The Search for the Best Financial

Performance Measure
Jeffrey M. Bacidore, John A. Boquist, Todd T. Milbourn, and Anjan V. Thakor

Refined economic value added (REVA) provides an analytical framework for

evaluating operating performance measures in the context of shareholder value
creation. Economic value added (EVA) performs quite well in terms of its correlation
with shareholder value creation, but REVA is a theoretically superior measure for
assessing whether a firm’s operating performance is adequate from the standpoint of
compensating the firm’s financiers for the risk to their capital. In this article,
comprehensive statistical analysis of both REVA and EVA is used to estimate their
correlation with and their ability to predict shareholder value creation. REVA
statistically outperforms EVA in this regard. Moreover, the realized returns for the
1988–92 period for the top 25 REVA firms were higher than the realized returns for
the top 25 EVA firms.

n the 1980s, shareholder activism reached even less impact on the stock price than the CEO.
I unprecedented levels and led to increased pres-
sure on firms to maximize shareholder value con-
Tying managerial compensation to stock price may
impose excessive risk on managers and may detract
sistently. For example, Time magazine summarized from the ability of such compensation to provide
this activism as “Angry investors closed out the incentive for managers to maximize shareholder
Decade of Greed with demands that executive com- wealth.
pensation should be tied to company performance” Any financial performance measure used in
(Smolowe 1996). The basic idea is that if managers managerial compensation, on the one hand, must
are offered compensation contracts that are tied to be correlated highly with changes in shareholder
shareholder wealth changes, their incentives will wealth and, on the other, should not be subject to
be better aligned with those of shareholders than is all of the randomness and “noise” inherent in a
the case for other types of contracts. In designing firm’s stock price. This dichotomy is the fundamen-
such contracts, however, an important issue is tal tension a good performance measure must
which measure of shareholder performance to use
resolve. A recent example of a performance mea-
in designing the contract.
sure that seeks to resolve this tension is economic
The obvious metric for judging firm perfor-
value added (EVA). This measure, proposed by
mance is the stock price itself (see, e.g., Jensen and
Stern Stewart Management Services, creatively
Murphy 1990 and Milbourn 1996). Stock price, how-
links the firm’s accounting data to its stock market
ever (or returns based on stock price), may not be an
performance (Stewart 1991).
efficient contracting parameter because it is driven
Before examining the correlation between
by many factors beyond the control of the firm’s
executives.1 Moreover, as one moves down the orga- shareholder wealth and a performance measure,
nizational ranks, the inefficiencies of stock-based one must first define the appropriate way to mea-
compensation as a means of aligning managerial sure changes in shareholder wealth. We contend
interests with those of stockholders become even that shareholders are concerned with the abnormal
more evident because managers at lower levels have return they earn in any period—that is, the return
they earn in excess of what they expected to earn for
a firm within a given systematic risk class. When
this return is positive, shareholders have more than
Jeffrey M. Bacidore is a doctoral candidate in finance at covered their risk-adjusted opportunity cost of pro-
Indiana University. John A. Boquist is the Edward E. viding their capital. Conversely, when this return is
Edwards Professor of Finance at Indiana University.
Todd T. Milbourn is an assistant professor of finance at negative, they have been inadequately compen-
London Business School. Anjan V. Thakor is the sated for risk.2 Given this relationship, a good
Edward J. Frey Professor of Banking and Finance at the financial performance measure should correlate
University of Michigan. highly with abnormal stock returns.

Financial Analysts Journal • May/June 1997 11

 In this article, we present an empirical analysis Figure 1. The Components of Value
of the ability of EVA to predict abnormal return and
of the contemporaneous correlation between EVA Net Present Value
and abnormal return. In addition, we define a new of Current and
performance measure, a refinement of EVA, and Future Investment Market Value
Opportunities of Firm
examine its statistical properties. This new perfor-
mance measure, refined economic value added
(REVA), complements EVA in that it could be used
in conjunction with EVA, with the choice of mea-
sure dictated by the level of the organization at
which the performance measure is used. We argue Economic Book
Assets in Place
that REVA is a better measure of performance for Value of Assets
top management, although EVA may be useful at
lower levels. Our empirical tests are extensive, and
we also focus on nonparametric tests of the predic-
tive abilities of EVA and REVA.
Although many companies use EVA for both
resource allocation and compensation purposes,
virtually no research reported in the published lit- Invested Capital, Strategy and Value
erature provides information about the statistical Figure 1 illustrates the components of firm
relationship of EVA to shareholder value.3 One goal value. The most transparent component of a firm’s
of this article is to fill that void. Moreover, much value is its physical assets in place: plant and equip-
remains to be learned about alternatives and com- ment, real estate, working capital, and so forth.
plements to EVA, particularly for managers at dif- Another component is the net present value of the
ferent levels of an organization. Although our firm’s current and future investment opportunities.
This component’s value is less tangible than its
efforts in this direction are by no means exhaustive,
physical assets, is driven significantly by the firm’s
we hope that the introduction of REVA is a useful
strategy, and is sizable for many firms. The total
first step.
value of the firm is the sum of these two compo-
nents of value. The question is: How do we deter-
STRATEGY, VALUE, AND THE CHOICE mine these values?
OF PERFORMANCE MEASURE The firm’s balance sheet contains one measure
An appropriate performance measure gauges how of the value of the firm’s assets in place. Consider
management strategy affects shareholder value as the accounting-based balance sheet in Table 1, for
measured by the risk-adjusted return on invested example. Because of a variety of accounting distor-
capital. An appropriate performance measure to tions, the total asset value on the typical balance
gauge the effectiveness of a given strategy must, sheet does not accurately represent either the liqui-
therefore, incorporate the required rate of return on dation value or the replacement-cost value of the
invested capital, accurately measure the amount of assets in place. Therefore, it is of limited use for
capital used by the company, and correlate highly asset valuation purposes and must be transformed.
with the risk-adjusted rate of return earned by The proponents of EVA, most notably Stern
shareholders. In this section, we formalize how Stewart, are careful to adjust this accounting bal-
invested capital and the required rate of return on ance sheet before arriving at an estimate of the
that capital should be measured, as well as what is value of the firm’s assets in place.4 The adjustments
meant by an appropriate risk-adjusted return to include netting the non-interest-bearing current
shareholders. liabilities against the current assets, adding back to

Table 1. Accounting-Based Balance Sheet

Assets Liabilities and Net Worth
Current assets Non-interest-bearing current liabilities (NIBCLs)
Net goodwill Interest-bearing current liabilities (IBCLs)
Fixed assets Long-term debt
Equity (net of write-offs)
Total assets Total liabilities and net worth

12 Association for Investment Management and Research

equity the gross goodwill (i.e., adding cumulative How Much Capital Is Invested in the
amortized goodwill back to total assets), restructur- Firm?
ing and other write-offs, capitalized value of R&D A good financial performance measure should
(and possibly advertising), LIFO reserve, and so ask how well the firm has generated operating
forth. The debt balance is increased by the capital- profits, given the amount of capital invested to
ized value of operating lease payments. The goal is produce those profits. The idea is that the firm’s
to produce an adjusted balance sheet that reflects financiers are free to liquidate their investment in
the economic values of assets in place more accu- the firm and invest the liberated capital elsewhere.
rately than the inherently conservative, historical- Thus, the financiers must earn at least their oppor-
cost-based balance sheet guided by generally tunity cost of capital on the invested capital. This
accepted accounting principles. condition implies that this cost of capital must be
After these adjustments, the typical firm’s subtracted from operating profits to gauge the
“economic book value” balance sheet would look firm’s financial performance. EVA, for that reason,
as it does in Table 2. Although this economic book defines net operating profit after tax (NOPAT) and
value balance sheet represents the value of the subtracts a capital charge for the economic book
assets in place more accurately than the balance value of assets in place. The economic book value
sheet produced by the firm’s accountants, it still of assets in place is the measure of the capital pro-
does not determine the total value of the firm, vided to the firm by its financiers. But does this
which includes the value of future opportunities. amount truly represent the capital used to generate
The total value of the firm—the sum of the two the operating profit?
components of value in Figure 1—is also equal to We believe the answer to that question is nega-
the market value of equity plus the market value of tive. At the beginning of any period, the financiers
debt. Thus, the difference between the market as a group could sell the entire firm for its market
value of the firm and the economic book value of value. They could then invest their proceeds in assets
its assets in place represents the market’s assess- identical in risk to the firm and earn an expected
ment of the value of the firm’s current and future return equal to the firm’s weighted-average cost of
investment opportunities. This difference can be capital (WACC).6 By not liquidating their holdings
considered an assessment of the value of the firm’s in the firm, these financiers are forgoing this oppor-
competitive strategy and its deployment of human tunity to earn the weighted-average cost of capital
resources. If we recast the firm’s balance sheet in on the market value of the firm at the beginning of
market value terms, we have a balance sheet such the period.
as that presented in Table 3.5 For the firm to create a true “operating” surplus
The market value of assets can be either above for its financiers in a given period, its operating
or below the economic book value of the assets in profit at the end of the period must exceed a capital
place. In particular, if the firm executes a poor charge that is based on the total market value of the
strategy in the opinion of the market or if it does capital used at the beginning of the period, not
not possess the human resources needed to imple- simply the economic book value of its assets in
ment a good strategy successfully, the market will place. The capital commitment of the firm’s finan-
lower the value of the firm’s assets, perhaps below ciers is represented by the total market value of the
the economic book value of the assets in place. firm, not simply the economic book value of the

Table 2. Economic Book Value Balance Sheet

Assets Liabilities and Net Worth
Current assets (with inventory at FIFO)—NIBCLs IBCLs
Gross goodwill Debt (+ capitalized leases)
Fixed assets Equity (+ adjustments for deferred taxes,
goodwill amortization, write-offs, LIFO
reserves, etc.)
Economic book value of assets in place Total liabilities and net worth

Table 3. Market-Value-Based Balance Sheet

Assets Liabilities and Net Worth
Market value of assets Market value of debt (including leases)
Market value of equity
Total market value of assets Total market value of debt and equity

Financial Analysts Journal • May/June 1997 13

assets in place. This “investment” in the firm in any (CAPM) captures the first two factors by specifying
given period constitutes the capital that the firm has that the expected return on a stock investment is
used to produce profits. This perspective motivates
E ( R j, t ) = Rft + β j [ E ( Rm t ) – R f ] ,
our development of REVA in the next section. t

where
Operating versus Trading-Based
Rf = the risk-free bond yield at time t
Performance t

Measures of shareholder wealth creation focus βj = firm j’s beta, a measure of the
on the firm’s stock price performance and seek to firm’s systematic risk
determine how much shareholders increase their
E ( Rm t ) – R f = the expected equity market risk
wealth from one period to the next based on the t
premium, usually taken as the
dividends they receive and the appreciation in the
long-run, average realized return
firm’s stock price. Essentially, such trading-based
on the market in excess of risk-free
performance measures assess how well an investor
bond returns
would have done if he or she had purchased a share
of stock at the beginning of the period and sold it The CAPM thus helps to determine the abnormal
at the end. This type of measure of shareholder return firm j earned in period t. We call this return
wealth creation is called a trading-based measure alpha and calculate it as
of performance. In contrast, performance measures α j, t = R j, t – E ( Rj, t ) .
such as EVA focus on the firm’s operating perfor-
mance from the standpoint of its financiers. Thus, α j, t measures the actual shareholder return
An operating measure of current performance in excess of the return that was expected in a period,
focuses solely on the performance of the firm in a given the firm’s systematic risk. In that sense, the
given period; a trading-based measure of perfor- return is abnormal. Alpha, we propose, is thus the
mance captures revisions in the market’s beliefs appropriate measure of shareholder wealth cre-
about the firms’s entire future stream of operating ation in any given period. Consequently, alterna-
performances. If stock markets are efficient and we tive operating performance measures will be
examine a sufficiently long time horizon, these two judged by their correlation with this abnormal
measures will converge. We usually assess perfor- return.
mance over shorter horizons, however: a year, a This measure is indeed a high hurdle for per-
quarter, or a month. Therefore, any operating mea- formance. One would not expect a large number of
sure of performance will diverge somewhat from a firms to achieve positive abnormal returns for
trading-based measure of performance. Obviously, many periods. In fact, because a firm with a string
we do not want to use a trading-based measure of of positive alphas is one that continues to produce
performance for compensating all of our managers, shareholder returns in excess of the risk-adjusted
particularly those at lower levels of the organiza- expected return, this firm is consistently “beating
tion, whose decisions have less impact on stock market expectations.” Firms that perform this well
price. We do, however, want a measure of perfor- are truly exceptional and quite rare. The question,
mance to be a barometer of shareholder wealth then, is: What operating measure of performance
creation against which we can judge the efficiency correlates highly with this measure of shareholder
of any operating performance measures. wealth creation?
Shareholders can earn a return on their invest-
ment in two ways: through dividends and through REVA AS A HIGH-LEVEL
capital gains. Over any period of time, t, the share- PERFORMANCE MEASURE
holder return for firm j can be specified as The operating measure of performance should
D j, t + ( Pj, t – Pj, t–1 ) obviously capture how well a given company has
Rj, t = -------------------------------------------------
-, performed in terms of operating income. Net oper-
P j, t–1
ating profit after taxes alone, however, is not an
where Dj,t is the dividends paid during the period appropriate measure because it neither captures
t – 1 to t and Pj,t is the price of the shares at the end how much capital is used to generate a given level
of period t. of income nor accounts for the required rate of
Many factors influence Rj,t, most notably, the return on invested capital. Thus, an appropriate
risk of the investment, the interest rates prevailing performance measure is one that includes net oper-
in the capital markets, and the expertise of the ating income after taxes, the amount of capital
firm’s managers. The capital asset pricing model invested, and the required rate of return on capital.

14 Association for Investment Management and Research

What Is REVA? t – 1 (or the beginning of period t) rather than on
EVA is defined as the economic book value of the assets in place.
The following example highlights the potential
EVA = NOPAT – kW (NA), (1)
differences in the shareholder wealth implications of
where kW is the weighted-average cost of capital EVA and REVA. Suppose an investor holds one share
and NA is defined as adjusted book value of net of XYZ Company stock at the beginning of the year.
capital at the beginning of the period. The stock is currently trading at $50 a share, and the
In Equation 1, NOPAT is defined as reported economic book value of the stock is $40 a share.7
net operating profits plus any increase in bad debt Suppose that the investor and the rest of the market
reserve plus any increase in the LIFO reserve plus expect this stock to earn 10 percent over the next year.
amortization of goodwill plus any increase in net The investor expects this 10 percent return on the $50
capitalized R&D plus other operating income market value of the firm, not its $40 economic book
(including passive investment income) minus cash value. That is, the investor expects a payoff of $5 a
operating taxes. share, not $4. The contrast between EVA and REVA
A firm’s weighted-average cost of capital in can now be seen in this context. Suppose the firm is
Equation 1 ( WACC ≡ k W ) is derived from the follow- completely equity financed. With all-equity financ-
ing formula: ing, the WACC for this firm is simply its cost of
DM EM equity of 10 percent. Further assume that the firm
k W = ------------------------ k D ( 1 – T ) + ------------------------ k E , (2) has 100 shares of stock outstanding at the beginning
DM + EM DM + EM
of the year (period t = 0). Therefore, the economic
where book value of the firm is 100 times $40, or $4,000, and
the market value of the firm is 100 times $50, or
DM = market value of the firm's total debt
$5,000, at t = 0. If during the year the firm generates
EM = market value of the firm’s total equity a NOPAT of $450 on the firm's invested capital, how
has the firm done in terms of EVA and REVA?
kD = pretax cost of debt
The firm has generated an EVA at time t = 1 of
T = the firm’s marginal tax rate
EVA1 = NOPAT1 – ($4,000 × 10%)
kE = cost of equity
= $450 – $400
A firm’s net asset base in Equation 1 is defined by a = $50.
company’s total assets minus non-interest-bearing
current liabilities, with the adjustments discussed According to the EVA perspective, this firm has
earlier. That is, net assets represent the total eco- created $50 of value for its shareholders. Now, con-
nomic book value of the firm’s assets in place. sider the firm’s REVA:
The motivation for the REVA refinement to REVA1 = NOPAT1 – ($5,000 × 10%)
EVA stems partly from EVA’s use of the economic
= $450 – $500
book value of assets when the capital charge for the
firm is derived from a market-based WACC. To = –$50.
make inferences about changes in shareholder According to REVA, this firm has destroyed $50 in
wealth, a market-derived cost of capital should be shareholder value.
applied to the market value of the firm’s assets. Should the firm’s management be satisfied
Thus, the REVA for a given period t is defined as with its performance over this period? It has gener-
REVA t = NOPAT t – k W ( MV t–1 ) , (3) ated a positive EVA and a negative REVA. The firm,
with a cost of capital of 10 percent, has produced an
where NOPATt is the firm’s NOPAT at the end of 8 percent return on the market value of its assets and
period t and MVt–1 is the total market value of the an 11.25 percent return on the economic book value
firm’s assets at the end of period t – 1 (beginning of of its assets. Because an investor in this firm could
period t). MVt–1 is given by the market value of the have taken his or her $50 a share and invested it in
firm’s equity plus the book value of the firm’s total another company of equivalent risk to generate the
debt less non-interest-bearing current liabilities, all 10 percent required return, the NOPAT of 8 percent
at the end of period t – 1. of the market value of the firm is, in our view,
inadequate compensation.
EVA and REVA Compared This example, although simple, illustrates how
The key distinction between EVA and REVA is a firm could be delivering less in terms of operating
that REVA assesses its capital charge for period t earnings than the shareholders require and yet
on the market value of the firm at the end of period appear to be creating shareholder value based on

Financial Analysts Journal • May/June 1997 15

EVA. This possibility is the first justification for of the period exceeds their opportunity cost of cap-
using REVA. ital. This condition does not hold for EVA—the
financiers could be receiving an operating-income-
Flows to Equity versus Total Flows to based return that is less than their opportunity cost
All Financiers of capital even when EVA is positive. Second, REVA
can be computed based on total operating flows to
Another difficulty with EVA provides a second
debt and equity or only on the flows to equity. This
justification for using REVA. Conceptually, EVA
capability is true for EVA only when the market
should be the same regardless of whether we use
values of debt and equity coincide with their
NOPAT and the WACC or (adjusted) net income
respective economic book values.
and the equity cost of capital.8 However, as we
show below, there is no such equivalence when the
market value of a firm differs from its book value. Organization Level and Choice of
The flows to equity-based EVA are defined as Financial Measure
EVAt(Equity flows) = Net incomet – kE(E),
REVA is a more appropriate performance mea-
sure than EVA when considering the shareholders’
where net income is simply the NOPAT less the view of the firm. Hence, the senior executives in the
firm’s after-tax interest expense and E is the eco- firm (e.g., the CEO and other members of the exec-
nomic book value of equity. We can then write utive committee) should be evaluated on the basis
Net incomet = NOPATt – kD(1 – T)D,
of the firm’s REVA performance. The firm’s value
derives both from its physical assets in place and
where D is the economic book value of debt. There- its strategy with respect to future opportunities.
fore, we should have Both of these values are the appropriate domain of
EVAt(NOPAT) = EVAt(Equity flows). (4)
the firm’s senior executives. The market value of
the firm—which is a component in REVA—
Expanding both the left hand side and the right includes the values of both the physical assets and
hand side of Equation 4 yields the strategy, whereas the economic value of the
firm—which is a component in EVA—represents
NOPAT t – k W ( NA t–1 ) = Net income – k E ( E ) . (5)
only the values of the physical assets in place. Strat-
Dropping the time subscripts and recognizing that egy is the primary responsibility of top manage-
Net income = NOPAT – kD(1 – T)D, we can write ment. The firm’s economic value is an adequate
Equation 5 as representation of invested capital from the stand-
point of those below top management. Thus, REVA
 DM  could be used to compensate senior management
NOPAT t – k D  ( 1 – T ) ---------------------------- ( D + E ) 
 ( D M + E M )  and EVA could be used to compensate divisional
managers and those below them.
EM
– k E ---------------------------- ( D + E )
( DM + EM ) EMPIRICAL ANALYSIS
= Net income – k D ( 1 – T )D – k E ( E ). The empirical analysis used the Stern Stewart Per-
formance 1000 database for the years 1982 through
Therefore, for the equivalence to hold, DM must 1992. We randomly selected 600 of the 1,000 firms
equal D and EM must equal E. That is, the market in the database and proceeded to calculate each
values of debt and equity should be equal to the firm’s EVA, REVA, total shareholder return, and
respective economic book values. It is uncommon risk-adjusted abnormal return for each year. These
for the market value of equity to be precisely equal calculations involved matching the 600 firms to
to the economic book value of equity, primarily corporate financial data for those years. All
because market value includes an estimate of the accounting and financial market data were taken
value of future opportunities. This conceptual dif- from Standard & Poor’s Compustat and the Uni-
ficulty does not exist with REVA, primarily because versity of Chicago’s CRSP databases, respectively.
it relies on the market value of capital. Calculating a firm’s yearly EVA and REVA
We can thus summarize two key advantages of requires estimates of its NOPAT, WACC, economic
REVA relative to EVA. First, whenever REVA is book value of assets, and market value of assets. In
positive, incremental shareholder value has been our calculation of REVA, we used the NOPAT data
created. The operating income flowing to financiers provided by the Stern Stewart database. The market
at the end of the period as a percentage of the value of capital was estimated by summing the mar-
market value of their investment at the beginning ket value of equity, book value of interest-bearing

16 Association for Investment Management and Research

liabilities, and book value of preferred stock.9 Book returns of each security on various combinations of
values were used for debt and preferred stock EVA and REVA measures to determine how well
because market values for these variables are not each performance measure explains abnormal
available.10 The WACC was estimated as the returns—that is, returns over and above the
weighted average of the cost of equity, debt, and expected return. Both EVA and REVA were scaled
preferred stock, where the weights are the market- by the market value of equity to create two new vari-
based capital structure weights obtained from the ables, EVARET and REVARET. This step was done
capital structure components. To estimate the firm’s to make both variables consistent with the abnormal
cost of debt, we assigned the bond yield as reported return variable, which was measured as a percent-
in Standard & Poor’s Industrial Bond Guide for com- age. The market value of equity was used because
mensurate bond ratings. The after-tax cost of debt both EVA and REVA gauge the value creation for
was estimated by multiplying the firm’s cost of debt shareholders. First, we regressed abnormal returns
by its marginal tax rate, which was estimated by on EVARET and REVARET individually. Our
dividing the reported tax expense by the firm’s pretax results indicate that, on an individual basis, both
income. The cost of equity was estimated using the EVARET and REVARET are positively related to
CAPM.11 The cost of preferred stock was estimated abnormal returns at the 1 percent significance level
as the average of the cost of equity and debt.
(see Table 4). A 1 percent increase in EVARET results
To determine the cost of equity, estimates of the in a 0.27 percent increase in abnormal returns. Sim-
risk-free rate, the expected market risk premium, ilarly, a 1 percent increase in REVARET leads to a
and the firm’s beta are needed. The yield to matu-
0.58 percent increase in abnormal returns. There-
rity of a one-year discount bond taken from the
fore, on average, an increase in either EVA or REVA
CRSP Fama–Bliss files was used as a proxy for the
leads to an increase in shareholder wealth.
risk-free rate. To gauge the expected market risk
■ What about past EVA and REVA as predictors of
premium, historical averages of realized annual
future abnormal returns? Our next set of tests
market risk premiums from 1926 to the year prior
to the observation year were calculated. The real- involved adding lagged values of EVARET and
ized market risk premiums were taken from Ibbot- REVARET into the abnormal returns regressions.
son Associates, 1995 Yearbook of Stocks, Bonds, Bills, These tests examined whether past realizations of
and Inflation.12 Firm betas were calculated using the EVA and REVA have a significant effect on abnor-
methodology outlined by Fama and French (1992). mal returns. In the regression of abnormal returns
This methodology involves first assigning all on EVARET and lagged EVARET, EVARET is signif-
NYSE, Amex, and Nasdaq firms available on CRSP icantly positively related to abnormal returns and
to 10 size-based portfolios. Each size portfolio is lagged EVARET is significantly negatively related to
then portioned into 10 beta portfolios. The monthly abnormal returns (see Table 5). In a multiple regres-
returns of each of the 100 equally weighted portfo- sion framework, the coefficient on lagged EVARET
lios were used to estimate the portfolios’ betas represents the sensitivity of abnormal returns to
using the techniques outlined by Dimson (1979).13 changes in lagged EVARET, which are uncorrelated
The portfolio beta was assigned to each firm within with contemporaneous EVARET. Similarly, the
its portfolio and used to calculate the firm’s cost of coefficient on EVARET measures the sensitivity of
equity. abnormal returns to changes in EVARET, which are
EVA was estimated using NOPAT and the eco- uncorrelated with last period’s EVARET. Thus, the
nomic book values of capital from Stern Stewart, coefficient on EVARET represents how abnormal
and the WACC used to calculate REVA was also
used to re-estimate EVA. This step was to isolate the Table 4. Abnormal Returns as Explained by
true difference between REVA and EVA—that is, EVARET and REVARET
the different capital bases. Using different WACCs (t-statistics in parentheses)
could provide results driven solely by the method Coefficient EVARET REVARET
used to estimate WACC and, as a result, would Intercept 0.12047 0.16232
cloud the true underlying relationship between the (16.683)** (19.366)**
variables in question.14 Abnormal return 0.26620 0.57997
(5.942)** (11.212)**
Empirical Results 2
R 0.0114 0.0393
■ How well do EVA and REVA correlate with
Note: Sample size = 3,076.
shareholder wealth creation? The first set of tests
involved regressing the CAPM-based abnormal **Significant at the 1 percent level.

Financial Analysts Journal • May/June 1997 17

 Table 5. Abnormal Returns on Contemporaneous and Lagged EVARET and
REVARET
(t-statistics in parentheses)
EVARET REVARET
Coefficient Current Lagged Current Lagged
Intercept 0.11969 0.16549
(15.357)** (16.138)**
Abnormal return 0.47639 –0.12918 0.75111 –0.10277
(6.555)** (–1.819)* (9.588)** (–1.408)

R2 0.0205 0.0396
Note: Sample size = 2,574.
*Significant at the 10 percent level.
**Significant at the 1 percent level.

returns vary with unexpected changes in EVA, where between REVA and EVA. REVA is positively related
expectations are based on the previous period’s to abnormal returns, but any revaluation based on
EVA. The results above are consistent if one assumes past period’s REVA is impounded into contempo-
that the market uses EVA not only to assess abnor- raneous REVA directly; this result occurs because
mal returns this period but also to predict future REVA is a function of the market value of equity. A
performance. If the firm has a large EVA this period, positive REVA in a given period that leads to revi-
the market may revise its valuation of the firm sions in the market’s expectations regarding future
upward significantly, reflecting revised expecta- performance will “raise the hurdle” by increasing
tions about future profitability. If the market’s the market value of equity and, as a consequence,
beliefs are not confirmed in the next period, how- next period’s capital charge. Thus, REVA is a more
ever, the value of the stock could fall. Thus, even if a appropriate compensation measure because it is a
firm has a positive EVA this period, the stock may truer measure of whether the firm has surpassed
not earn positive abnormal returns. the market’s expectation and thereby added share-
The first key implication of this analysis is that holder value.
the market appears to reward unexpected increases in We also regressed abnormal returns on EVARET
EVA and bases expectations of future profitability on and REVARET simultaneously. The purpose of such
this period’s EVA. The second key implication is that a regression was to determine which performance
EVA may be inappropriate to use as a compensation measure does better in explaining a firm’s abnormal
measure for top management because the market is returns when the impact of the other measure is
really rewarding only the unexpected portion of EVA. accounted for. The coefficient on REVARET is again
Basing compensation on total EVA may result in significantly positive; the EVARET coefficient is now
rewarding managers for subpar performance. significantly negative (see Table 6). This result indi-
A similar regression was run using REVARET cates that REVA contains information not captured
and lagged REVARET. The coefficient on contem- in EVA that is relevant for predicting abnormal
poraneous REVARET is again significant, although returns and that increases in EVA that are uncorre-
lagged REVARET is insignificant (see Table 5). This lated with changes in REVA result in significantly
result highlights a key empirical difference lower abnormal returns.

Table 6. Abnormal Returns as Explained by Both EVARET and REVARET

(t-statistics in parentheses)
Coefficient EVARET REVARET
Intercept 0.16746
(19.642)**
Abnormal return –0.20407 0.75450
(–3.162)** (9.981)**

R2 0.0424
Note: Sample size = 3,076.

**Significant at the 1 percent level.

18 Association for Investment Management and Research

 In summary, EVA is significantly related to wealth. The simpler measure, REVA, goes even far-
abnormal returns. As shown in Table 6, however, the ther, however. REVA appears to include all informa-
information provided by REVA subsumes the infor- tion relevant to a firm’s abnormal return that is
mation implied by EVA realizations. Therefore, con- contained in EVA. Moreover, the statistical power of
sistent with the theory, REVA better explains REVA holds up over long time horizons and out-of-
changes in shareholder wealth and also manage- sample tests. We formed three portfolios of stocks
ment performance at the top level of the firm.15 from our sample on the basis of the top 25 EVA firms
over the 1982–87 period, the top 25 REVA firms over
Nonlinear Relationships Estimation the same period, and a value-weighted market index
■ How often does a positive REVA (EVA) predict for the same period. We then examined the total
a positive excess return? One potential problem with portfolio return over the 1988–92 time period. The
the regression analysis is that both REVARET and results of this process, shown in Table 7, document
EVARET may be related to abnormal returns in a that the high-performance REVA firms outper-
highly nonlinear manner. An additional methodol- formed both the high-performance EVA firms and
ogy was used to account for any nonlinearities: the market index.
nonparametric tests to capture how well REVA and
EVA predict the sign of the abnormal returns. This CONCLUSION
approach is equivalent to calculating the probabil- This analytical framework for evaluating operat-
ity that given that a firm has a positive EVA (or ing performance measures in the context of share-
REVA), the firm generates a positive abnormal holder value creation indicates that the most
return. appropriate measure of shareholder value is the
Our test involved splitting the sample into two return shareholders earn through price apprecia-
groups: those with positive EVA and those with zero tion and dividends in excess of that required to
or negative EVA. The proportion of observations in compensate shareholders for systematic risk. We
each group that had positive abnormal returns was conclude that EVA does quite well in terms of its
then calculated. If EVA is a good predictor of abnor- correlation with this measure of shareholder value
mal returns, the proportion of firms that experienced creation.
positive abnormal returns in the positive EVA sub- EVA, however, views the economic book value
sample should exceed the proportion of firms that of the physical assets in place as the capital finan-
experienced positive abnormal returns in the nega- ciers commit to the firm. We propose that a more
tive EVA subsample. In other words, Prob(abnormal appropriate measure of the capital used in the firm
return > 0, given that EVA > 0) should be greater than for any period of time is the market value of the
Prob(abnormal return > 0 given that EVA < 0). We found firm at the beginning of the period. This approach
that 68.05 percent of the positive EVA sample exhib- led us to a refinement of the EVA measure, REVA.
ited positive abnormal returns whereas only 56.01 REVA assesses a capital charge for a period equal
percent of the negative EVA sample exhibited posi- to the weighted-average cost of capital times the
tive abnormal returns. This procedure was repeated market value of the firm at the beginning of the
using REVA. Positive REVA implied a positive period. This premise permits computation of REVA
abnormal return 77.22 percent of the time; a negative using either flows to equity or flows to all finan-
REVA implied a positive abnormal return only 58.48 ciers, which is not possible with EVA unless market
percent of the time.16 and economic book values happen to be equal by
These results indicate that the proportion of chance.
positive REVA that correspond to positive abnor-
We conducted a comprehensive statistical
mal returns is significantly higher than the same
analysis of both EVA and REVA to estimate their
proportion for EVA.17 Thus, although EVA on its
correlation with and ability to predict shareholder
own predicts abnormal returns fairly well, REVA
value creation. REVA statistically outperforms EVA
performs significantly better.18 This finding is
in this regard. Moreover, the realized returns for the
important because senior management should seek
top 25 REVA firms were higher than the realized
a performance measure with the greatest ability to
predict correctly directional changes in shareholder Table 7. Portfolio Analysis of EVA and REVA
wealth. Our results imply that REVA serves both of (percentages)
these roles, as displayed in Table 6 and in the non- Portfolio Total 1988–92 Portfolio Return
parametric tests.
Top 25 EVA firms (1982–87) 15.804%
■ How do REVA and EVA stock portfolios per-
Top 25 REVA firms (1982–87) 17.013
form? Our extensive empirical tests document that
EVA on its own can predict changes in shareholder Value-weighted market index 12.861

Financial Analysts Journal • May/June 1997 19

returns for the top 25 EVA firms for the 1988–92 organization. Future research in this area should
period. We concluded that REVA could be used to address the important issues of optimal compensa-
compensate senior executives and EVA could be tion design using both EVA and REVA for firms of
used to compensate those at lower levels in an hierarchical organizational design.19

NOTES
1. Noise trading, portfolio insurance, and other factors unre- values of debt and preferred stock approximate market values.
lated to the firm’s performance may induce randomness in 11. See Sharpe (1964).
stock prices. See Milbourn (1996) for a theoretical and em- 12. See Ibbotson (1995).
pirical examination of some of these issues. 13. As in Fama and French (1992), we estimated beta as the sum
2. The theoretical statistical expectation of the abnormal return of the slope coefficients in a regression of portfolio returns
is zero. on the contemporaneous market return and one-period
lagged market return.
3. Articles in the popular press have mentioned the correlation
14. We chose to re-estimate the WACC for each firm each year
of EVA to measures of shareholder value, but the underlying because the WACCs reported in the Stern Stewart Perfor-
research has not been reported. mance 1000 are based on rolling three-year averages.
4. Stern Stewart considers more than 250 accounting adjustments 15. Because beginning-period market value of equity appears
in moving to EVA. In practice, however, most firms find that in the denominator of both the independent and dependent
no more than 15 adjustments are of material significance. variables, the independent and dependent variables may be
5. For nonpublicly traded firms or divisions, a typical ratio of spuriously correlated. As a result, one might find a signifi-
market value to book value for comparable firms could be cant relationship between the independent variables and
abnormal returns even if the underlying variables are un-
used to convert existing book values into market values.
correlated. Because the market value of equity undergoes a
6. That is, the firm’s weighted-average cost of capital is defined nonlinear transformation, however, the spurious correla-
as the financiers’ opportunity cost of capital. tion should be reduced. To control for any residual spurious
7. This disparity between economic book value and market correlation, we ran instrumental variable regressions in
value is representative of the sample of firms used in our which the lagged values of EVARET and REVARET were
subsequent empirical analysis. Specifically, we document an used as instruments. The conclusions drawn from the in-
average ratio of adjusted economic book value to market strumental variable regressions were qualitatively the same
value to be 78.19 percent. as those drawn from the ordinary least squares regressions.
16. Both of these differences are significant at the 1 percent level.
8. Finance companies, for example, may find it appropriate to
17. This difference is significant at the 1 percent level.
use this flows-to-equity approach.
18. These nonparametric results fortify our intuition that a firm
9. As recommended in the calculation of EVA and as we define could have a positive EVA and still be destroying shareholder
REVA in Equation 3, we did not include a firm’s non-interest- value.
bearing current liabilities. 19. We would like to thank W. Van Harlow III for his helpful
10. Following most finance researchers, we assumed that the book comments on an earlier draft of this article.

REFERENCES
Dimson, Elroy. 1979. “Risk Measurement When Shares Are Milbourn, Todd T. 1996. “The Executive Compensation Puzzle:
Subject to Infrequent Trading.” Journal of Financial Economics, Theory and Evidence.” IFA Working Paper No. 235, London
vol. 7, no. 2 (June):197–226. Business School.
Fama, Eugene F., and Kenneth R. French. 1992. “The Cross- Sharpe, William F. 1964. “Capital Asset Prices: A Theory of
Section of Expected Returns.” Journal of Finance, vol. 47, no. 2 Market Equilibrium under Conditions of Risk.” Journal of
(June):427–65. Finance, vol. 19, no. 3 (July):425–42.
Ibbotson Associates. 1995. 1995 Yearbook of Stocks, Bonds, Bills, Smolowe, Jill. 1996. “Reap as Ye Shall Sow.” Time, vol. 147, no.
and Inflation. Chicago: Ibbotson Associates. 6 (February 5):45.
Jensen, Michael, and Kevin Murphy. 1990. “Performance Pay Stewart, G. Bennett III. 1991. The Quest for Value. New York:
and Top-Management Incentives.” Journal of Political Economy, Harper Business.
vol. 98, no. 2 (April):225–62.

20 Association for Investment Management and Research