the risk borne by the firm, and thus, the greater the likelihood of
performance variance, the greater the compensation risks. In an efficient managerial labor
market, executives will demand, and receive, compensation for bearing risk, thus leading
to a positive relationship between compensation and the risk borne by the firm (Fama,
1980; Harris & Raviv, 1979; Winn & Shoenhair, 1988).
More recently, it has been suggested that top manager pay, in part, reflects risks borne
by the firm (Hadlock & Lumer, 1997; Hall & Liebman, 1998; Kroll, Simmons, & Wright,
1990). To facilitate the efficient measurement of risk, the firms COC is used as a proxy for
risk. Managers choosing competitive strategies exposing the firm to greater risk will be
charged higher costs by investors for the risk through higher capital costs. It is not
surprising that all asset-pricing models predict an upward-sloping relationship between the
level of risk borne by the firm and its COC (Lehn & Makhija, 1996; Uyemura et al., 1996).
In our cross-sectional tests, we control for firm risk differences using the COC as a proxy
for risk. While equity beta coefficients are often used as a measure of risk, they only
measure the risk borne by the shareholders. Since we are concerned with the overall risk of
the firm, we need to incorporate the risk borne by other security holders in the firm as well.
The weighted-average COC can therefore serve as a reasonable proxy for the firms risk.
We obtain COC estimates from the Performance 1000 database.
The extent of a firms global presence is measured by overseas (foreign) sales, as a
proportion of the firms total sales, and is denoted by FSALES. Both total and overseas
sales are obtained from the Compustat database.
In combining the samples of firms obtained from the compensation and the
performance databases, we require that data be available for all variables to be used
in tests of our hypothesis. This results in a final sample size of 1965 observations,
where each observation is a firm-year. The numbers of firms in each year from 1992 to
1995 are 432, 550, 502, and 481, respectively. Overseas sales were only available for
119 firms, and only for the 1995 year. Hence, our tests of the pay-for-performance
hypotheses that account for the global nature of the firms operations are limited to this
subsample.
Since EVA, MVA, and compensation measures are measured in dollars and therefore
related to firm size, we control for firm-size differences in our cross-sectional tests. Firm
size is measured by the capital employed at the beginning of the year (BOYCAP),
obtained from the Performance 1000 database. Using this measure of firm size, salary is
redefined as salary/BOYCAP. Other compensation measures are adjusted for firm size
similarly. We also define the firm-size standardized EVA (denoted as SEVA) as EVA/
BOYCAP. Since MVA is the total market value added to the stock of capital, it is a
cumulative measure. The incremental market value added during year t, after adjustment
for differences in firm sizes, is defined as:
DMVA
t

MVA
t
MVA
t1
BOYCAP
t
1
A. Fatemi et al. / Global Finance Journal 14 (2003) 159179 163
Table 1 provides descriptive statistics for all variables used in our cross-sectional tests.
On average, salary constitutes about 46% of TDC, and bonus constitutes about 22% of
TDC. Further, the usual accounting measures of performance (ROA and ROE) are positive
on average. The average EVA is, however, $95.74 million, while the average DMVA is
about $3.6 billion. Since EVA is defined as the product of the capital employed and
difference between the return on capital and the COC, a negative EVA implies that the
return on capital is less than the COC on average. In our sample, the average return on
capital, obtained from the Performance 1000 database, is 10.95%. This is less than the
average COC of 11.68%.
It is tempting to conclude, on the basis of the negative average EVA, that our sample
firms have been engaged in suboptimal decision-making with respect to resource
allocation. However, the positive incremental market value added (DMVA) suggests a
different implication. For strategic reasons, firms may commit resources to investments
with expected payoffs in the distant future. Investments in these real options are
valuable, and their value is reflected in a higher market value of the firm. Thus, even
though the EVA is negative, the market views these investments favorably and
consequently revalues the firm upwards. Of the 1965 observations in our sample,
1247 (63.5%) had a negative EVA. However, 60.5% of these negative EVA firms had
positive increments to their market value. Interestingly, this proportion of positive
DMVA firms is comparable to the proportion of positive DMVA firms in the total
sample (60.7%).
In Table 2, we provide simple correlation coefficients between the variables used in this
study. Panel A reports correlation coefficients between our measure of firm size
Table 1
Descriptive statistics on measures of compensation, risk, and firm performance for the sample of 1965
observations between 1992 and 1995
a
Variable Mean Standard deviation
Salary
b
409.39 216.46
Bonus
b
272.48 333.47
TDC
b
1609.38 3311.88
(Salary/TDC)
c
46.35 24.31
(Bonus/TDC)
c
22.22 16.01
CAPITAL
d
4736.95 10 323.29
COC
c
11.68 2.33
ROE
c
12.77 86.12
ROA
c
5.13 7.48
EVA
d
95.74 652.40
MVA
d
3599.90 9028.39
SEVA
c
0.65 9.20
DMVA
c
29.77 162.47
FSALES
c
14.05 13.35
a
Statistics for FSALES are based on a sample of 119 firms for 1995.
b
In thousands of dollars.
c
In percent.
d
In millions of dollars.
A. Fatemi et al. / Global Finance Journal 14 (2003) 159179 164
(CAPITAL) and the three compensation measures (unadjusted for firm-size differences).
All correlation coefficients are positive and significantly different from zero, indicating
that executives of large firms receive greater compensation than do executives of smaller
firms.
Panel B shows the correlation coefficients between the size-adjusted compensation
measures and the risk of the firm as measured by the COC. All correlation coefficients are
positive and significant at the 1% level. These coefficients indicate a strong positive
relationship between firm risk and executive compensation. In subsequent tests of the
payperformance relationship, we control for both firm size (by normalizing all dollar-
denominated measures using firm size) and for risk (by including our proxy for risk as an
explanatory variable).
In Panel B, we also report the correlation coefficients between measures of performance
and the size-adjusted compensation components. Generally, there is a strong indication of
a positive relation between firm performance and compensation. While compensation is
not correlated with ROE, the strong positive correlation between compensation and ROA
may suggest that compensation policy is more a function of enterprise performance
rather than simple return to equity holders.
3. Results
We investigate the relationship between executive compensation and firm perform-
ance using cross-sectional regression analysis. The dependent variables in all our
cross-sectional regressions are the three components of compensation (salary, bonus
and TDC) adjusted for firm size. For brevity, we refer to compensation components
generically as COMP. In all regressions, the t statistics are based on Whites (1980)
heteroskedasticity-consistent estimators of the standard errors of the models parame-
ters.
Table 2
Correlation coefficients between selected measures of compensation and firm size, risk, and performance for the
sample of 1965 observations between 1992 and 1995
a
Salary Bonus TDC
Panel A
CAPITAL 0.436*** 0.252*** 0.295***
Panel B
COC 0.211*** 0.250*** 0.224***
ROA 0.132*** 0.258*** 0.156***
ROE 0.012 0.030 0.015
SEVA 0.118 0.193*** 0.101***
DMVA 0.235*** 0.228*** 0.300***
***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.
a
Compensation variables used to compute the correlation coefficients with Capital are not adjusted for
differences in firm size. In all other computations in the table, compensation variables are scaled by the capital
employed at the beginning of the year.
A. Fatemi et al. / Global Finance Journal 14 (2003) 159179 165
3.1. Determinants of executive compensation
Our first test examines the relationship between the components of compensation and
each of the four measures of firm performance (ROE, ROA, SEVA, and DMVA) in turn,
while simultaneously controlling for firm risk. Thus, we estimate the following model:
COMP
i
a
0
a
1
PERF
i
a
2
COC
i
e
i
2
where the performance variable (PERF) is ROE, ROA, SEVA or DMVA. Estimates of this
models parameters are reported in Table 3.
Regardless of the compensation measure and performance measure used in the models
estimation, the estimates of a
2
are all positive and statistically significant at the 1% level.
This positive relationship between compensation and firm risk is consistent with the
arguments presented by Fama (1980), Harris and Raviv (1979), and Winn and Shoenhair
(1988) that, in an efficient managerial labor market, executives demand and receive
compensation for bearing risk.
After we control for firm size and risk, we find a statistically significant relationship
between all three measures of executive compensation and the change in the market value
(DMVA) and ROA. The estimates of a
1
are positive and statistically significant in all of
Table 3
OLS estimates of the relation between compensation and firm performance after controlling for risk, for the
sample of 1965 observations between 1992 and 1995
a
Compensation
measure
a
0
a
1
a
2
F statistic
( P value)
Adjusted
R
2
Panel A: COMP
i
=a
0
+a
1
DMVA
i
+a
2
COC
i
+e
i
Salary 0.397*** (4.73) 0.110*** (3.71) 6.652*** (8.65) 97.15 ( < .01) .09
Bonus 0.467*** (6.83) 0.077*** (3.03) 5.986*** (9.24) 112.86 ( < .01) .11
TDC 6.201*** (5.94) 1.217*** (3.48) 66.470*** (6.87) 150.92 ( <.01) .13
Panel B: COMP
i
=a
0
+a
1
SEVA
i
+a
2
COC
i
+e
i
Salary 0.491*** (3.33) 0.275 (0.26) 7.720*** (5.76) 46.52 ( < .01) .04
Bonus 0.407*** (4.71) 0.954* (1.77) 5.715*** (7.23) 87.60 ( < .01) .08
TDC 5.511*** (5.46) 12.678** (2.14) 64.374*** (6.64) 90.14 ( < .01) .08
Panel C: COMP
i
=a
0
+a
1
ROE
i
+a
2
COC
i
+e
i
Salary 0.461*** (5.27) 0.001 (0.24) 7.474*** (9.02) 45.57 ( < .01) .04
Bonus 0.511*** (7.18) 0.013 (1.35) 6.538*** (9.52) 65.75 ( < .01) .06
TDC 6.904*** (5.71) 0.055 (1.12) 75.527*** (6.58) 62.10 ( < .01) .06
Panel D: COMP
i
=a
0
+a
1
ROA
i
+a
2
COC
i
+e
i
Salary 0.400*** (4.28) 0.582*** (2.72) 6.692*** (7.34) 48.02 ( < .01) .05
Bonus 0.351*** (4.77) 1.518*** (4.38) 4.515*** (6.17) 97.93 ( < .01) .09
TDC 5.553*** (5.39) 12.757*** (3.80) 58.427*** (6.21) 78.17 ( < .01) .07
***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.
a
COMP
i
is the measure of executive compensation, TDC is total direct compensation, COC
i
is the cost of
capital, DMVA
i
is the change in market value added, SEVA
i
is the standardized economic value added, ROE
i
is
the return on equity, and ROA
i
is the return on assets for firm i. t statistics are in parenthesis.
A. Fatemi et al. / Global Finance Journal 14 (2003) 159179 166
these cases. However, we find no evidence of a relationship between compensation and
ROE, indicating that executive compensation is more a function of enterprise performance
than the return to equity holders. Further, we find only a weak relationship between
compensation and economic value added (SEVA). While bonus and TDC are related to
SEVA at the 10% and 5% levels, respectively, the relationship between salary and SEVA is
not significant.
Taken together, the results presented in Table 3 indicate that there is a positive
relationship between executive compensation and broad measures of firm performance
(ROA and MVA), after controlling for differences in firm size and risk. However, it can be
argued that a firms competitive position, and hence, its performance, may be a function of
the extent of its global operations. In a recent survey of U.S. multinational companies only
6% of the firms did not provide some type of long-term incentive plan (Freedman, 2000).
However, with higher levels of performance-based pay reportedly being awarded to global
managers, it is unclear whether differences in performance of global firms will result in
disparate changes in the pay of global versus domestic managers (Platt, 2002). Clearly, the
level of international sales has been suggested to have an effect on the components of
executive compensation.
To examine the effect of the firms global nature on the pay-for-performance relation-
ship, we estimate the following model on the subsample of 119 firms for which we were
able to obtain the ratio of overseas sales to total sales:
COMP
i
a
0
a
1
PERF
i
a
2
COC
i
a
3
FSALES
i
e
i
3
where FSALES is the ratio of overseas sales to total sales. Estimates of this models
parameters are presented in Table 4.
From Table 4, we find that even after controlling for the extent of the firms global
nature, there is a significant positive relationship between executive compensation and
firm performance. Our estimates of the coefficient on performance (a
1
) are all positive and
significant except in the case where ROE measures firm performance. Only bonus is
positively related to ROE in Table 4, whereas salary and TDC are not. We also find that all
measures of compensation are significantly positively related to EVA for these 119 firms,
indicating that EVA is an important determinant of compensation for global firms.
The model in Eq. (3) also allows us to address the issue of whether executive
compensation is related to the extent of the firms global nature. From Table 4, the
coefficients on FSALES are significantly different from zero when ROE measures
performance. Further, when performance is measured by DMVA or ROA, there is a
significant positive relationship only between TDC and FSALES. When performance is
measured by SEVA, the positive relationship is observed for salary and bonus, but not
TDC. These results suggest that TDC is higher for executives of firms with a greater global
presence, but that this relationship is also dependent on how firm performance is
measured.
A number of previous studies have discussed the relationship between executive
compensation and ROA (see for example, Agarwal, 1981; Deckop, 1988; Kroll et al.,
1990; Leonard, 1990; Pavlik & Belkaoui, 1991; Winn & Shoenhair, 1988). Although EVA
and MVA are measures of performance more indicative of contribution to shareholder
wealth, it is not clear whether these broader measures are better predictors of the cross-
A. Fatemi et al. / Global Finance Journal 14 (2003) 159179 167
sectional variation in executive compensation. To investigate this issue, we estimate the
following model:
COMP
i
a
0
a
1
COC
i
a
2
ROA
i
a
3
PERF
i
e
i
4
where PERF is either SEVA, or DMVA.
The estimates of this model for the total sample of 1965 observations are presented in
Table 5. With PERF defined as DMVA, and for all components of compensation, the
coefficients of COC, ROA, and DMVA are positive and significant at the 1% level.
Table 4
OLS estimates of the relation between compensation and firm performance after controlling for risk, for the
sample of 119 observations in 1995
a
Compensation
measure
a
0
a
1
a
2
a
3
F statistic
( P value)
Adjusted
R
2
Panel A: COMP
i
=a
0
+a
1
DMVA
i
+a
2
COC
i
+a
3
FSALES
i
+e
i
Salary 0.045
(0.42)
0.051**
(2.30)
2.042**
(2.04)
0.298
(1.19)
13.16
( <.01)
.24
Bonus 0.153
(1.39)
0.040*
(1.69)
2.304**
(2.39)
0.401 (1.18) 6.80
( <.01)
.13
TDC 1.379
(1.37)
0.647***
(7.28)
12.704
(1.48)
4.215***
(2.91)
28.70
( <.01)
.41
Panel B: COMP
i
=a
0
+a
1
SEVA
i
+a
2
COC
i
+a
3
FSALES
i
+e
i
Salary 0.047
(0.37)
0.722***
(4.36)
2.061*
(1.89)
0.344*
(1.88)
12.43
( <.01)
.23
Bonus 0.037
(0.26)
1.234***
(6.09)
1.134
(0.91)
0.400*
(1.91)
17.60
( <.01)
.30
TDC 1.607**
(2.03)
8.697***
(3.76)
14.908**
(2.05)
4.856
(1.61)
21.36
( <.01)
.34
Panel C: COMP
i
=a
0
+a
1
ROE
i
+a
2
COC
i
+a
3
FSALES
i
+e
i
Salary 0.190
(1.43)
0.086
(1.19)
3.348***
(2.98)
0.386*
(1.96)
5.78
( <.01)
.11
Bonus 0.264
(1.65)
0.172**
(1.98)
3.131**
(2.31)
0.465*
(1.96)
5.40
( <.01)
.10
TDC 3.222***
(2.74)
0.370
(0.58)
30.519***
(3.07)
5.353***
(3.08)
7.69
( <.01)
.15
Panel D: COMP
i
=a
0
+a
1
ROA
i
+a
2
COC
i
+a
3
FSALES
i
+e
i
Salary 0.750
(0.56)
1.250***
(3.17)
1.802
(1.50)
0.300
(1.56)
9.04
( <.01)
.17
Bonus 0.049
(0.41)
2.347**
(2.29)
0.248
(0.17)
0.304
(1.02)
13.88
( <.01)
.25
TDC 1.958*
(1.71)
13.663***
(4.02)
12.570
(1.21)
4.387***
(2.66)
14.01
( <.01)
.25
***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.
a
COMP
i
is the measure of executive compensation, TDC is total direct compensation, COC
i
is the cost of
capital, DMVA
i
is the change in market value added, SEVA
i
is the standardized economic value added, ROE
i
is
the return on equity, ROA
i
is the return on assets for firm i, and FSALES
i
is the ratio of overseas sales to total
sales. t statistics are in parenthesis.
A. Fatemi et al. / Global Finance Journal 14 (2003) 159179 168
Moreover, the adjusted R
2
values fall in the range .09.14. When only COC and ROA are
used as independent variables, the adjusted R
2
values are in the range .05.09 (see Panel
D, Table 3). Thus, it can be concluded that MVA is indeed a significant predictor of cross-
sectional variations in executive compensation and provides additional information in
explaining the cross-sectional variation in executive compensation.
In the second panel of Table 5, PERF is defined as SEVA. Regardless of how executive
compensation is measured, the firm-size adjusted EVA has no additional explanatory
power. In fact, the explanatory power of ROA is also virtually eliminated. We suspect that
this is due to the high correlation between ROA and SEVA (q=.66). The high degree of
multicolinearity makes interpretation of the significance levels of the models parameters
difficult. In the last panel of Table 5, performance is defined to encompass all three
measures of performance (ROA, EVA, and MVA). Once again, while the coefficients on
SEVA are insignificantly different from zero, those on MVA are significant at the 1%
level. Since MVA is a measure of the increase in shareholder wealth, our results indicate
that cross-sectional variation in executive compensation can best be explained by the
extent to which their actions increase shareholder wealth.
To further examine the relationship between compensation and firm performance in the
context of the firms global position, we replicate the analysis presented in Table 5 for the
Table 5
Tests for the relative explanatory power of performance measures for the sample of 1965 observations between
1992 and 1995
a
Compensation
measure
Intercept COC ROA SEVA DMVA F statistic
( P value)
Adjusted
R
2
Salary 0.345***
(3.87)
5.988***
(7.08)
0.496**
(2.41)
0.109***
(3.68)
66.04
( < .01)
.09
Bonus 0.314***
(4.27)
4.034***
(5.47)
1.459***
(4.44)
0.074***
(3.07)
97.05
( < .01)
.13
TDC 4.956***
(5.35)
50.666***
(6.02)
11.813***
(3.91)
1.201***
(3.52)
111.39
( < .01)
.14
Salary 0.418***
(4.08)
6.448***
(6.74)
1.388
(0.95)
0.949
(0.54)
36.48
( < .01)
.05
Bonus 0.343***
(4.54)
4.609***
(6.38)
1.206*
(1.85)
0.367
(0.47)
66.60
( < .01)
.09
TDC 5.332***
(5.48)
61.258***
(6.13)
3.399
(0.56)
11.025
(1.31)
60.53
( < .01)
.08
Salary 0.372***
(3.94)
5.222***
(4.66)
2.297
(1.34)
2.152
(1.03)
0.142***
(2.59)
66.50
( < .01)
.12
Bonus 0.317***
(4.26)
3.926***
(4.96)
1.713**
(2.22)
0.303
(0.34)
0.079**
(2.57)
73.40
( < .01)
.13
TDC 4.943***
(5.40)
51.013***
(5.61)
10.996*
(1.74)
0.976
(0.12)
1.186***
(3.32)
83.55
( < .01)
.14
***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.
a
TDC is total direct compensation, COC
i
is the cost of capital, DMVA
i
is the change in market value added,
SEVA
i
is the standardized economic value added, and ROA
i
is the return on assets for firm i. t statistics are in
parenthesis.
A. Fatemi et al. / Global Finance Journal 14 (2003) 159179 169
sample of 119 observations in 1995 for which we were able to obtain data on the ratio of
overseas sales to total sales. In Table 6, we report our estimates of the following model:
COMP
i
a
0
a
1
COC
i
a
2
ROA
i
a
3
PERF
i
a
4
FSALES
i
e
i
5
In the first panel, firm performance is measured by MVA. Our estimates of a
2
and a
3
are
all positive and significantly different from zero when firm performance is measured using
DMVA. These results are similar to those reported in Table 5 for the full sample: MVA is a
significant predictor of cross-sectional variation in executive compensation and provides
additional explanatory power beyond the traditional accounting measure of performance
(ROA). Moreover, none of our estimates of a
4
(the coefficient on FSALES) are significant
in this case. Thus, the pay-for-performance relationship is not dependent on the extent of
the firms global position if performance is measured by the MVA.
In the second panel of Table 6, we find that when firm performance is measured by the
size-adjusted economic value added (SEVA) in addition to ROA, compensation is
significantly positively related to the former. Thus, unlike the results presented in Table
5, where we found no significant relationship between compensation and SEVA, a positive
relationship is observed when we control for the extent of the firms global nature. Finally,
in the last panel of Table 6, where firm performance is measured by ROA, SEVA, and
Table 6
Tests for the relative explanatory power of performance measures for the sample of 119 observations in 1995
a
Compensation
measure
Intercept COC ROA SEVA DMVA FSALES F statistic
( P value)
Adjusted
R
2
Salary 0.502
(0.41)
0.590
(0.51)
1.120***
(3.05)
0.049***
(4.48)
0.222
(1.25)
12.92
( < .01)
.29
Bonus 0.041
(0.27)
0.618
( 0.45)
2.254***
(5.10)
0.035***
(2.66)
0.249
(1.16)
12.72
( < .01)
.28
TDC 0.342
( 0.47)
2.859
( 0.30)
12.004*
(1.91)
0.619**
(2.49)
3.402
(1.42)
29.32
( < .01)
.49
Salary 0.025
( 0.20)
1.634
(1.40)
0.482
(1.06)
0.649***
(3.06)
0.317*
(1.71)
9.61
( < .01)
.23
Bonus 0.020
(0.14)
0.011
(0.01)
1.269**
(2.48)
0.911***
(3.84)
0.328
(1.58)
15.34
( < .01)
.33
TDC 1.388
( 1.28)
10.634
(1.10)
4.830
(1.27)
7.463***
(4.23)
4.852***
(2.97)
16.51
( < .01)
.34
Salary 0.051
(0.40)
0.653
(0.56)
1.006**
(2.14)
0.105
(0.39)
0.045***
(3.16)
0.231
(1.28)
10.29
( < .01)
.28
Bonus 0.030
(0.20)
0.114
( 0.08)
1.335**
(2.43)
0.842***
(2.70)
0.006
(0.34)
0.317
(1.51)
12.20
( < .01)
.32
TDC 0.342
( 0.47)
2.876
( 0.29)
12.036*
(1.78)
0.029
(0.01)
0.620***
(2.10)
3.400
(1.49)
23.25
( < .01)
.49
***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively.
a
TDC is total direct compensation, COC
i
is the cost of capital, DMVA
i
is the change in market value added,
SEVA
i
is the standardized economic value added, ROA
i
is the return on assets for firm i, and FSALES is the ratio
of overseas to total sales. t statistics are in parenthesis.
A. Fatemi et al. / Global Finance Journal 14 (2003) 159179 170
DMVA, we find that MVA is, in general, a stronger determinant of the cross-sectional
variation in salary and TDC, but not bonus.
3.2. Compensation and decision-making
The evidence presented above suggests that executive compensation is positively
related to MVA and to a lesser extent to EVA. Since MVA is the present value of future
expected EVA, and since the relationship between MVA and compensation is strong, why
then is the relationship between compensation and EVA weaker? To explore this issue
further, we sort our sample into groups based on SEVA and DMVA. As argued below, this
classification of the sample enables us to distinguish firms based on their growth
opportunities and economic returns earned.
All firms in the sample are ranked independently based on these two variables. The
sample is then divided into three groups of equal size based on SEVA values. A similar
partitioning of the sample into three equal-sized groups is carried out using DMVAvalues.
Combinations of rankings based on SEVAand DMVAthus yields nine subsamples. We then
analyze the following four subsamples further: high SEVA/high DMVA; high SEVA/low
DMVA; low SEVA/high DMVA; low SEVA/low DMVA. We denote these four subsamples
as HH, HL, LH, and LL, respectively (the first letter in the subsample designation refers to
the SEVA ranking, while the second letter refers to the DMVA ranking).
We view these four subsamples in the following manner. The HH subsample consists of
firms that generate high economic returns with a high market value; firms which may be
positioned to enjoy a prolonged period of economic rents. The HL subsample firms
generate high economic return with a relative low market value; firms with limited growth
prospects and low expected economic profits (an alternative interpretation is that these
firms are window dressing their current income). The LL subsample represents firms
whose current and future ability to generate economic profits is severely limited. Finally,
the LH subsample represents firms with low current profits making investments in
valuable real options.
Table 7 provides descriptive statistics on the components of executive compensation for
these four subsamples. Compensation variables have been normalized by beginning-of-
year capital to account for firm-size differences. The most striking result from this table is
the difference in compensation levels between the HH and LL subsamples. For example,
the average salary of executives in the HH subsample is 0.74, while that for executives in
the LL subsample is 0.32. Similar differences are observed for all other components of
salary. All of the differences in compensation between these two subsamples are
significant at the 1% level.
These differences in compensation levels suggests that, on average, the compensation
policy of these firms is rational: executives that generate high economic and market value
for the firm are compensated significantly higher than executives that do not.
We next examine the relationship between compensation, EVA and MVA for each of
the four subsamples. The first four panels of Table 8 report our estimates of the model in
Eq. (5) using SEVA as the measure of firm performance. The next four panels contain the
results when DMVA is used as the measure of performance. Results reported in the first
two panels, the HH and HL subsamples, indicate that the estimates of the coefficients on
A. Fatemi et al. / Global Finance Journal 14 (2003) 159179 171
SEVA are positive and generally significant. Thus, for firms that generate high economic
profits, EVA is a significant determinant of compensation.
The next panel contains results of the regressions for the low SEVA/high DMVA
subsample, i.e., firms making large investments in valuable options with long-term
payoffs. For these managers, compensation is negatively correlated with EVA. The
estimates of the coefficients on SEVA are all negative and significantly so. It appears
that foregoing opportunities to report higher current-period income results in a compen-
sation penalty. Moreover, the adjusted R
2
values are the highest of all for subsample
regressions, ranging between .30 and .47. Our results for the LL subsample, reported in the
fourth panel of Table 8, further indicate that SEVA is not a significant determinant of pay
for firms limited current or future growth possibilities.
The next four panels report the estimates of the equation where DMVA is used as a
measure of performance. The results indicate that this measure of performance is, in
general, positively correlated with compensation, especially for the subsamples where the
MVA is above average.
A similar pattern emerges when we include both SEVA and DMVA as performance
measures for the firm. In Table 9, we report the results of the estimation of the model with
both explanatory variables (along with a control for risk differences). MVA emerges as a
significant determinant of compensation in all but the high SEVA/low DMVA subsample.
Further, in the subsample where EVA is below average but MVA is above average, the
results are similar to those presented in Table 8. The estimates of the coefficients on
DMVA are all significantly positive, while those on SEVA are all significantly negative,
i.e., higher SEVA brings in a compensation penalty, but higher DMVA is rewarded
(Further, the adjusted R
2
values are the highest of all four subsamples.) These results
therefore run counter to the argument that managers are rewarded for myopic behavior.
Table 7
Descriptive statistics (means) for subsamples of firms based on SEVA and DMVA values
a
High SEVA/high DMVA High SEVA/low DMVA
N 233 234
SEVA 0.0968 0.0707
DMVA 2.4140 0.9965
Salary 0.7402 0.4232
Bonus 0.5832 0.3121
CashComp 1.3234 0.7353
TDC 3.2179 1.4199
Low SEVA/high DMVA Low SEVA/low DMVA
N 225 198
SEVA 0.0801 0.0752
DMVA 1.2463 0.8697
Salary 0.4549 0.3232
Bonus 0.2406 0.1295
CashComp 0.6954 0.4527
TDC 1.4549 0.9646
a
Compensation variables are expressed as percent of BOYCAP.
A. Fatemi et al. / Global Finance Journal 14 (2003) 159179 172
Table 8
SEVA and DMVA regressions for HH/HL/LH/LL groups (3 3 sorting, only 4 used in estimation)
Salary Bonus TDC Salary Bonus TDC
High SEVA/high DMVA High SEVA/low DMVA
N 233 233 233 234 234 234
Intercept 0.539
(1.40)
0.630
(1.50)
3.972
(1.44)
0.020
(0.11)
0.207
(1.09)
1.823**
(1.70)
COC 0.094***
(3.15)
0.084**
(2.59)
0.509**
(2.38)
0.025**
(1.76)
0.029**
(1.95)
0.227**
(2.57)
SEVA 0.606
(1.21)
1.240**
(2.28)
6.144**
(1.72)
1.191***
(3.29)
2.099***
(5.52)
5.328**
(2.21)
F statistic
( P value)
6.92
( < .01)
7.64
( < .01)
5.50
( <.01)
8.32
( <.01)
19.83
( <.01)
11.08
( <.01)
Adjusted R
2
.05 .05 .04 .06 .14 .08
Low SEVA/high DMVA Low SEVA/low DMVA
N 225 225 225 198 198 198
Intercept 2.035**
(2.26)
1.077**
(2.49)
6.436**
(2.45)
0.330***
(2.83)
0.276**
(2.11)
2.186**
(2.44)
COC 0.062**
(2.54)
0.061***
(2.62)
0.276***
(2.74)
0.052***
(4.48)
0.039***
(3.89)
0.279***
(4.02)
SEVA 22.507**
(1.98)
8.031
(1.62)
60.388**
(1.87)
0.480
(0.79)
0.796
(1.17)
1.843
(0.40)
F statistic
( P value)
101.72
( < .01)
49.46
( < .01)
74.07
( <.01)
17.84
( <.01)
8.08
( <.01)
8.12
( <.01)
Adjusted R
2
.47 .30 .39 .15 .07 .07
High SEVA/high DMVA High SEVA/low DMVA
N 233 233 233 234 234 234
Intercept 0.317
(0.86)
0.542
(1.29)
2.289
(0.87)
0.007
(0.04)
0.254
(1.27)
1.943**
(1.72)
COC 0.063**
(2.16)
0.074**
(2.25)
0.272
(1.32)
0.032**
(2.13)
0.038**
(2.34)
0.252***
(2.70)
SEVA 0.103***
(4.86)
0.070***
(2.91)
0.825***
(5.50)
0.032
(0.75)
0.092**
(1.99)
0.184
(0.53)
F statistic
( P value)
18.61
( < .01)
9.34
( < .01)
19.60
( <.01)
3.07
(.05)
6.09
( <.01)
7.03
( <.01)
Adjusted R
2
.13 .07 .14 .02 .04 .05
Low SEVA/high DMVA Low SEVA/low DMVA
N 225 225 225 198 198 198
Intercept 1.494***
(2.84)
0.904***
(3.56)
5.039***
(3.17)
0.236**
(2.15)
0.218**
(1.70)
1.783**
(2.02)
COC 0.007
(0.15)
0.040**
(1.79)
0.123
(0.89)
0.054***
(4.57)
0.041***
(4.10)
0.288***
(4.17)
SEVA 1.503***
(7.74)
0.565***
(6.03)
4.115***
(7.02)
0.087
(1.38)
0.157**
(2.57)
0.745**
(1.77)
F statistic
( P value)
31.76
( < .01)
23.26
( < .01)
27.69
( <.01)
18.99
( <.01)
10.89
( <.01)
9.72
( <.01)
Adjusted R
2
.22 .17 .19 .15 .09 .08
A. Fatemi et al. / Global Finance Journal 14 (2003) 159179 173
Rather, they are consistent with the argument that executive compensation is closely linked
to shareholder value creation.
Taken together, the results presented in Tables 79 indicate executive compensation
levels are related to the level of EVA and MVA generated by the firm. Moreover, both
EVA and MVA as measures of firm performance help explain the cross-sectional
variation of compensation within and across subsamples formed on the basis of the
levels of EVA and MVA generated. In all cases, except for the high SEVA/low DMVA
subsample, compensation is significantly positively related to DMVA. More importantly,
when MVA is high, compensation can be a negative function of EVA. This negative
relationship suggests that for firms with valuable growth options, those managers that
create value through these options are compensated in proportion to the value created.
Those managers that forego the opportunities in favor of improving current-period EVA
are penalized.
1
Table 9
SEVA and DMVA regressions for HH/HL/LH/LL groups (3 3 sorting, only 4 used in estimation)
Salary Bonus TDC Salary Bonus TDC
High SEVA/high DMVA High SEVA/low DMVA
N 233 233 233 234 234 234
Intercept 0.323
(0.68)
0.540
(1.29)
2.326
(1.41)
0.020
(0.11)
0.208
(1.10)
1.825**
(1.69)
COC 0.066**
(1.67)
0.073**
(2.20)
0.296**
(2.12)
0.024
(1.64)
0.025
(1.61)
0.220**
(2.44)
SEVA 1.330
(1.49)
0.436
(0.63)
8.586
(0.92)
1.174***
(3.21)
2.032***
(5.32)
5.211**
(2.36)
DMVA 0.139***
(2.68)
0.058**
(1.89)
1.060**
(1.93)
0.015
(0.36)
0.063
(1.43)
0.110
(0.36)
F statistic
( P value)
14.25
( < .01)
6.35
( <.01)
14.59
( < .01)
5.57
( < .01)
13.96
( < .01)
7.45
( <.01)
Adjusted R
2
.15 .06 .15 .06 .14 .08
Low SEVA/high DMVA Low SEVA/low DMVA
N 225 225 225 198 198 198
Intercept 2.486***
(2.67)
1.253***
(2.82)
7.689***
(2.80)
0.265**
(2.52)
0.151
(1.10)
1.603**
(1.69)
COC 0.016
(0.50)
0.043
(1.62)
0.147
(1.24)
0.053***
(4.48)
0.042***
(4.17)
0.289***
(4.19)
SEVA 19.821**
(2.16)
6.983**
(1.72)
52.929**
(2.02)
0.410
(0.68)
0.931
(1.38)
2.472
(0.53)
DMVA 0.947**
(2.55)
0.370**
(1.92)
2.630**
(2.16)
0.084
(1.30)
0.164***
(2.67)
0.762**
(1.80)
F statistic
( P value)
91.84
( < .01)
41.71
( <.01)
64.54
( < .01)
12.78
( < .01)
7.93
( < .01)
6.55
( <.01)
Adjusted R
2
.55 .35 .46 .15 .10 .08
1
While it would be interesting to replicate the results in Tables 8 and 9 in the context of the firms global
presence, the limited sample size (119) for which we have data on FSALES prevents us from doing so.
A. Fatemi et al. / Global Finance Journal 14 (2003) 159179 174
3.3. Compensation and the reward/incentive hypothesis
Finally, we examine whether executive compensation serves as a reward for
managerial effort based on superior past performance or as an incentive for improved
future firm performance. These tests are conducted using lagged and leading values of
firm performance. If compensation rewards managers for superior performance in the
current period only, then neither the lagged nor the leading values of firm performance
would be significant in explaining current period compensation. If compensation rewards
managers for prior superior performance, then only the lagged performance variables
would be significant. Conversely, if compensation is used to motivate managers to
improve performance in the future, only the leading values of performance would be
significant.
We examine compensation on two lagged and two leading values of SEVA and DMVA
after controlling for differences in firm risk. Since a total of 5 years of firm performance is
required to test this hypothesis, the sample size is reduced to 1871 observations. We use
the compensation data from 2 years (1992 and 1993). Thus, year t is defined as one of
these 2 years and lagged and leading measures of performance are measured relative to
this base year. We estimate the following model:
COMP
i;t
a
0
a
1
COC
i;t
a
2
PERF
i;t2
a
3
PERF
i;t1
a
4
PERF
i;t
a
5
PERF
i;t1
a
6
PERF
i;t2
e
i
6
Table 10 presents the estimates of this models parameters using SEVA as the measure
of firm performance. The estimates of the coefficients on contemporaneous and lagged
values of SEVA are insignificantly different from zero. Thus, at least when firm
performance is measured by SEVA, there is no evidence to support that compensation
rewards managers for past or current performance. However, the results reported in Table
10 provide limited support for the incentive pay hypothesis. The estimates of a
5
are
positive and significant for the cash components of compensation (salary and bonus).
Table 10
Test of the reward/incentive hypothesis for EVA as a determinant of compensation, for the sample 1871
observations in 1992 and 1993
a
Salary Bonus TDC
Intercept 1.776*** ( 4.38) 1.497*** ( 4.65) 6.073*** ( 3.98)
COC
t
0.373*** (9.65) 0.258*** (8.51) 1.167*** (7.80)
SEVA
t 2
3.937 (1.52) 1.485 (0.90) 1.609 (0.24)
SEVA
t 1
0.760 (0.00) 2.205 (0.00) 10.039 (0.00)
SEVA
t
7.345 ( 0.71) 1.352 (0.42) 16.390 ( 0.69)
SEVA
t + 1
12.255** (1.80) 7.893*** (2.92) 7.002 (0.40)
SEVA
t + 2
4.485** ( 2.07) 0.139 ( 0.11) 15.677** (2.17)
F statistic ( P value) 38.45 ( <.01) 48.72 ( < .01) 9.72 ( < .01)
Adjusted R
2
.11 .13 .03
***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively.
a
t statistics are in parenthesis.
A. Fatemi et al. / Global Finance Journal 14 (2003) 159179 175
Moreover, the estimate of a
6
is also significantly greater than zero for total compensation,
but not for bonus, and in fact is significantly negative for salary. Thus, at least in the short
run, salary and bonus act as incentives for maximizing EVA, while TDC does so in the
longer term.
In Table 11, we present the estimates of the model using DMVA as the measure of firm
performance. Our estimates of the coefficient on DMVA
0
(a
4
) are positive and signifi-
cantly different from zero for all three measures of compensation. Thus, executives are
rewarded for maximizing current period MVA. Moreover, the estimates of the coefficients
on DMVA
t 2
(a
2
) are also positive and significantly different from zero for bonus and
TDC. These results are therefore consistent with the hypothesis that compensation rewards
managers for current and past performance measured in terms of the shareholder wealth
added.
The coefficients on the leading terms of DMVA (a
5
and a
6
) are insignificant for salary
and bonus. However, we find a significant positive relationship between TDC and
DMVA
t + 2
, as evidenced by our estimates of a
6
. Since TDC includes salary, bonus, and
other noncash compensation, the significance of the coefficient in the TDC regression is
driven by the noncash components of compensation. Since the noncash component
includes stock options granted to executives, the value of these options increases with
the market value of the firm, resulting in higher executive compensation.
Taken together, our results suggest that executives are rewarded for their efforts to
create economic and market value for the firm. Further, their compensation also acts as an
incentive for creating additional market value in the future.
4. Summary and conclusions
In this study, we examine the relationship between executive compensation and
measures of performance capturing the economic profit earned by the firm (EVA and
MVA). Consistent with prior studies examining top manager compensation, we deseg-
regate the compensation package of top managers into cash, merit pay (bonus), and TDC,
Table 11
Test of the reward/incentive hypothesis for MVA as a determinant of compensation, for the sample 1871
observations in 1992 and 1993
a
Salary Bonus TDC
Intercept 1.537*** ( 2.71) 1.732*** ( 4.00) 9.697*** ( 5.00)
COC
t
0.264*** (5.49) 0.172*** (4.93) 0.832*** (5.12)
DMVA
t 2
0.072 (1.34) 0.170*** (2.87) 0.516** (2.27)
DMVA
t 1
0.254 (0.00) 0.224 (0.00) 0.171 (0.00)
DMVA
0
0.532** (1.73) 0.673*** (3.72) 2.687*** (3.00)
DMVA
t + 1
0.054 (0.29) 0.058 ( 0.24) 2.450*** ( 2.74)
DMVA
t + 2
0.280 (1.55) 0.059 (0.28) 2.172*** (2.95)
F statistic ( P value) 55.82 ( < .01) 102.79 ( < .01) 59.84 ( < .01)
R
2
.15 .25 .16
***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively.
a
t statistics are in parenthesis.
A. Fatemi et al. / Global Finance Journal 14 (2003) 159179 176
which includes long-term incentives such as stock options. We also investigate the causal
direction between the wealth creation activities of the firm and the compensation of its top
managers.
We document that executive compensation is positively related to the level of risk borne
by the firm. We find that the MVA to the firm is a significant determinant of executive
compensation. Comparing traditional measures of firm performance and MVA, we find
that including MVA in assessing executive compensation provides additional information
about the nature of top manager compensation. In general, we find that EVA and MVA are
better predictors of cross-sectional variation in top manager pay than traditional perform-
ance measures such as ROA, although the relationship between EVA and compensation is
found to be weaker.
We also examine whether executive compensation is an increasing function of the
extent of the firms global activities measured by the ratio of overseas sales to total sales.
We find that, in general, the basic pay-for-performance relationship is unaffected by the
extent of the firms global activities. However, we also present limited evidence, albeit
based on a smaller sample, that executives of firms with significant overseas operations
enjoy somewhat higher compensation.
Perhaps one of the most significant findings of our study is arrived at when we divide
our sample firms into four groups based on their rankings of market and EVA. By
grouping firms into cohorts based on relative levels of performance for both performance
measures, we identified four distinct groups. The performance cohorts may be referred to
as winners (high MVA and EVA), losers (low MVA and EVA), holders of real
options (high MVA/low EVA), and problem children (low MVA/high EVA). We find
that when EVA is achieved at the expense of MVA (i.e., the problem children group),
such behavior brings in a compensation penalty. Accordingly, it can be inferred that
compensation contracts are generally set in a manner that encourages managers to act in
the long-term interest of the shareholders, even when doing so may mean lower short-
term profits.
Finally, we assess whether top manager pay is an incentive for future performance or
reward for past behavior. Our evidence suggests that top managers are not only incented to
increase the EVA of the firm, but also rewarded for current and past additions to MVA. We
also assess the causal direction of the pay-for-performance relationship. We demonstrate
that achieving superior wealth creation for owners is accomplished by linking top manager
pay to the economic value created in the current period and by the ability of managers to
signal the marketplace about the progress (real options) being created by the firm for future
periods. Thus, our study indicates that the best wealth-creating top managers are
responsible for multidimensional behavior of the firmEVA additions today as well as
in the future (and hence, MVA).
An interesting extension of our work would consider including additional information
regarding analysts estimates of future earnings as a measure of future MVA performance.
This would help clarify whether the marketplace expects the performance of winners, or
executives are rewarded for performance surprises. Further, while we provide small
sample-based evidence of the pay-for-performance relationship in an international envi-
ronment, extending this study would help assess whether compensation plans for interna-
tional managers can be structured in the same manner as for domestic managers.
A. Fatemi et al. / Global Finance Journal 14 (2003) 159179 177
References
Agarwal, N. (1981). Determinants of executive compensation. Industrial Relations, 20(1), 3645.
Becker, G. (1964). Human capital. New York: National Bureau for Economic Research.
Carpenter, M., Sanders, W., & Gregersen, H. (2001). Building human capital with organizational context: the
impact of international assignments experience on multinational firm performance and CEO pay. Academy of
Management Journal, 44(3), 493511.
Copeland, T., Koller, T., & Murrin, J. (1995). Valuation: measuring and managing the value of companies. New
York: Wiley.
Deckop, J. R. (1988). Determinants of chief executive officer compensation. Industrial and Labor Relations
Review, 41(2), 215226.
Demski, J. S., & Sappington, D. E. (1989). Hierarchical structure and responsibility accounting. Journal of
Accounting Research, 27(1), 4058.
Dodd, J. L., & Chen, S. (1996). EVA: a new panacea? Business and Economic Review, 42(4), 2628.
Dodd, J. L., & Johns, J. (1999). EVA reconsidered. Business and Economic Review, 45(3), 1318.
Fama, E. (1980). Agency problems and the theory of the firm. Journal of Political Economy, 88, 288307.
Freedman, R. (2000). Long-term incentive plans: U.S. multinational firms erase the borders. Journal of Com-
pensation and Benefits, 16(5), 2628.
Gomez-Mejia, L., Tosi, H., & Hinkin, T. (1987). Managerial control, performance and executive compensation.
Academy of Management Journal, 30(1), 5170.
Grossman, S. J., & Hart, O. D. (1983). An analysis of the principal agent problem. Econometrica, 51, 745.
Hadlock, C. J., & Lumer, G. B. (1997). Compensation, turnover, and top management incentives: historical
evidence. Journal of Business, 70(2), 153187.
Hall, B., & Liebman, J. B. (1998). Are CEOs really paid like bureaucrats? Quarterly Journal of Economics,
113(2), 653691.
Harris, M., & Raviv, A. (1979). Optimal incentive contracts with imperfect information. Journal of Economic
Theory, 20, 231259.
Hodak, M. (1994). How EVA can help turn mid-sized firms into large companies. Journal of Applied Corporate
Finance, 98102.
Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: managerial behavior, agency costs and ownership
structure. Journal of Financial Economics, 3(4), 305360.
Jensen, M., & Murphy, K. (1990). Performance pay and top-management incentives. Journal of Political
Economy, 98(2), 225264.
Kramer, J. K., & Peters, J. R. (2001). An interindustry analysis of economic value added as a proxy for market
value added. Journal of Applied Finance, 11(1), 4149.
Kroll, M., Simmons, S., & Wright, P. (1990). Determinants of CEO compensation following major acquisitions.
Journal of Business Research, 20, 349366.
Lehn, K., & Makhija, A. (1996, May/June). EVA and MVA as performance measures and signals for strategic
change. Strategy and Leadership, 3438.
Leonard, J. S. (1990). Executive pay and firm performance. Industrial and Labor Relations Review, 43,
1329.
Mehran, H. (1995). Executive compensation structure, ownership and firm performance. Journal of Financial
Economics, 38(2), 163184.
Miller, D. (1995). CEO salary increases may be rational after all: referents and contracts in CEO pay. Academy of
Management Journal, 38(5), 13611385.
Murphy, K. J. (1985). Corporate performance and managerial remuneration: an empirical analysis. Journal of
Accounting and Economics, 7, 1142.
Pavlik, E. L., & Belkaoui, A. (1991). Determinants of executive compensation. Westport, CT: Quorum Books.
Platt, G. (2002). CEOs find performance does matter. Global Finance, 16(4), 2627.
Porac, J., Wade, J., & Pollock, T. (1999). Industry categories and the politics of the comparable firm in CEO
compensation. Administrative Science Quarterly, 44(1), 112144.
Rappaport, A. (1986). Creating shareholder value: the new standard for business performance. New York: Free
Press.
A. Fatemi et al. / Global Finance Journal 14 (2003) 159179 178
Richard, J. (2000). Global executive compensation: a look at the future. Compensation and Benefits Review,
32(3), 3538.
Shiely, J. (1996). Is value management the answer? Chief Executive, 119, 5457.
Spinner, K. (1995, November). Signed and sealedbut can EVA deliver? CFO Magazine, 9395.
Stewart, G. (1991). The quest for value: a guide for senior managers. New York, NY: Harper Collins.
Tosi, H., Werner, S., Katz, J., & Gomez-Mejia, L. (2000). How much does performance matter? A meta-analysis
of CEO pay studies. Journal of Management, 26(2), 301339.
Tully, S. (1993, September). The real key to creating value. Fortune, 3844.
Uyemura, D. G., Kantor, C. C., & Pettit, J. M. (1996, Summer). EVA for banks: value creation, risk management,
and profitability management. Journal of Applied Corporate Finance, 94113.
Weaver, S. C. (2001). Measuring economic value added: a survey of the practices of EVA proponents. Journal of
Applied Finance, 11(1), 5060.
Werner, S., & Tosi, H. L. (1995). The effect of ownership on compensation strategy and managerial pay.
Academy of Management Journal, 38(6), 16721691.
White, H. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direst test for heteroskedas-
ticity. Econometrica, 48, 817838.
Winn, D., & Shoenhair, J. D. (1988). Compensation-based (dis)incentives for revenue-maximizing behavior: a
test of the revised Baumol Hypothesis. Review of Economics and Statistics, 154158.
A. Fatemi et al. / Global Finance Journal 14 (2003) 159179 179