-
The Role of the CFO, Performance Measurement, and Incentive
Design
THE FIRST PART OF THE BOOK developed the theory of finance from
first principles-the theory of choice (utility theory) and objects
of choice (investment frontiers that were constructed of uncenain
choices in a multiperiod setting). The equilibrium asset pricing
theories that resulted are extremely useful tools for financial
decision makers at the microeconomic level and shall be used
throughout the remainder of the book. Now, however, we shift our
orientation to the practitioner. We take the decision maker's point
of view. In this chapter, we tackle three questions. How sho.uld I
measure the performance of my company vis-a-vis competitors? What
drives my stock ppce? How shall I design ineentive.s to reduce the
agency costs between pwner, lll!lll{l,ger,' and, gtll.er s
hold,ets?. In r:remaitring pters we discuss the financial economics
ofthe following questions (as well as mall others). liow much is a
target company worth in a takeover? Hmv canllegally avoid taxes and
how much is it worth to do so? How should J manage the mix of debt
and equity to maximize the value of my firm? What dividend or
sharerepurchase policy should I recommend? Is leasing better than
borrowing? How do I manage my company's pension fund? How do I make
deci.sions about allocating our resources abroad? How hould I
manage my compap.y1s risk?
The two pansof the bboktire interactive andteffective. Whenever
possible, we shall point . out where the theory of finan helps
practitioners and where there are gaps--either because the theory
is inconsistent with practice or where the theory simply is devoid
of content to aid day-to-day decisions. Formultiperiod irfye
iltlent dedsio in th ( ce o f ncmainty the developments made in the
appli 18. ! re ;} t i!Yive;h:l Jlk,IJIJquSly. We seriously
considered moving thatAAI nano .. boel,Jmt i t i so rondaifienW we
put it up front
This chapter wjtJt, o f .the e f llllCi pfficer ( FO) of a large
multibusiness,IDliJtinatioo11l COifii'WlJ}lll\ : \iide yariety o f
decisions and reSP9IlSibilities':that he o r she may b a v f
I } ' i n tl}e (lotn irf Qf financial economics, for
; iamp1e, capital .
y. 0 t h clpsely lilted, but somehow missing
-
470 Chapter 13: The Role of the CFO, Performance Measurement,
and Incentive Design
from textbooks, for example, perfonnance measurement and
incentive design. Others are learned on the job, for example,
investor relations and how to manage the budgeting process. So
let's begin. What does a CFO do anyway?
A. (~e Role of the Chief Financial Officer In the middle of the
last century, the 1950s and 1960s, the CFO was primarily a
financial record keeper, in charge of internal accounts and of the
company's financial reporting to shareholders and regulators. This
role is still common in family-owned businesses in Asia, Latin
America, Africa, and Eastern Europe. But in large multibusiness,
multinational companies such as Toyota, International Business
Machines, British Air, Deutsche Bank, CVRD, Telefonica, and
Norilsk, the CFO is one of the top three decision makers, standing
alongside the.chief executive officer (CEO) and the chief operating
officer (COO), Figure 13.1 attempts to display the decisions for
which the CFO may be responsible. Most finance textbooks stress the
first two categories: investment and financing. However, there is
an important third category, namely, managerial decisions such as
performance measurement and incentive design. Let's review each
category in greater detail.
Figure 13.1 Decisions for which the CFO may be responsible.
Investment decisions Large capital expenditures
Research and development Mergers and acqusitions Ownership
structure Capitaleffidency Working capital
management
Management decisions
Financing decisions
-
B. Performance Measurement 471
Managerial decisions made by the CFO include the performance
measurement of the business units of the firm, setting and
reviewing budgets, designing incentives that are perf onnance
com-patible, and investor relations to communicate aspirations and
results to the external community. The CFO is also usually
responsible for being sure that the company conforms to all
regulations (environmental, health and safety, tax, agency, and
legal). Other top executives from planning, budgeting, human
resources, and legal staff will report to the CFO on these matters.
Much of the remainder of this chapter will deal with performance
measurement and incentive design.
The second area is financial decisions. The CFO is responsible
for the audited financial state-ments of the firm; hence the
comptroller reports to him as do the external auditors. He is also
responsible for the sources and uses of funds. This means that he
must recommend capital stnJ.c-ture (the mix of debt and equity) and
dividend policy (the percentage of dividends paid out) to the board
of directors. For this purpose, the corporate treasurer usually
reports to the CFO. Closely related is the risk management
function, which of course includes financial risk. The CFO is
usu-ally responsible for the insurance position, the hedging
position, and the net risk exposare of shareholders. He will also
deal with bond rating agencies that provide an assessment of the
credit
risk of the company. Also, the chief planning officer often
reports to the CFO. There are many types of plans, for example, tax
plans, short-term (annual and quarterly) budgets, and strategic and
long-range plans.
Finally, there are investment decisions. The CFO is often
responsible for reviewing all capital expenditures above a certain
limit, say, $2 million. A myriad of details are involved: What
methodology should be used (traditional net present value or real
options analysis)? How should the cost of capital be adjusted for
differences in project and country risk'? How should cash flows be
defined? For firms with substantial research and development
budgets, the CFO is often assigned the responsibility for final
allocations. In addition to internally generated growth, most firms
have a development officer in charge of mergers and acquisitions,
joint ventures, and divestitures, who reports to the CFO. Closely
related are issues that affect the ownership structure of the
firm-dilution of ownership value, violation of debt covenants,
equity carveouts (initial public offerings of ownership in a
business unit), issuance of tracking stock (whose value is based on
the income of a business unit), and issuance of executive stock
options. Finally, there is working capital management (inventory,
payables, and receivables policy).
The remainder of this chapter covers performance measurement,
budgeting, communicating with the external investment community,
and incentive design. The remainder of Part II of the book covers
most of the other topics.
B. 9erforrnance Measurement 1. Measures Based on Earnings
Performance measurement is one of the most important management
responsibilities of the CFO because it subtly affects the way
people behave. Owners of the firm want perfonnance measures to be
aligned with maximizing shareholder wealth, a goal that is easy to
articulate but difficult to implement.
Figure 13.2 defines a wide variety of performance measures that
companies can use. Choice of the one to actually use is not
immediately obvious. Let's go through them one by one. First, we
have earnings per share or growth in earnings per share. It suffers
from every knock-out criterion. It contains no balance sheet
information-a deficiency that implies that if one firm requires
two
-
472 Chapter 13: The Role of the CFO, Performance Measurement,
and Incentive Design
Figure 13.2 Comparison of perfonnance metrics. ...
0 i::: :: 0 ,::
;;z .c ,:: 0 .J:J fl:) >.
ell .::r::: ..... .... ,2 i::: E-< :s .s ;:: i
"' ;;1 ~ 0 'i > 0 0 ... t.:: .c ,:: 13 (/) .c 0
"'
... ::I 0 ,:: e3 u II) ] I i::: e > o:! B ,:j ::I 0 .J:J 0 u
t:: ,.'!.i u 0 0 i i::: ..c: a ~ z fl:) ~
'
Earnings per share EPS =NII Number of shares Growth in earnings
per share G(EPS):: (EPS1-EPS1_ 1) /EPS1
--
Return on invested capital ROIC = EBIT II Return on equity
ROE=NI/E Spread ROIC WACC
EP (ROIC WACC)I Economic Profit or EV A Grow1hinEP (EP1-EP1_ 1)
/EP1_ 1
---
Expectations-Based-Management EBMTM = Actual EP - Expected EP
--
Discounted Cash Flow DCF ---
Real Options Analysis ROA Most complicated measure
dollars of capital to generate one dollar in earnings while at
the same time another requires only one dollar of capital to
produce the same earnings, then the market would assign the same
value to both because !;hey report identical earnings. Take a look
at Figure 13.3 for an example.
The first company, called Longlife, invests $3,000 every three
years in equipment that lasts three.years. The other company,
called Shortlife, invests $1,000 every year. They also have timing
diffr-rences in working capital investments, but other than that
they are identical. They both are assumed to have a 10% cost of
capital. In particular, their net income is exactly the same each
year. Therefore, if one were to believe that value depended only on
net income, then they should have the same value.
We have already shown, in Chapter 2, that value depends on
discounted cash flows (net income plus depreciation minus capital
expenditures minus investment in working capital). Although the
undiscounted total cash flows are $1,800 for both companies, there
are significant timing differences. Consequently, the DCF value of
Longlife is $808, while for Short1ife it is $1,296-a 60%
difference. This clearly illustrates why net income or earnings per
share is an inadequate performance measure.
Another problem is that earnings is short tenn in nature. We
once sat in a division manager's office while she received a phone
call from her CEO asking that her division supply more
fourth-quarter earnings so that the firm could meet its year-end
consolidated earnings target. After hanging
-
B. Performance Measurement 473
Figure 13-3 Two companies with identical earnings but a 60%
difference in value.
Longlife 2 3 4 5 6 Total
Revenues 4,000 4.000 4.000 4,000 4.000 4.000 24.000 Cash costs
2,000 2,000 2.000 2,000 2.000 2,000 12.000 Depreciation 1,000 1,000
1,000 1,000 1,000 1,000 6,000 Taxes@ 50% 500 500 500 500 500 500
3,000
Net income 500 500 500 500 500 500 3,000
Capital expenditure 3,000 0 0 3,000 0 0 6,000 Increase in
working capital 600 50 50 300 100 100 1.200
Cash flow -2,100 1,450 1,450 1,800 1,400 1,400 1,800
Discount factor .909 .826 .751 .683 .621 .564 Present value
-1,909 1,198 1,089 -1,229 869 790
Value of Longlife given 10% cost of capital = 808
Shortlife 2 3 4 5 6 Total
Revenues 4,000 4.000 4,000 4,000 4,000 4,000 24,000 Cash costs
2,000 2,000 2,000 2,000 2,(X)(J 2.000 12.000 Depreciation 1,000
1,000 1,000 1.000 1.000 1.000 6,000 Taxes@ 50% 500 500 500 500 500
500 3,000
Net income 500 500 500 500 500 500 3,000
Capital expenditure 1,000 1,000 1,000 1,000 1,000 1,000 6.000
Increase in working capital 200 250 250 100 200 200 1,200
..... -~--
Cash flow 300 250 250 400 300 300 1800
Discount factor .909 .826 .751 .683 .621 .564 Present value 273
207 188 273 186 169
Value of Shortlife given 10% cost of capital = 1,296
up the phone she remarked that she would supply the earnings by
slashing advertising expenses, but would lose customers during the
following year and would have to pay great expense to win them
back. Although she believed the dictum from above would destroy
value, ~he did in fact produce the required extra earnings.
Finally, as we shall soon see, neither earnings per share nor the
growth in earnings per share is highly correlated with the total
return to shareholders.
2. Measures Based on Rates of Return Return on invested capital
(ROIC) is comprehensive because it is the product of two key value
drivers (operating margin and capital turnover). The definition of
pretax ROIC is
EBJT sales ROJC:::: - x ----sales invested capital
(13.1)
-
474 Chapter 13: The Role of the CFO, Performance Measurement,
and Incentive Design
Figure 13-4 ROIC tree.
x
Operating margin = Operating profit I sales
ta! turns= sales I invested capital
Physical capital +
Operating cash +
Inventories +
I Accounts receivable
Accounts payable
Materials cost
Energy cost Number of sales personnel
Sales force Compensation package
Staff Sales per salesperson
Capacity utilization
Design specifications
Credit policy per product line
Delinquent accounts
Method of payment
where EBJT is earnings before interest and taxes. When used to
evaluate business unit performance, however, ROIC encourages
capital harvesting behavior. It is easier for a manager to allow
the capital under her control to depreciate than it is to invest
new capital profitably. As she harvests the business, ROIC rises
because the amount of invested capital is assumed to drift upward.
What happens, however, before much time passes, is that the capital
base devalues and, assuming that the decline in sales is not as bad
as the decline in the capital base, then ROIC will be fine. The
spread of the ROIC over the w~ighted average cost of capital (WACC)
has the same problem. Management cannot influence WACC, so the
easiest thing to do to raise ROIC is to allow the base of invested
capital to depreciate-that is, to harvest the business.
As part of expectations-based management, however, ROIC is
useful because it traces back into all of the line items in the
income statement and balance sheet as shown in Figure 13.4. It is
possible to trace individual lines on the financial statement all
the way down to individual product line information for the purpose
of performance measurement.
Return on equity (ROE) is also a poor performance measure.
First, it is possible to artificially raise the return on equity by
using 100% debt to purchase another company that earns less than
its cost of capital but more than the cost of debt The result will
be to increase the return on equity, but to increase risk faster,
thereby decreasing the price per share of the company:
ROE= (EB/T-kbD)(l-T)/E, (13.2)
-
where
B. Performarice Measuremerit 475
k1, == the interest rate good on the face value of debt,
D the face value of debt. T the cash tax rate, E == the book
value of equity.
3. An Economic Profit Measure The definition of EVA, a trademark
of Stern Stewart, also called economic profit (EP), is the spread
between the return on invested capital (ROIC) and the weighted
average cost of capital (WACC)multip!ied
bythedollaramountofinvestedcapital (l).EP = EVA = (ROIC - WACC)l.
It helps to alleviate harvesting behavior by multiplying the spread
by the amount of invested capital. One often reads that when
economic profit is positive, that is, when a business unit earns
more than its weighted average cost of capital, the business unit
in question creates value for
. shareholders. It turns out that positive EVA is neither a
necessary nor sufficient condition for creation of shareholder
value either at the company or the business unit level.
Furthermore, the level of EVA as well as the growth in EVA are not
highly correlated with the total return to shareholders (TRS). More
detail will be given shortly. Perhaps an example will sutlice in
the meantime. In October 1998, Intel, a computer processor
manufacturer. which was earning roughly 50% return on invested
capital. and which had a cost of capital equal to roughly 10%.
reported that its earnings were up 19% over the year before.
Clearly, Intel's EVA was positive, as was its growth in EVA. Yet on
the announcement, its share price fell 6%. Why? Because the
consensus analyst expectation was that earnings were to have gone
up 24%. Hence. expectations were not met, and Intel's price was
adjusted downward. EVA does not attempt to incorporate
expectations.
Chapter 14 shows that the discounted value of EP. when added to
the book value of assets in place, equals the DCF value of the
firm. A short example is shown in Table 13.1. Either way. whether
we look at DCF orat discounted EPplus assets in place. the value of
the entity is $1.088.60. Given this value, the expected return to
investors is 10% (i.e., WACC). To create value in year I, for
example, EP must exceed the $50 that is expected, or EBIT (I T)
must exceed $150. The same requirement is true for any and all
years. For example, in year 2 if EBIT (1 T) turns out to be $200
instead of $220, the value of the firm will fall, even though ROIC
> WACC and EP is positive.
4. Measurement Based on Expectations To create shareholder value
it is necessary and sufficient for a company or a business unit to
exceed shareholder expectations. As shown in Figure 13.2,
expectations-based management (EBMr" is a trademark of the Monitor
Group) is defined as the difference between actual and expected
economic profit:
Actual EP Expected EP == [Actual ROIC - Expected ROICJ x l
[Actual WACC Expected WACC] x I
+ [ROIC WACC][Actual I Expected I]. (13.3)
Equation 13.3 has three parts. The first may be interpreted as
earning more than expected on the company's invested capital,/. The
second says that value is created when the actual cost of capital
is lower than the expected cost of capital. The third term says
value is created when the company
-
476 Chapter 13: The Role of the CFO, Performance Measurement,
and Incentive Design
Table 13.1 Comparison of EP with DCF for Valuation After-tax
Discount
Year ROIC EBIT(l - T) WACC EP FCF Factor PV(FCF) PV(EP) 0 $1,000
10% 1.000 $1,000.0
15% $1,100 $150 10% $50 $50 0.909 $45.5 $45.5 2 20% $1,300 $220
10% $110 $20 0.826 $l6.5 $90.9 3 10% $1,400 $130 10% $0 $30 0.757
$22.5 $0.0 4 5% $[,800 $70 10% -$70 -$330 0.683 -$225.4 -$47.8 5
10% $2,000 $180 10% $0 -$20 0.621 -$12.4 $0.0
CV 10% $200 10% $0 $2,000 0.621 $1,241.8 $0.0
$1,088.6 $1,088.6
Notes: gcv 5.0% (continuing value growth) FCF free cash flow
EBIT (I - T) - changes in new (J) investment CV= continuing value
EBIT (1-T)(I- gcy/r)/(WACC- gcv)
$200(1- .05/.10)/(.1 - .05) = $2,000 I= book value of assets
WACC l 0% ( weighted average cost of capital)
invests more than expected and does so profitably. In other
words, it earns more than the cost of capital on new capital
invested. Overall, expectations-based management is the best
short-term measure of management performance. It is also the
measure that is most highly correlated with the total return to
shareholders.
Table 13.2 shows the results of regressions where the annual
market-adjusted total return to shareholders is the dependent
variable and the independent variables are the earnings per share
(EPS) .scaled by the share price at the beginning of the year in
the first row, the growth in EPS scaled by the
beginning-of-the-year share price in the second row, the economic
value added (scaled) in the third row, and the growth in economic
value added (scaled) in the fourth row. The sample was taken from
the S&P 500 for the years 1992-1998. None of the r-squared
statistics exceed 6%, although according to the i:statistics that
are shown in parentheses, all of the independent variables are
statistically significant.
Table 13.3 brings analyst expectations into the picture. As
before, the dependent variable is the total return to shareholders,
adjusted for the market return the same year. The equation for the
market-adjusted return, MARi,,
MAR _nn+ JI -
CT( I+ 'm,) ( 13.4)
takes the ratio of the product of the one-month returns for the
ith stock in a given year and divides it by the product of the
monthly market returns in the same year. This is then regressed
against three expectations-based variables in the following
structural equation:
MAR;1 =a+ b ln[E(l, 1)/ E(l, 0)]+ cln[E(2, 1)/ E(2, 0)]+ d[E(L,
I) - E(L, O)]+ E11 (13.5)
-
B. Performance Measurement 477
Table 13.2 Market-Adjusted Total Return to Shareholders and
One-Period Return Measures A B c D
Intercept -0.099 -0.0441 -0.003 -O.lJ04 (14.02) (-8.33) ( 5.59)
(-6.871
EPS,/SH J.086 (12.89)
!::,,PS,/S1_ 1 0.669 (11.10)
EVA/S1-1 0.001 (6.786)
!::,,VA/S1-1 0.003 00.02)
Number of observations 2,582 2.579 2,194 2.185 Adjusted
R-squared 0.06 0.05 O.D2 0.04 F-statistic 166.1 123.3 46.1
100.4
Source: Copeland, Dolgoff, and Moel (2003) Panel data forthe
S&P 500 compar.ies for the years 1992-1998. EVA data from Stem
Stewart. Company EPS and EPS growth data from Compustat, market
return data from CRSP. 1-stntistics in parentheses.
The first term is analyst expectations about this year's
earnings as revised this year (between time O and time 1 ). The
second term is the revision of analyst expectations of earnings
next year, but observed this year. In other words it is the
difference between analyst expectations of next year's earnings at
the end of this year versus the expectations of earnings next year
at the beginning of this year. The natural logarithm of the first
two tenns is used because the data are supplied in units of dollars
per share and we wanted to convert to percentages. The third term,
the change in expectations about long-term growth as observed this
year, was already given as a percentage. The results of this
multiple regression are given in Table 13.3.
In this regression the r-squared is 47% when all variables are
included (column I). When only two independent variables at a time
are included, the lowest r-squared is 28%. Clearly, the total
return to shareholders and changes in analyst expectations are
closely related with each other. Upon further examination, the
first column of Table 13.3 tells us that there is no significant
relationship between changes in expectations this year about this
year's earnings and TRS, when longer-tenn expectations are also in
the multiple regression. However, changes in expectations this year
about next year's earnings and about long-term earnings growth are
both highly significant. Even more interesting is the fact that the
impact of changes in expectations regarding long-tern1 growth
(3.269) is 8.4 times larger than the impact of changes in
expectations about earnings next year. If the value of a company in
the market is conceptuaiized as a discounted cash flow, the
multiple regression makes perfect sense because most of the value
is derived from cash flbws beyond the first two years.
So far we have only looked at changes in earnings expectations.
What about the cost of capital and capital expenditures? They also
appear in Eq. (13.3). Unfortunately, most analysts do not record
their expectations about these two additional variables. Therefore,
we have to build our own expectation models. The CAPM was used to
estimate the difference between the beginning and end-of-year cost
of equity and that was assumed to be the change in expectations
about the cost
-
478 Chapter 13: The Role of the CFO, Performance Measurement,
and Incentive Design
Table 13,3 Market-Adjusted Total Return to Shareholders and
Changes in Analyst Expectations A B c D
Intercept 0.360 0.430 0.420 0.461 (18.108) (19.81) (18.88)
(20.74)
ln[E(l, 1)/E(I, 0)) O.Dl8 0.324 (0.94) (22.25)
ln[E(2, 1)/ (2, 0)] 0.389 0.280 (18.22) (23.25)
E(L, 1) E(L, 0) 3.269 4.105 (16.08) (19.79)
In(!/ S1_ 1) 0.117 0.142 0.139 0.159 (18.82) (20.80) (19.88)
(22.85)
Number of observations 2,318 2,491 2,365 2,560 Adjusted
R-squared 0.47 0.31 0.32 0.28 f-statistic 512.9 567.7 562.1
489.3
!\0111re. Copdand. Dolgoff, and Moel (2003). Panel data for the
S&P 500 companies for the years 1992-1998. fapeclation, uala
frum Zach, market return data from CRSP. I-statistics in
parentheses.
of capital. Changes in expectations about capital expenditures
were estimated by calculating the difference between actual capital
expenditures as a percentage of sales and capital expenditures
predicted by a simple time-series regression based on the last five
years of history. The results. shown in Table 13.4, indicate that
percentage changes in expectations of the cost of equity ks,t+I are
statistically significant and have a negative sign as predicted.
Unexpected changes in capital expenditures, however, are not
significant.
What are the implications of expectations-based management for
CFOs? First of all, perfor-mance of business units must exceed
expectations in order to create shareholder value. Consider the
following example. Two business units have the same cost of
capital, let's say 10%. During the past year business unit A earned
8% ROIC and business unit B earned 15%. Which created more value
for shareholders? The answer is that we cannot make any judgment
unless we know what the business units were expected to earn. If,
for example, unit A was expected to lose 5%, then it has exceeded
expectations by 13% and created shareholder value. On the other
hand, ifunit B was expected to earn 25%, then it has fallen short
of expectations by 10% and destroyed shareholder value. Slightly
more subtle are decisions that involve new investment. Again, let's
take a simple example. Suppose that a company has informed the
market that it has two new investments, each expected to earn 30%
while the cost of capital is 10%. The market believes what it has
heard and incorporates the good news into the share price, which
has risen as a consequence. Suddenly, management learns that one of
the two projects will earn only l 5% instead of the expected 30%.
Should management cancel the new project? This is a thorny problem,
requiring careful thought. If management does proceed to undertake
the project, then the market's expectations will be revised
downward and surely the stock price will fall. Remember though,
that every decision in economics must be weighed in light of the
next-best alternative, which in this case is deciding not to go
ahead with the 15% return project If management decides not to
invest, then shareholders are presumed to earn exactly the
opportunity cost of capital, namely, 10%. Consequently, if the 15%
project
-
B. Performance Measurement 479
Table 13-4 Market-Adjusted Return and Expectations, the Cost of
Equity, and Capital Expenditures Independent Variable A B c D E F
G
Intercept 0.009 0.360 0.359 0.358 0.359 0.361 0.361 (2.23)
(!8.11) (17.73) ( 17.96) (17.74) (1333) i13.67)
ln[E( I. 1)/ E(l, OJI -0.006 0.018 0.026 0.017 0.026 0.01 i
0.007 (-0.34) (0.94) (l.39) (0.92) (l .37) (0.48) (0.33)
ln[E(2, IJ/ (2, O)] 0.465 0.389 0.391 0.391 0.390 0.416 0.421
(22.30) (18.22) (18.19) (18.28) (18.15) (15.89) ( 16.26)
E(L, I) - E(L, 0) 3.492 3.269 3.263 3.273 3.260 2.771 2.679
(17.73) (16.08) (15.27) (16.10) (15.27) (I 0.36) (10.38)
ln(l/Sr-1) 0.117 0.118 0.118 0.118 0.123 0.123 (18.82) (18.60)
(18.83) (18.59) (14.70) ( 15.06)
(kr+I - kr)/ k1 -0.087 -0.014 (-2.30) (-0.30)
(rJ;r+l (r1.r+1))/ E(rr,+1) -0.036 (-1.53)
(/Jt+J /31)//3, -0.128 (-2.99)
UCAPEX -0.002 --0.002 (-0.38) (-0.40)
Number of observations 2,699 2,318 2,185 2,318 2.185 1,384 L457
Adjusted R-squared 0.38 0.47 0.48 0.47 0.48 0.47 0.47 F -statistic
558.7 512.9 403.8 411.0 405.2 206.0 254.6
Source: Copeland, Do!goff, and Moel (2003). Panel data for the
S&P 500 rnmpanies for the years 1992-1998. k, is the n,D.rket
cost of equity. 11.,+1 is the ;1cmJl lO-year Treasury spot rate at
the end of the year, E (r f.r+ 1) is the one year forward I 0-year
rate as of the beginning of the year. Betas are BARRA betas.
UC.4PEX stands for scaled unexpected capital expenditures.
Expectations data from lacks. market return data from CRSP.
I-statistics in parentheses.
is rejected, shareholders will earn only 10% and the stock price
will fall even more. Therefore, when it comes to new investments,
the rule of thumb remains the same as in Chapter I. Maximi:e
shareholder wealth ljy accepting all new investments that earn more
than the market-determined opportunity cost of capital.
Figure 13.5 shows a chart with data for Chevron, one of the
world's largest integrated oil companies from 1990 to 2000. Notice
that in 1994-95 earnings rose from approximately $2.50 per share to
$3.00 per share. Nonnally, we would expect this to be good news,
but the total return to shareholders, relative to the market, fell
the whole year. The rationality of this is difficult to understand
until analyst expectations are plotted on the same chart. The
squiggly line that starts in January 1993 and ends in the square
earnings box in December 1995 represents consensus analyst
expectations. In January 1993 analysts were expecting about $3.50
per share for 1995. During the two years that followed they
continuously reduced their expectations, and it was this downward
revision that matched the decline in the market-adjusted total
return to shareholders. Note also that during 1995, expectations of
earnings the following year (1996) were also adjusted downward,
although in early 1996 they began to turn around.
-
480 Chapter 13= The Role of the CFO, Performance Measurement,
and Incentive Design
Figure 13.5 Chevron 1990-1998:TSR,annual earnings, and market
expectations. Source: Monitor Analysis, Zacks, Compustat.
Chevron Corp. Market-adjusted TSR vs. analyst earnings
estimates, 1991 1998
5.5 IA
5 l.2
4.5
4 [/) Gl 3.5
3
2.5
2 Note: EPS and analyst expectations 02 exclude extraordinary
items
1Tmmmmmrmmrrrrrnmmmmmmmmmmmrnmmmmmmmmmrrrmmmmrrrrr!O 1.5 0 0
.... .... N N (") (") 'SI' 'SI' II") II") IC \0 I:" I:" 00 00
OI
OI ....,
~ ~ OI OI OI ~ OI ~ OI OI OI OI OI OI OI OI OI OI I . . I I . I
I I I I I I ; "3 c:: .... c:: '3 c:: "3 c:: '3 c:: '3 a '3 ; '3 '3
(,;! ::: (,;! (,;! (,;! el -,
...., ..., ...., .., ...., ..., ...., ...., ...., ...., ...,
...., ..., ...., ..., ...., ...,
The art of setting expectations requires that management
communicate appropriately with investors in the market place, and
formulate and monitor performance vis-a-vis expectations
internally. Let's discuss external communications first. Define the
signal-to-noise ratio as
SNR = [actual EP - expected EP]/a(analyst). (13.6)
The signal, in the numerator, is the difference between actual
and expected economic profit. Noise, in the denominator, is the
dispersion of analyst expectations (measured by the standard
deviation).
For a given strength signal, we hypothesize that as the amount
of noise increases, there will be three effects. First, greater
noise implies greater variation in the resulting message, namely,
the total return to shareholders. Second, the reaction of share
prices to any single message becomes smaller because the
information in the message becomes obscured by the noise. For
time-series data these two effects will be difficult if not
impossible to separate. The third effect is the relationship
between the signal-to-noise ratio and the level of TRS. Since
greater noise implies greater variance in TRS and greater
difficulty interpreting the signal, we expect that the m.irrket
will require higher returns on average in order to be compensated
for poorer-quality signals. The conclusion is that if management
seeks to maximize shareholder wealth, it should attempt to reduce
noise when communicating with the market. In practice this implies
an unbiased signal. If there is a bias ( e.g., consistently setting
expectations lower than management expects it can achieve), and if
the bias is constant, then the market simply adjusts for it and the
resulting signal is the same as if there had been no bias at all.
The opposite is an uncertain bias-sometimes pessimistic and at
other times optimistic. This lack of constancy results in a noisier
signal and consequently a lower share price as investors require a
higher expected return.
Not only is it advisable to have an unbiased signal, but it is
wise to pay attention to other attributes. If multiple members of
the top management team are responsible for communicating with the
market, their messages should be mutually consistent-to reduce
noise. The messages should be verifiable through independent
sources (such as customers, suppliers, competitors,
-
B. Performance Measurement 481
regulators, and former employees of the company). And there
should be "color" to the signal. In other words it should contain
in-depth explanations for what is being communicated.
The complement to external communications with investors is the
process of building and monitoring internal performance
expectations. Finally. it is desirable to align compensation with
performance-a topic thar is discussed in greater detail later in
this chapter. Usually the budgeting process is the way that
expected performance is established and monitored. Without delving
far into an unnecessary level of granularity, it is worth simply
stating a few rules of thumb that set apart good from bad budgeting
practices:
1. Redo the budget whenever the operating environment changes.
2. Work to set reasonable stretch targets. 3. Assign decision
rights to the people who have the right skills and necessary
information. 4. Don't overreach a manager's span of control. 5. Be
aware that the process of setting expectations is not necessarily
incentive compatible with
actual performance. 6. When decision rights are assigned, be
sure to assign responsibility and accountability with them. 7. Be
aware of externalities among business units and seek market-based
solutions. 8. Be aware of multi period problems with measuring
performance and designing incentives.
The first of the eight rules of thumb is almost always violated
by bureaucrats who insist that budgets be submitted at the same
time each year whether revision is needed or not. The reality is
that when information anives that materially changes the
opportunity set facing a company. then it is time to revise the
plans that are embedded in the budget and to resubmit. An annual
budgeting cycle is meaningless most of the time. Rules 2 and 5 are
closely related. Michael Jensen [2002] has been quoted as saying
that "the budgeting process is just an incentive for lying: The
logic is straightforward, but not compelling. Lower-level
management has an incentive to set cxoectations low-low enough that
it becomes easy to surpass easy benchmarks-a process called
sandbagging. There are two ways to overcome this problem. The first
is to be better informed. In order to have a two-way discussion
between top management who must set expected performance standards,
and lower-level management who must perform, it becomes necessary
for top management to take the time and effort to understand the
business that they are evaluating. Critical information may be
gathered from former managers of the business, from suppliers, from
customers, from engineers. and from government agencies. An
informed discussion helps to reduce sandbagging. The second piece
of advice is to create an incentivi;-compatible compensation
design. There is more on this point later on in the chapter. Rules
3, 4, and 6 are also related. Together they imply that the manager
of a business unit should have complete control over its resources.
For example, a performance measurement system based on earnings
assigns no responsibility for managing capital efficiently. It is
hardly surprising that when managers are not charged for using the
firm's capital, that the sales revenue per dollar of capital
employed begins to decline. Rules 7 and 8 are easy to state but
difficult to implement.
5. Measurement Based on DCF Valuation The main shortcoming of
all of the aforementioned perfonnance measures is that they are
basically one-period views of the world. Consequently, at the
corporate level and the business unit level, where it is reasonable
to assume that forecasts of both the income statement and of the
balance
-
482 Chapter 13: The Role of the CFO, Performance Measurement,
and Incentive Design
sheet information are available, it is advisable to take a look
at discounted cash flow valuations. The expected cash flows in
these models might be based on analyst forecasts and provide a
market perspective of the company's value, or they might be based
on management forecasts to provide an inside-out perspective. If
the difference between the two is significant, it may lead to
board-level decisions. For example, if the market value is less
than management believes the stock is worth, then a share
repurchase program can be initiated.
Because valuations are multiperiod in nature, they are also
useful for setting one-period per-formance targets, when one is
using an expectations-based management system. To create value it
is necessary to exceed expectations, and these are not always
constant over the life of a busi-ness unit. Often there is a
start-up period following a major capital expenditure when the
expected ROJC is low or negative, followed by a normal return
period, followed by higher expected returns when the capital base
depreciates. If these life cycle stages are predictable, then it is
not diffi-cult to set expectations of performance in an appropriate
manner. A similar story can be told for tum-around activities and
cyclical businesses. Still other businesses perpetually earn high
ROIC because they do not use much capital, for example, a
consulting firm. In all instances, the mul-tiperiod a,;pect of DCF
is useful when used in conjunction with the short-term measures
listed earlier in the chapter.
6. Measurement Based on Real Options Real options analysis is a
superset of DCF, and all that was said there is applicable here.
Most companies gain insight from ROA primarily at the project
level, although decisions concerning whether to exit and reenter a
line of business, and certain aspects of mergers and acquisitions
(the option to expand or abandon an acquired business), are common
applications.
This section discusses the design of incentive systems. Given
that there is often a separation of ownership from the control of
operations, how can incentive design help to alleviate the
resulting agency cost? For example, you might own a farm, but fail
to possess the farming skills necessary to own it: Naturally, you
hire a farmer to run the day-to-day operations. How should you
compensate her? If you use a form of profit sharing, the farmer
will maximize this year's crop output. But as owner that may not be
in your best interest because it requires overfarming the land and
therefore exhausting it for future use. The result is that there is
a conflict between maximizing this year's crop yield and maximizing
the value of the farm for resale.
There is a short list of thorny issues in incentive design.
First, how should pay be linked to performance, given the objective
of shareholder (i.e., owner) wealth maximization? Second, given
that executives change jobs frequently, how should compensation be
distributed over time? Third, how should incentive design vary
between line officers and staff, and as one moves from top
management to middle-level management and lower'?
1. Why Maximize Shareholder Wealth? Before diving into these
topics, there is first the question of why it makes sense to
organize resources into corporations with a limited liability
residual claimant called the shareholder. Why not organize as an
Athenian democracy where all stakeholders can vote on policies and
decisions, or into an
-
C. Incentive Design 483
autocracy run by an absolute dictator, or into a socialist firm
where all ownership rests in the hands of workers who are
represented by decision-making committees? The broad answer is that
a firm may be thought of as a set of contracts designed to share
risk and maximize output. As residual claimants (with limited
liability) the shareholders need to satisfy all other
higher-priority claimants and to accept all residual risk. In order
to maximize their own wealth, shareholders must be sure that
customers are well cared for, that the appropriate labor force can
be recruited and retained, that suppliers are content, that bank
covenants are adhered to, that interest and debt principal payments
are made as promised, and that the government receives the taxes
that are due. Suppose that there were no residual claimant to whom
decision rights are assigned. If so, then stakeholders with limited
information might take the helm. For example, employees who are
typically risk averse find it difficult to separate the risk of
their human capital (e.g., losing their jobs) from risk taking in
business and, consequently, they might be overly conservative
decision makers. The assignment of residual risk to shareholders
who are free to sell their shares in an open market and free to
spread their company-specific risk across diversified portfolios,
and who do not have their human capital tied up in the corporation,
results in efficient risk sharing. The fact that they have control
of the company that they are said to own gives them decision rights
that are aligned with the interests of all of the stakeholders
ahead of their residual claim.
Jensen [ 1998] adds four related attributes to these basic
themes. First, when risk changes over time (as it always does),
then it is costly to write and rewrite contracts for sharing rules
among the diverse claimants of the firm. Second, shareholders who
are not employees. and who are granted limited liability by the
government, find that it is easier and less cost! y for them to
bear the company risk than it is to write detailed contracts
regarding risk sharing. It has been argued that common stock (i.e.,
wealth from residual claimants) is ideal for outright ownership of
assets by a firm. having lower transactions costs than rental
contracts for use of the same assets. Third, residual claims that
are alienable (i.e., that can be sold and bought) make it easier to
separate ownership from control. This, in turn, makes it easier to
hire managers with firm-specific skills because they are not
required to tie up both their human and their financial capital
under the roof of the firm for which they work. Finally, the
existence of shares as residual claims does not hamper decision
making of management because the separation theorem discussed in
Chapter I of this textbook can be applied-namely, take all projects
until the marginal rate of return on the last dollar spent just
equals the market-determined cost of capital.
Often it is alleged that the existence of residual claims held
by shareholders creates conflicts with other stakeholders,
especially labor. The allegation is that greater profits for
shareholders are obtained by laying off labor. Copeland, Koller,
and Murrin [ 1994] study the relationship between employment growth
and shareholder wealth creation. over the 1983-1991 time interval.
Their sample is collected at the company level and aggregated by
industry in three countries: Japan, Germany, and the United States.
They find no evidence that labor suffers to benefit shareholders.
Within an industry, and in the long run, companies that are winners
are more productive, create more shareholder wealth, and experience
greater growth in employment. In the auto assembly industry, for
example, all countries experienced growth in employment, but the
Japanese, who were the most productive, had the greatest growth in
employment. Similarly, in steel, an industry with overcapacity and
shrinking employment due to technological changes, Germany had
greater shareholder wealth creation and better employment results
than the United States. The empirical evidence clearly indicates
that creating an organization where residual claims and control of
the company both are held by shareholders actually aligns the
economic interests of all stakeholders. All are afloat in the same
ship, and prosper or sink with it.
-
484 Chapter 13: The Role of the CFO, Performance Measurement,
and Incentive Design
2. Alignment of Pay with Performance The first half of this
chapter makes a strong case that the performance measure most
closely related to the total return to shareholders is the
difference between actual and expected (i.e., forecasted) results.
This is without doubt a subjective measure, and it becomes
necessary to deal with its subjectivity. Jensen [1998, p. 206]
argues that "performance in most jobs cannot be measured
objectively because joint production and unobservability mean that
individual output is not quantifiable." He goes on to criticize
objective merit systems because ( 1) any misspecification of the
performance mea~urement system can be easily gamed by employees,
(2) the measurement system becomes difficult to change because
inevitably any alteration is detrimental to some employees, and (3)
the threat of increasing objective standards higher in response to
better-than-expected performance this year-the so-called
expectations treadmill-causes a behavioral reaction where employees
deliberately restrict output so that they can earn their bonuses.
There are reasons against subjective systems as well. Subjective
systems are unpopular because employees may not trust superiors to
evaluate their performance accurately. Given the negative
attributes of objective systems and the lack of trust that defeats
subjective systems, Jensen concludes: 'The compensation system that
results from this set of forces appears to be one with little or no
pay for performance."
There are, however, some basic principles that help to link pay
with performance. The first is to keep the relationship between
performance and compensation as linear a~ possible. Figure 13.6
illustrates a "kinked" relationship between pay and performance
that is typical. Although the relationship is linear around the
average or expected level of performance, there is a lower bound
represented by basic salary, and an upper bound that represents
salary plus maximum bonus. To understand how the kinks in the
curve, combined with the one-period nature of compensation, result
in perverse behavior, suppose that a manager finds herself at point
A on the curve near the end of the year. It makes sense to take
large risks because she cannot do worse on the downside, and if the
risk pays off she can earn much more on the upside. If we study her
behavior at point B instead, and if it is early in the year, then
from her point of view she is incented to be very conservative
because she can lose a lot on the downside and gain little on the
upside because she is already at the maximum bonus point. Clearly,
a linear pay schedule is much better because she gets a constant
percentage of the value created through her actions. By the way, we
note that if performance is measured relative to expectations, it
is just as easy to construct a linear pay schedule.
Figure 13.6 "Kinked" relationship between pay and
performance.
Pay Linear incentives
B
I Kinked incentive structure J i // \ ~~"'"' to
be,ttraconsm,ti,c _______ A/ { I . ______
""_Pr_o_pe_n-si-ty_t_o_ta-ke_l_ar_g_e __ "~l1tornlanice :1 _ risk
at thmd of the Y'" "
-
C. Incentive Design 485
3. The Use of Stock Options It is often argued that if enough of
an executive's wealth is tied to the performance of the company,
then there is no agency problem between the owner and the managers
of the firm. The manager will want to join the owner in seeking to
maximize the value of the stock. Some would say that the manager
should invest a substantial portion of his wealth in the stock or
in call options on it to align the manager's interests with their
own. An executive stock option plan is a form of long-term
compensation contract that depends on corporate performance. It
usually gives managers the right to purchase a specified number of
shares for a specified period of time (called the maturity date of
the option) for a specified price (called the exercise price). The
first stock option plan to receive favorable tax treatment was
called the Restricted Stock Option plan of the Revenue Act of 1950.
There have been many changes since then, but since 1994 the tax
laws govern two types of plan: incentive stock options (ISOs) and
nonqualified stock options (NQSOs).
Tax consequences are as follows: At the time of the grant there
are no tax consequences for either the executive or the company. At
exercise, the executive who was granted NQSOs pays the ordinary
income tax on the spread (Sx - X) between the stock price and the
exercise price; the company reduces X and can deduct S - X. When
the stock is sold, the executive pays a capital gains tax on Sr - S
x, and there is no consequence for the company. If the executive
was granted an ISO, taxation is deferred until the stock is sold,
when the capital gains rate is applied to Sr S0, the gain from the
time of grant, and there is no tax consequence for the company.
The accounting treatment is also important. Both NQSOs and ISOs
must be exercised sequen-tially. There is a definite drawback to
the sequential exercise requirement. Suppose the firm's share price
was $50, fell back to $30, then rose again to $40. If options were
issued at each stage with the exercise price equal to the stock
price, executives would be required to exercise the earlier $50
options (that are $10 out-of-the-money) before they could exercise
the options that were issued at $30 (that are $10
in-the-money).
Both ISOs and NQSOs have a maximum life of l O ye:m from the
date of issue. The exercise price ofISOs must be greater than or
equal to the stock price at the time of issue, but NQSOs can have
an exercise price as low as 50 percent of the stock price.
Two reasons for using stock options at all are ( 1 )that they
augment salaries with a call option that makes executives' total
compensation more closely tied to shareholder wealth creation and
(2) that stock options are a more tax-efficient way of delivering
after-tax dollars to executives. Nevertheless, options pay off on
the upside and are simply left unexercised on the downside;
consequentially they are part of a compensation schedule that is
"kinked" on the downside. However, it can be argued that the
practice of resetting the exercise price after the stock p;:ice
declines has the effect of straightening the kinks, for example,
cash payments.
Miller and Scholes [1982] demonstrate that, when compared with
salaries, options are tax neutral from the firm's point of view and
are tax dominant from the manager's point of view. To illustrate
their argument, suppose that a firm expects $140 million of cash
flow in a good state of nature before paying management salaries,
and $120 million in a bad state. The finn' s tax rate is 50%, and
management salaries amount to $100 million. Table 13.5 shows the
firm's net payoffs after management salaries and taxes.
If the probability of the good state is 50% and if we assume a
zero discount rate, then the present value of the firm in this
one-period example is $15 million. Assume IO million shares
outstanding so that the price per share is $1.50. An at-the-money
call option will sell for $0.25, the average of its good and bad
state payouts, as shown in Table 13.6.
-
486 Chapter13: The Role of the CFO, Performance Measurement, and
Incentive Design
Table 13,5 Expected Cash Flows of the Finn (millions of dollars)
Good State Bad State
Cash flow before salaries $140 Salaries -100
Cash flow before taxes 40 Taxes@ 50% -20
Net cash flow $20
Table 13.6 Call Option Payouts at Time 1
Share price Less _the exercise price
Payout
Good State
$2.00 -1.50
$0.50
Table 13.7 Subsidiary Balance Sheet (millions of dollars)
$120 -100
20 -10
$10
Bad State
$1.00 -1.50
$0.00
Beginning Good State Bad State
Assets $15 Debt $12.5 Assets $15 Debt $15 Assets $10 Debt Equity
2.5 Equity 0 Equity
Total $15 Total $15.0 Total $15 Total $15 Total $10 Total
$10 0
$10
Now, suppose that the company offers managers an option plan
that is tax neutral from the finn's point of view. What will its
terms be? The plan will need to have a present value of $2.5
million and will be constructed from a portfolio of 10 million
at-the-money options granted to management. In return, management
would have to agree to reduce its salary by $2.5 million-an equal
dollar amount.
To handle the potential liability created by the plan, we assume
that the finn creates a subsidiary that will have a balance sheet
with $2.5 million in equity (the salary expense reduction) and will
borrow $12.5 million via a zero-coupon bond with a face value of
$15 million. The $15 million in cash is used to purchase 10 million
shares of stock. Table 13.7 shows the current balance sheet of the
subsidiary as well as its balance sheet in the good and bad states
of nature. If the good state occurs, the options will be exercised,
whereupon the subsidiary receives $15 million in cash and delivers
stock worth $20 million. The $15 million in cash is used to pay off
the loan and the equity in the subsidiary is worthless, leaving the
parent firm with a $2.5 million tax-deductible loss. If the bad
state occurs, the options will not be exercised, and the stock held
by the subsidiary will be worth $10 million and turned over to the
bank. The equity in the subsidiary will be worthless, once again
leaving the parent with a tax-deductible loss. The bank lends $12.5
million, and its expected payoff is $12.5 million. The parent ends
up with a $2.5 million loss either way.
-
C. Incentive Design 487
Table 13.8 The Finn's Cash Flow, Given the SAR Plan (millions of
dollars) Good State Bad State
Cash flow before salaries $140.0 $120.0 Salaries --97.5
-97.5
Cash flow after salaries 42.5 22.5 Investment loss -2.5 -2.5
Cash flow before taxes 40.0 20.0 Taxes -20.0 -10.0
Net cash flow $20.0 $l0.0
Table 13.8 shows the firm's expected cash flows given the terms
of the option plan and its payouts from the subsidiary. Comparing
the net cash flows of Tables 13.8 and 13.5, we see that the firm is
completely indifferent between the two alternatives.1
From the management perspective, after-tax salary could be
invested in stock options in order to have the same pattern of
future risky payouts. The main advantage of the stock option plan
is that taxes are deferred. Given that the plan is tax neutral from
the firm s perspective and rax preferred by management, there is a
strong incentive to use stock option plans.
Management, whose compensation is based on straight salary,
cannot benefit from undertaking risky positive net present value
projects unless their salaries are adjusted ex post to reflect good
decisions. Stock option plans can help to correct this
underinvestment problem because the options, and the stock, are
immediately more valuable when risky positive net present value
projects are initiated. There are several possible drawbacks,
however. Management cannot easily diversify the greater
firm-specific risk imposed by stock option plans and may require a
higher level of expected compensation. From the shareholders' point
of view, the cost of higher expected compensation may offset the
benefit of reducing the underinvestment problem. Furthermore, stock
options are not protected against dividend payments. This may
change the behavior of management to have a bias against the
payment of dividends and in favor of share repurchases. This may
conflict with the wishes of some groups of shareholders and support
the desires of others. Stock option plans may also bias management
toward decisions that increase the variability of the stock price
in an effort to increase the value of their stock options.
Studies of the announcement effect of the inception of stock
option plans for management indicate that shareholders react
favorably, considering the plan to be a net benefit Larker [1983]
finds a significant positive residual on the day following receipt
of the first shareholder proxy statement. Brickley, Bhagat, and
Lease [1985] find a significant positive 2.4% cumulative return
between the board of directors meeting and the Security and
Exchange Commission stamp date for the proxy statement. Lemgruber
[1986] used monthly return data for a sample of 119 firms with no
other infonnation in their proxy statements except for the election
of board members. For the interval between the board meeting and
the release of the proxy statement, he found a significant 2.7%
abnormal return.
I Note that we have not argued that the finn will use the tax
shelter that has been created if it can write off the difference
between the stock price and the exercise price in the good state of
nature.
-
488 Chapter 13: The Role of the CFO, Performance Measurement,
and Incentive Design
That the market reacts favorably to the inception of executive
stock option plans is consistent with the benefits of the plan
exceeding its costs from the shareholders' point of view. Call this
the incentive hypothesis. It is also consistent with a signaling
hypothesis. If managers have superior information concerning the
future prospects of the firm, they would desire the implementation
of a stock option plan when they believe the firm will do well. The
market would respond favorably to their action. Positive
announcement effects are also consistent with a tax hypothesis,
namely, that the after-tax payoffs of a salary plus the stock
option plan dominate those for a salary and bonus plan.
Consequently, the value of the firm will rise following the
inception of a stock option plan because total costs fall.
lt is not easy to separate the three aforementioned hypotheses.
They are all consistent with the observation of positive abnormal
returns at the inception of a stock option plan. They do, however,
make different predictions regarding management behavior. For
example, the incentive hypothesis predicts greater investment and
higher leverage once the plan is started. The signaling and tax
hypotheses predict greater earnings. All three hypotheses predict
lower dividend payout. The empirical evidence seems to lend little
support to the incentive hypothesis. For example, although
Lemgruber [ 1986] found positive announcement returns, he found no
significant changes in the rate of investment, in financial
leverage, or in the variance of the stock price. Lambert and
Larcker [19851 found that the variance decreased (a result that is
inconsistent with Lemgruher). Lemgruber did, however, find
significant decreases in dividend payout after the beginning of
stock option plans. Tehrani an and Waegelein [ 1985] find that
abnormal returns after the adoption of short-term compensation
plans are associated with positive unexpected earnings. In sum, the
empirical work suggests that stock option plans are adopted more
for tax or signaling reasons than to reduce agency costs between
owners and managers.
Although options are a tax-efficient way of delivering
compensation to executives, and help to align the interests of
owners and managers. a problem arises. namely. that a manager who
has accumulated a sizeable amount of wealth in the form of options
has an incentive to manipulate the market based on his inside
information. If he can convince the market that he has received
good news, the stock price will rise, thereby giving him the
opportunity to exercise his options at a higher price. To mitigate
this problem it is advisable to make the options exercisable only
at certain points of time during the year. This separates the
manager's ability to exercise the options from the freshness of
inside information that he might have. Even better are "clawback"
provisions that empower the board of directors to take back all
gains from the exercise of stock options.
Very large stock option awards also become material for
accounting reporting of earnings. FASB now requires that the
present value of option plans be estimated and reported in a
footnote. and that a fully diluted earnings estimate also be
reported.
4. Total CEO Compensation Total CEO compensation has many
separate components: salary, stock grants, stock options, pension
and health benefits. perquisites, and the duration of the CEO's
time in office. Additionally, there are contingent payment schemes
such as evergreen contracts. golden parachutes, and golden
handcuffs. Let's review the whole package.
Graef Crystal [1988] studied the top 100 of Fortune's 500
companies and found that CEO pay (including salary, bonus, the
value of perquisites when available, and 20% of realized gains from
long-term incentives, e.g., stock options) varies in line with a
number of seemingly rational factors. See Table 13.9. The only
problem is that shareholder returns is not one of them. The
variables in his index of company performance (profits, return on
equity, the market-to-book value of equity,
-
C. Incentive Design 489
Table 13.9 Poor Alignment between CEO Pay and Shareholder
Returns
Factor Definition/Comment Effect of a 10%
Increase on CEO Pay
Company size Index combining sales. assets. book equity. and
number of employees
+2.QC/c
Company perforn1ance Index based on profits, 5-year average ROE.
market-to-book value of equity, and 5-year average total return to
investors
+31.0%
Company risk Government regulation Tenure of CEO Location
Beta Regulated companies pay less Longer tenure results in
relatively lower pay More pay in high cost-of-living areas (e.g.,+
7% for New York and I 0% for Los Angeles)
+5.0% NIA
-1.2% NIA
CEO age No effect on pay Shares owned by CEO No effect on
pay
0.0% 0.0%
Note: CEO compensation includes salary. bonus, value of
perquisites when available. and 20% of realized gains from
Jong-term incentives (e.g., stock options). Source: Graef Crystal,
Fortune. June 1988.
and the historic return to investors) are virtually uncorrelated
with shareholder returns. Of course, he did not test to see whether
changes in expectations regarding these variables had an impact
mainly because he had no data on expectations.
Research by Jensen and Murphy [ 1990] studies the relationship
between CEO bonus and salary compensation and various explanatory
factors including the change in shareholder wealth. Table 13.10
shows the results. First, there is a statistically significant
relationship between CEO compensation and firm-specific changes in
shareholder wealth, even after accounting for the company
performance relative to the industry (row 3) and relative to the
economy (row 4). Second, the change in CEO compensation is
incredibly small relative to what happens to the firm. For example,
a $1,000 change in accounting profits results in a I 7. 7 cent
increase in CEO pay (row 5, column 3). Overall, the results in
Table 13.10 indicate that CEO compensatio11 is linked (but not very
strongly) to changes in shareholder wealth, changes in accounting
income, and changes in sales, but not to performance relative to
the market or industry. Holding the effects of accounting profit
and sales constant, a $1 ,000 increa,e in shareholder's wealth
results in a. .74 cent increase in CEO compensation.
The relationship between the level of CEO compensation and the
size of the company is much stronger, as shown in Table 13 .11. The
r -squared is between 50 and 70%. The average elasticity is 30%.
Thus, a CEO of a $200 million sales revenue company earns about 30%
more than the CEO of a $100 million company.
Jensen and Murphy [1990] also estimate the total effects of all
estimatable origins on CEO wealth. Note this is not salary and
wage, but an estimate of the wealth of the CEO. Note that it
includes the wealth effect of not being dismissed due to poor
performance. Also, the relationship between pay and performance
seems to be much larger for small firms than large. The largest
single effect is changes in value of the company's stock with
respect to stock ownership by the CEO, who had considerable skin in
the game.
-
490 Chapter 13: The Role of the CFO, Performance Measurement,
and Incentive Design
Table 13.10 Sensitivity of CEO Pay to Changes in Various Stock
and Accounting-Related Measures of Performance( 1990 by The
University of Chicago Press. Reprinted with permission.)
Regression Coefficients b Independent Variable a (1) (2) (3) (4)
(5) Intercept 31.5 31.9 32.5 31.0 32.8 t,.(shareholder wealth)
.0000140 .0000126 .0000074 .0000120 .0000074
(7.5) (4.8) (4.3) (7.1) (4.4) M wealth net-of-industry) c
-.0000012
(-.7) M wealth net-of-market) c .0000013
(.4) t,.(accounting profits) .000177 .000187
(17.2) (15.7) t,.(sales) .0000122 -.0000034
(7.2) (-1.7) R2
.0083 .0082 .0449 .0148 .0453 Sample size 7,747 7,747 7.721
7,721 7,721
Note: The sample is constructed from longitudinal data reported
in Forbes on 1,668 CEOs serving in 1,049 firms, 1974-1986;
I-statistics are in parentheses. a. The variables are all measured
in thousands of 1986 dollars. b. The dependent variable is
'.(salary+ bonus), measured in thousands of 1986 constant dollars.
The qualitative results are unchanged when '.(total pay) is used as
the dependent variable. c. '.(wealth net-of-industry) is defined as
(r, - i,)V,_ 1, where r, is shareholder return, V,_ 1 is
beginning-of-period market value. and i1 is the value-weighted
return for all other firms in the same two-digit industry.
Similarly. '.(wealth net-of-market) is defined as (r, -m;)V,_ 1,
where m, is the value-weighted return for all NYSE stocks. Source:
Jensen and Murphy (1990).
Table 13.11 Estimated Elasticity of CEO Pay with Respect to
Sales( 1990 by The University of Chicago Press. Reprinted with
permission.)
Year
1973 1975 1979 1981 1983 R2
Manufacturing .313 .296 .297 .287 .285 .60 Retail trade .253
.271 .230 .306 .298 .53 Gas and electric utilities .331 .236 .347
.313 .314 .67 Commerical banking .337 .329 .367 .372 .404- .68
Insurance .313 .277 .299 .372 .345 .69
5. Set Stretch Targets In 1986 the board of directors of Ralston
Purina announced that 491,000 shares of stock would be awarded to
Ralston's top 14 executives (160,000 to the CEO) if the stock
closed at or above $100 per share for 10 consecutive days any time
within the next 10 years. The current stock price was $63. An
increase from $63 to $100 represents a wealth gain of $811 million
for shareholders, and the stock awarded to management would
represent 6.05% of the gain in shareholder wealth. Is this
-
Figure 13-7 Cumula-tive abnormal perfor-mance of Ralston Purina.
(Reprinted from Camp-bell and Wasley [ l 999]. with permission from
Elsevier.)
1501 125 IOO
75 50 25
C. Incentive Design 491
- Ralston return market adjusted --Ralston return minus peer
group
O+IJ-,A!L_-~~~.-f..'.='lo&~~'..\-.1--l-.:!....\:
-25
an appropriate incentive for management? At first it seems so
because both parties (management and shareholders) would gain-a
win-win proposition.
Campbell and Wasley [ 1999] use the CAPM to estimate that had
Ralston Purina mere! y earned its cost of equity, it would have
taken four years and ten months to achieve the target share price.
It actually took four years and five months. During the expected
time to the award, Ralston's stock slightly underperformed the
S&P 500 and greatly underperformed comparables. Figure 13. 7
shows the cumulative abnormal performance starting in June of 198
I, five years prior to the announcement of September 1986, and
continuing beyond the time of the award in February 1991 up to June
of 1992, which was the expected time of the award.
During the four years and five months that it took to earn the
award, Ralston announced share repurchases seven times amounting to
$1 .541 billion. Dividends plus share repurchases averaged 140% of
net income between 1987 and 1990. As before, and once again, the
moral of the story is that performance should be set relative to
expectations. In this case management received a handsome reward
for earning less than the cost of equity, for performance that fell
below expectations. Stretch targets need to be set relative to
expectations.
6. Incentive Design across Multiple Time Periods The multiperiod
aspects of decison making are what make incentive design so
difficult. Managers change jobs frequently, thereby creating the
problem of giving proper attribution during the current
compensation period for the effect of their decisions made years
earlier in a different business unit Managers vary in the number of
years before retirement, and this creates an end-game problem.
Incentives based on the company's stock price may be less important
for executives near the end of their career than those who are just
starting. Also, one-year performance measures should be "guided" by
a long-term view to avoid behavior that maximizes short-term
performance at the expense of the long-run.
There is not much good theory to provide advice about a solution
for this multi period problem. We offer two thoughts-creation of a
"compensation bank" system, and control of share ownership based on
expected time to retirement.
The compensation bank concept takes the awarded bonus received
by an executive this year and puts it into a corporate trust ( or
makes it a general obligation of the firm), where it vests to
the
-
492 Chapter 13: The Role of the CFO, Performance Measurement,
and Incentive Design
Table 13.12 An Example of a Compensation Bank Year1 Year2
Year3
Bonus award $500 $100 $300 Cash paid out a $100 $ 20 $ 60
Deposit to compensation bank $400 $ 80 $240 Amount in bank
(end-of-year) $400 $520 $622 Less withdrawal b $ 0 $173 $207 Total
cash compensation $100 $193 $267 Bank balance forward c $400 $347
$415 a. 20% bonus b. 33.3% of balance c. Earns 10% interest
executive over a period of time. Vesting may be granted by a
review board or be automatic over time. Table 13. I 2 provides a
short example.
7. Designing Incentives for Different Layers of Management The
incentive structure that is appropriate for top management is quite
different than that of successively lower layers of management. Top
management can directly impact the share price of the company.
Business unit leadership cannot, nor can sub-business unit,
project, or plant management. Everyone, of course, will be happy to
accept a stock option plan if the grants are supplementary to their
existing level of compensation, but when faced with the more
difficult choice of whether they want to give up an amount of
existing income in order to receive an equivalent value of stock
options, few employees decide to switch. The most important reason
is that further down in the organization, managers have little or
no direct influence on the company's stock price. They would.
rather have their bonus compensation tied to something that they
control and where they have a reasonable chance of exceeding
expectations. A second reason is that they cannot separate the
value of their human capital from the value of the company;
therefore stock options do not allow them to diversify, whereas a
cash bonus for superior performance on their value drivers can be
reinvested elsewhere, thereby allowing diversification.
Middle management usually receives a salary plus bonus. The
problem with most bonus systems is that they do not require forced
ranking, and consequently everyone is placed in the top two or
three categories. A study by Medoff and Abraham [ I 980] surveyed
two large companies and found that the first, with 4, 788
employees, placed 94.5% in the top two of four ratings. The second
company rated 2,841 managers into six categories and 98.8% fell
into the highest three ratings. Further analysis indicated that
earnings differentials between jobs were much more significant than
earnings differentials (i.e., bonus differences) within job
categories. In other words, promotion is much more important as a
differentiator than are bonuses. Murphy [1985] found that for a
sample of vice presidents of large companies, promotion implied an
average 18.8% pay increase while the average pay increase given no
promotion was only 3.3%. There are two implications. First, bonus
systems can be more effective signals if rank ordering is required.
But since the more meaningful
-
C. Incentive Design 493
economic signal is promotion, it is wise to place a great deal
of effort in doing the evaluation well.
Lower-level management incentive pay, called gain sharing,
actually results in productivity gains. Mitchell, Lewin, and Lawler
[ 1990] study the relationship between the hourly wages earned (the
dependent variable in a multiple regression) and two independent
variables, the proportion of workers with incentives and the
existence of an incentive plan. Based on 716 companies they
obtained a cross-sectional r-squared of 68%, and both independent
variables were highly significant. The concept of linking the
performance of employees to value drivers under their control
actually works quite well.
8. The Issue of Relative Performance-Who Is Responsible?
Arguably, one drawback of incentive plans tied to changes in the
stock price is that not all stock price movements are attributable
to management decisions. Why should management be responsible for a
change in the market rate of interest as driven by the Federal
Reserve Bank? One point of view would argue that senior management
should, in fact be held responsible for the total movement in stock
prices. It is possible to hedge. For example, a gold mine can be
thought of as a portfolio of risky assets that is long in gold and
other precious metals. An offsetting position can be created by
shorting futures on gold. The net effect would be to reduce the
mining company's exposure to fluctuations in the price of gold-and
were the hedge perfect, the equity of the gold mining company could
theoretically be turned into a low-risk or even risk-free bond. In
a sense, if management decides to be in the gold business, it is
making a conscious decision to do so on behalf of shareholders.
There is no risk that cannot be managed in the long run-if not by
hedging or diversifying, then by simply deciding to get out of the
business. Therefore, there is a strong philosophical point of view
that favors total management responsibility for any movement in the
stock price, whatever the cause. The maxim is simple. Maximize
shpreholder wealth.
An alternative point of view argues that management of a gold
mine should have a comparative advantage in operating gold mines
and in finding and developing new sources of gold. Manage-ment
should not be expected to also specialize in hedging. Therefore, if
the stock price of the gold mining company changes because interest
rates rise or because the economy goes into a recession, management
should be indemnified. One of the implications, of course, is that
if man-agement is compensated on its relative (not on its absolute)
performance, then it may receive high compensation in a depressed
economy or low compensation in a buoyant economy.
To some extent it is possible to measure the performance of a
company after removing the effect of exogenous factors that are
judged to be beyond the control of management. We show, in Chapter
14, the details of how to build a spreadsheet discounted cash flow
model of a company. It is based on assumptions made by analysts
concerning value drivers such as the nominal rate of growth in
revenues, operating margins, capital turns, and the cost of
capital. This type of model typically produces a valuation that is
85% ( or higher) correlated with the actual market price. Given
this degree of accuracy, the DCF model can then be manipulated to
show the separate effects of exogenous variables on the stock from
management performance relative to expectations. While still more
an art than a science, nevertheless, this process is one way of
infonning the judgment of the decision makers who arbitrate the
level of total compensation for relative perfonnance. In this way,
if so desired, it is possible to remove the effect of a change in
the price of gold from the actual change in the stock price of a
gold mine.
-
494 Chapter13: The Role of the CFO, Performance Measurement, and
Incentive Design
We started this chapter with the role of the chief financial
officer-a role that has increased in complexity with the emergence
of multibusiness, multinational companies. Years ago the CFO was an
accountant, but today he or she is one of the top decision makers
in the company and often is elevated to the role of CEO. The CFO's
responsibilities are broad, including managerial decisions such as
performance measurement, incentive design, public relations, and
the management of human resources. Financial decisions include risk
management, capital structure and the cost of capital, dividend
policy, tax policy, and working capital management. The CFO is also
usually responsible for major investment decisions including
research and development spending, large internal projects, and
mergers and acquisitions.
The CFO is confronted with the problem of choosing a measure of
performance that is highly correlated with the change in
shareholders' wealth. We saw that for one-year time intervals the
best measure was the difference between actual and expected
economic profit, the keystone for expectations-based management.
Closely related is a multiperiod measure, namely, discounted
cash.flow valuation. In the next chapter we shall prove that the
sum of discounted economic profits, when added to the book value of
assets, equals the discounted cash flow value of the company.
Given that expectations drive the share price of companies, we
then discussed the myriad of thorny problems concerning the design
of incentive programs-what the agency problems are, how they are
usually handled, and the realization that there is no perfect
solution.
PROBLEM SET 13.1 Prove the following theorem: "The present value
of forecasted economic profit. when dis-counted at the weighted
average cost of capital, plus the book value of assets, equals the
DCF value of the firm." 13.2 Suppose that a firm has a project that
was started last year, and it is expected to earn less than its
cost of capital if left unchanged. Management comes up with these
suggestions .
. (a) Invest in a debottlenecking project that will raise
economic profit, but not up to the cost of capital. (b) Cut
operating costs but not enough to earn the cost of capital. (c)
Sell the unprofitable business unit for a premium over its book
value.
Some numbers are given in Table Q13.2. Value each alternative
and compare them in terms of value creation. 13.3 Some companies
measure performance by requiring both "top line" growth and "bottom
line" growth (i.e., growth in revenues and growth in net income).
Is this perfon11ance measure consistent with shareholder value
creation? Why or why not? 13.4 Why are EPS, the percent change in
EPS, EVA, and the change in EVA unrelated to TRS (total return to
shareholders) in Table 13.2? 13.5 Management has informed the
market that it expects to earn 30% on project A and 40% on project
B, and these expectations are already baked into the firm's current
stock price. The cost of capital for both projects is 10%. The
amount of investment required for project A is $5 million, and
project B requires $30 million. Just before it is about to invest
in project B, the company learns that the expected rate on the
project will be 30%, not 40%.
-
References 495
Table Q13.2 Data for Q 13.2 Base Case (a) Debottleneck (b) Cut
Costs (c) Sell for 10% Premium
WACC I 0
-
496 Chapter 13: The Role of the CFO, Performance Measurement,
and Incentive Design
Miller, M., and M. Scholes, "Dividends and Taxes: Some Empirical
Evidence," Journal of Political Economy, December 1982,
1118-1141.
Mitchell, D., F. Lewin, and D. Lawler, "Alternative Pay Systems,
Firm Performance and Productivity;' in A. Blinder, ed., Paying for
Productivity, 1990, Brookings Institution.
Murphy, K., "Corporate Performance and Managerial Remuneration,"
Journal of Accounting and Economics, April 1985, 11-42.
Smith, C., and R. Watts, "Incentive and Tax Effects of Executive
Compensation Plans," Australian Journal of Management, 1982, Vol.
7, No. 2, 139-157.
Tehranian, H., andJ. Waegelein, "Market Reaction to Short-Term
Executive Compensation Plan Adoption," Journal of Accounting and
Economics, April 1985, 131-144.