Vous êtes sur la page 1sur 24

A contingent-claim integration of

cost-volume-profit analysis with


capital budgeting*

JOSEPH K. CHEUNG Simon Fraser University and George Mason University


JOHN HEANEY Simon Fraser University

Abstract. This paper integrates cost-volume-profit (CVP) analysis into the theory of cap-
ital budgeting by modeling the profit fimction in the CVP relation as put and call options
on sales revenue. Such an approach is shown to be particularly useful when the profit
function is piecewise linear, such as when there are multiple break-even points, or when
the profit fimction is truncated. The results are of general applicability because such an
approach does not require making assumptions about the decision maker's risk attitude.
This claim is proven by showing that certain well-known decision models in the extant
literature can be derived as specific cases from the results. The paper also shows how the
CVP analysis can be extended to the case in which there is a dependence of the firm's
cash flows on macroeconomic variables. Specifically, it applies to the CVP analysis the
state-contingent claim approach to capital budgeting of Banz-Miller (1978) and relates
the option valuation approach to Banz and Miller's framework. A numerical example
using the state prices of Banz-Miller is provided.

Resume. Les auteurs associent l'analyse cout-volume-profit (CVP) a la theorie de


l'etablissement du budget des investissements en integrant la fonction de profit dans
r interaction CVP a titre d'option de vente et d'achat sur le produit des ventes. Cette
methode semble particulierement utile lorsque la fonction de profit est lineaire mais mor-
celee. II en ainsi, par exemple, lorsqu'il existe plusieurs seuils de rentabilite ou lorsque la
fonction de profit est tronquee. Les resultats peuvent etres generalises puisqu'une methode
de ce genre n'exige pas la formulation d'hypotheses relatives a I'attitude du decideur a
regard du risque. Cette affirmation est etayee par les auteurs qui montrent que certains
modeles decisiotmels bien connus dans la documentation existante peuvent etre derives
des resultats obtenus sous forme de cas precis. Les auteurs montrent egalement comment
l'analyse cout-volume-profit peut etre etendue au cas oii les flux monetaires de Tentreprise
dependent de variables macroeconomiques. De fa^on precise, ils appliquent a l'analyse
cout-volume-profit la methode des creances conditiormelles a un etat proposee par Banz
et Miller pour etablir le budget des investissements, et ils relient la methode d'evaluation
de Foption a la formule de Banz et Miller. Les auteurs proposent un exemple numerique
dans lequel ils recourent aux prix de Fetat de Banz-Miller.

* We thank Pao Cheng, members of the Simon Fraser University finance seminar, and the
referees of this journal for their comments and suggestions. All remaining errors are ours.
This research was completed by J. Cheung at Simon Fraser University. Financial support for J.
Cheung was provided by a research grant from the Social Sciences and Humanities Research
Council of Canada.

Contemporary Accounting Research Vol. 6 No. 2 - II pp 738-760


A Contingent-Claim Integration of Cost-Volume-Profit Analysis 739

Introduction
Cost-volume-profit (CVP) analysis can be viewed as a specific type of break-
even analysis. In a typical break-even analysis, the net present values (NPV)
of a project, conditional on alternate outcomes, are compared. For example, in.
the classic study by Reinhardt (1973), the NPVs for Lockheed's TriStar project
are compared for different numbers of TriStars sold. However, although such
information is valuable to management, it does not provide a solution to the
capital budgeting problem. In the TriStar example, for instance, the number
of aircraft that would be sold was uncertain at the time tbe investment was
undertaken, and such information would be revealed over a time period that
was uncertain. Any adequate solution to the capital budgeting problem would
require the calculation of a single NPV that appropriately accounts for both
uncertainties.' The purpose of this paper is to integrate traditional CVP analysis
into the theory of capital budgeting using contingent-claim analysis.
The idea of integrating CVP analysis with capital budgeting has been ad-
vocated by others including Manes (1966). However, the issue has failed to
receive much attention. Writers in accounting t5^ically ignore NPV in break-
even analysis, although finance texts address break-even analysis in NPV terms.
We attribute this lack of attention to two reasons. One is that the accounting
literature usually deals with complicated models of the CVP relation. Exam-
ples are the multiple break-even points and the newsboy problem, in which the
profit function is truncated above. In these models, the associated profit graph
is at best piecewise linear, which complicates a present value determination via
traditional techniques. Finance texts, on the other hand, simply deal with linear
profit functions.
The second reason seems to be that the traditional CVP analysis requires a
specification of the decision-maker's risk-return tradeoff, such as risk-aversion
in .Adar, Baroea, and Lev (1977), and risk-neutrality in Shih (1979). Imposing
a preference structure on the decision maker makes the firm's optimal decision
dependent on the manager's preferences. In contrast, the logic underlying the
NPV rule is such that the optimal decision is independent of the decision maker's
preferences and depends on the capital market's risk attitude instead. Magee
(1975) has investigated the CVP relationship in the context of mean-variance
capital market equilibrium, and the validity of his results are conditional on
investors' having quadratic utility.
This paper shows that valuing the profit function in a CVP framework as a
contingent claim can overcome these difficulties. In regard to the first problem,
piecewise linear cash flows lend themselves naturally to valuation as a portfolio
of put options, call options, and T-bills.^ In regard to the second problem,
the standard paradigm for valuing options, such as that by Black and Scholes
(1973) and by Merton (1973) in continuous time, assume only that no arbitrage

1 Reinhardt (1973) has only one source of uncertainty because the production rate is assumed
fixed.
2 See Cox and Rubinstein (1985, pp. 371-375), for example.
740 J.K. Cheung J. Heaney

opportunities exist in financial maikets. In discrete time, such as Rubinstein


(1976) and Brennan (1979), an additional assumption about the representative
investor's utility function is required. In either case, the NPV determined via the
contingent-claim approach does not depend on the decision maker's subjective
attitude toward risk. Therefore, the present approach not only integrates capital
budgeting concepts into the CVP relation but also produces results of general
applicability.
However, the capital budgeting approach to CVP analysis raises an addi-
tional consideration that is generally ignored in the traditional approach to CVP
decisions—an exception being Magee (1975). Namely, how does the depen-
dence of the firm's cash flows on macroeconomic variables affect the CVP
decision? It is weli known that in capital market equilibrium the valuation of
cash flows depends on their correlations with macroeconomic variables such
as aggregate wealth in the capital assets pricing model, aggregate consumption
in consumption-based capital asset pricing model, or economy-wide factors in
arbitrage pricing theory. If one assumes that CVP decisions are based on max-
imizing the manager's objective function, one generally requires knowledge of
the unconditional probability distribution function of future cash flows or, in
the case in which the manager is a mean-variance maximizer, just the uncon-
ditional means and variances of the future cash flow. However, if decisions are
based on maximiang NPV, solving the CVP problem would seem to require
some knowledge of the joint-distribution function between the cash flow and
macroeconomic variables.^ In some cases, such as when all investors are risk
neutral, the dependence of cash flow on macroeconomic variables may legiti-
mately be ignored. But in the general case, we may wish to model explicitly the
dependence of CVP decisions on macroeconomic variables.
The central purpose of this paper is to show how the traditional CVP analysis
can be absorbed into an option-pricing formulation of the capital budgeting
problem. Additionally, it also shows how the CVP analysis can be extended
to account for the effect of macroeconomic variables by applying the state-
contingent claim approach to capital budgeting of Banz and Miller (1978).^^ In
Banz and Miller, the relevant macroeconomic variable is assumed to be the retum
on the market portfolio, and the firm's cash flow projections are considered to
be conditional on the current state of the market such as depression, normal,
boom, and so on. This contrasts with the traditional CVP analysis in which the
cash flow projections are assumed to be independent of the current state of the
economy.^

In many cases, only partial knowledge of the joint distribution is required. For example, if the
representative investor has quadratic preferences, a knowledge of the expectation of per capita
wealth conditional on the firm's cash flow is sufficient
Gehr (1981) and Bierman and Smidt (1988, ch. 19) have also applied the state contingent-
claim approach to capital budgeting.
It appears that there is an additional distinction between the traditional CVP analysis and the
state-contingent claim approach to capital budgeting in that the former assumes a continuous
A Contingent-Claim Integration of Cost-Volume-Profit Analysis 741

Two cracial observations should be made at this point. First, the analysis is
not concerned with the agency problem, in the sense that management act on
their own behalf rather than in the best interests of the firm's financial claimants.
Instead, it is assumed that the agency problem has been solved, that the capital
markets are (sufficiently) complete, and that management and shareholders are in
equilibrium with the capital markets. These assumptions guarantee that wealth-
maximizing decisions can be made independently of management's preferences
and that these decisions would be unanimously approved by shareholders.^
Second, our concern with present values is prompted by the assumption that,
in the absence of the agency problem, management will act to maximize share-
holder wealth. We assume that a current decision of the firm concerning a project
will affect the project's future cash flows. The profile of such cash fiows is given
by traditional CVP analysis, but their present value is obtained by contingent-
claim analysis.
The paper will proceed as follows. The next section formulates the profit
function in the CVP relation as a contingent claim on snd-of-period sales reve-
ntte. It also illustrates how the basic result can be modified to accommodate
common issues such as unequal effective tax rates for profits and losses, semi-
fixed costs, and the newsboy problem. The third section shows via the newsboy
problem that the results are consistent with those reported by previous writers.
Specifically, it demonstrates that both Magee's (1975) and Shih's (1979) de-
cision rules in connection with the newsboy problem can be derived from the
formulation presented here. Also using the newsboy problem for illustration,
the fourth section shows how the Banz and Miller (1978) approach to capital
budgeting can be applied to the CVP analysis. The final section summarizes
the paper and offers some caveats concerning the application of the theory in
practice.

Cofitingent-claim formulation of the CVP relation


This section first shows how the profit function in the traditional CVP relation
can be formulated as a combination of put and call options on end-of-period
sales revenue. The results are then extended to accommodate (1) asymmetric
taxation of profits and losses, (2) semifixed costs, and (3) the newsboy problem.

distribution of the firm's cash flows while the latter assumes a discrete distribution. But this
distinction is somewhat superficial because state-contingent claim analysis is not restricted
to discrete distributions. The Black and Scholes (1973) option pricing formula is a classic
example of applying state-contingent claim analysis to a continuous price distribution. In fact,
the Banz and Miller approach itself is a discretized version of the Black and Scholes equation.
In a world of certainty, this result is referred to as the Fisher separation theorem. See, for
example, Copeland and Weston (1988, ch. 1) and Brealey and Myers (1984, ch. 2). The
result has been generalized to the uncertain world with a complete set of markets for state-
contiBgent claims by Debru (1959, ch. 7) and Hirshleifer (1970, ch. 9). Ekem and Wilson
(1974) and Radner (1974) have provided conditions for the result to hold under uncertainty
with incomplete markets.
742 J.K. Cheung I. Heaney

The analysis makes no specific assumptions about probability distribution of the


sales revenue variable. This discussion is deferred to the end of the section.

Basic model
A typical CVP analysis assumes that the number of items sold is known at the
end of a single period. The assumption that uncertainty is resolved at a given
instant in the future considerably simplifies the task of determining the NPV for
the project. Although such an assumption would be inappropriate for projects
whose uncertainty is resolved over substantial periods of time, it may well be
appropriate for short-run production, planning, pricing, and related decisions. It
is the purpose here to show that contingent pofits, coupled with the assumption
of a single time period, allow for the application of standard call and put option
theory to calculate the project's NPV.
The typical CVP relation can be stated as follows:
F (1)
where
T is the profit before taxes, Q is the number of units produced and sold, and
p, V, and F are, respectively, unit selling pdce, unit variable cost, and fixed
cost. Q and hence T are random variables whose values are revealed at the
end of the period, and p, v, and F are assumed to be known parameters at
the beginning of the period. We assume that Q and hence T are continuous or
piecewise continuous. By expressing sales in dollars (i.e., letting X = pQ), (1)
becomes

T = ^^^X~F (2)
P
Equation (2) gives the relation between end-of-period values of sales revenue X
and profit T. Let Xo be the present value of X and e~'' be the discounting factor
for the period, at the risk-free rate r. Then the present value of the project as of
the beginning of the period is
To = ^ ^ Xo - e-'F (3)
P
It should be emphasized that risk neutrality is not assumed in (3). F is
discounted at the risk-free rate because this value is assumed to be known at the
beginning of the period. X, on the other hand, is unknown as of the begirming
of the period so that its value at that time, Xo, depends on the systematic risk of
X (i.e., on X's covariance with market-wide factors at the end of the period).^
In addition, Xo may depend on the current state of the economy. We will retum
to this point later.
It frequently occurs that T in (1) is piecewise linear. When this is the case, T
can be more conveniently valued as a contingent claim on X. Figure 1 illustrates

7 Sales revenue X itself can be valued using a one-period valuation model, such as that of
Rubinstein (1976) or Lintner (1965). See Brealey and Myers (1984, ch. 9).
A Contingent-Claim Integration of Cost-Volume-Profit Analysis 743

this idea. In Figure I, T is given by line segment ABD, which is OBD minus
ABG. But the present values of OBD and ABG are, respectively, those of a long
call option and a short put option on X, with an exercise price OB. Therefore,
T in (2) can be valued as

P
where H = pF/(p - v) and C(H) and P{H) are, respectively, the current value
of a call and a put option on X, with an exercise price //.^ Note that (3) and
(4) together imply the well-known put-call parity result:

Xo+P{H) = C{H) + e-''H (5)


Admittedly, the put-call option formulation of the profit function as in (4) is
largely superfluous at this point, for (3) is more straightforward when the profit
function is linear. However, the options approach will have wider applicability
when the profit function becomes more complicated, as the following shows.
Income taxes
Income taxes traditionally have received little attention in the CVP analysis. This
is understandable if the tax is strictly proportional (i.e., a constant proportion
of profits and losses). But income taxation is asymmetric to profits and losses:
profits are taxed when earned but losses are not immediately subsidized when
incurred. Although the taxpayer is allowed to offset tax losses under the loss
carryover privileges and can transfer them via loss-selling transactions, the tax
asymmetry cannot be eliminated completely (Majd and Myers, 1987; Cheung,
1989). As a result, losses are subsidized at a lower effective rate than profits
are taxed. Although this asymmetric tax structure does not alter the break-even
point, it nonetheless affects the NPV determination. Thus it would be proper
for the CVP analysis to account for these asymmetric effective tax rates on the
profit function.
Allowing for this asymmetric tax effect would result in a piecewise linear
profit function. Consequently, the options approach can be applied more easily.
Let X be the effective tax rate for profits, and X* for losses, with x > x*. Then,
from (4), the after-tax profit f will have a present value of

% ^ IZl [(1 _ x)C(//) - (1 - x*)P(H)] (6)


P
where

H = pF/ip - V).

8 For the standard call option, payoffs increase one for one with the stock price if the option
is in the money at expiry. For the profit function T in (2), however, a $1 increase in X leads
to an increase of (p — v)//) dollars in T. This accounts for the factor {p — v)lp in (4).
744 J.K. Cheung J. Heaney

Figure 1 The basic relation

Profit $T

G Salesrevenue$X

-$F

F = fixed cost
V = the unit variable cost
p = unit selling price

Note that if there is no tax asymmetry (i.e., x = t*), then (6) reduces to % =
(1 — x)To, tax being a constant proportion of profits and losses.

Semifixed costs
Semifixed costs present a problem for CVP analysis in that they result in multi-
ple break-even points. In such cases, a project's NPV can be a useful basis for
A Contingent-Claim Integration of Cost-Volume-Profit Analysis 745

evaluation, particularly when it is necessary to select among competing alter-


natives. However, (3) will not be applicable under such circumstances, because
the profit function is no longer linear.
To illustrate the use of put-call options valuation in such a case, consider the
semifixed cost example given in Deakin and Maher (1987, p. 434), where fixed
cost is Fi if Q <Q and F2 if Q > Q, Q being the number of units produced
and sold. Making the same assumptions as those underlying (1), (2) becomes

^ . (7)
= ^_ll X-F2 if X>pQ
P
This is illustrated in Figure 2, where line segment ABD graphs T for X < pQ
and D'B'C graphs T for X>pQ.
Based on (7) and Figure 2, the current value of T is given by

To = PV (AC) -- PV (CDD'C') (8)


where
PV = a present value operator, and
AC and CDD'C' = cash flows depicted in Figure 2.
Applying (3) to value AC, we have

PV{AC) = ^ ^ ^ Xo - e-'Fi (9)


P
and applying the result from Appendix A to value CDD'C, we have
dC{K)
Py (CDD'C') = -(F2-F1) (10)

where C{K) is the current value of a call option on X, with an exercise price
K. A cash flow such as CDD'C is referred to as the payoff fi^om a "truncated
r-bill."^ lo sum, where there are two break-even points due to semifixed costs
Fi arid F2, the value of the profit function is given by

+ {Fz ~ (11)
p K=pQ

9 The term truncated T-bill is due to Pao Cheng. For example, if C{K) can be valued as a Black
and Scholes (1973) type call option on X, with an exercise price pQ. an expiry date one year
hence, and a volatility of X equal to o^, then

PViCDD'C) = -IF, ~F,) -^N


N {{ ^"(VPg) ^'" -
t a J
intuitively, if capital markets are risk neutral, (10) can be interpreted as
PV (CDD'C) = -(F2 ~Fi)- e-^'ProblX > pQ]
However, no such assumption is made in the model.
746 J.K. Cheung J. Heaney

Figure 2 The CVP relation with semifixed costs

Profit ST

p-v
P

p-v

Sales revenue $X

-$F,

-$F,
A'

Fixed cost is F^ if Q < Q, F^ if Q > Q,


P is the unit selling price and v is the unit variable cost. The two break-even points are B and B'.

Note that the only unknown in (11) is the exact form of the partial derivative. To
can be determined precisely if the exact form of C(K) is known. Two versions
of the exact form of C(K) are discussed at the end of this section.

The newsboy problem


When demand is stochastic and overproduced goods are perishable, the firm is
said to face the newsboy problem. This problem has been the subject of rigorous
A Contingent-Claim Integration of Cost-Volume-ftofit Analysis 747

analysis, Magee (1975) and Shih (1979) being the primary examples. In such a
case, the firm's production quantity becomes a decision variable, and the CVP
relation as stated in (1) no longer applies. The present analysis shows that the
options approach to the newsboy problem differs from previous models in that
it provides a decision rule for the problem in a general form.
To show this, let demand Z) be a random variable, production Q be a decision
variable, and uoit variable cost v, unit selling price p, and fixed cost F all be
kEOwn constants. Rewrite D - X/p and Q = Y /p, such that X and Y are,
respectively, sales and production in dollar terms. Because unsold items are
worthless, sales is equal to Min[X/p, Y/p], and the profit function T is given
by

F ifX>Y (12)

Figure 3 graphs this relation. It is evident from Figure 3 that the cashflowACM
can be valued as
To=Xo~e-'-(vY/p + F)-CiY) (13)
where CiY) is a call option on X with an exercise price F, and Y is to be
determined endogenously. To determine the optimal production in dollars, we
solve for F that satisfies dTo/dY = 0. This leads to the following decision rule:

I (14)
dY p
In a later section, we will show that (14) is a general result of which certain
models in extant writings are special cases. Specifically, we will show that both
the decision mle in Magee (1975) and Shih (1979) can be derived from (14).
TIMS far, it has been shown that the profit function T can be valued as a
package of call and put options and T-bills. The results are in general form
because no assumptions have been made about the probability distribution of
sales revenue X, the underlying asset for all contingent claims, or about the
market's risk attitudes. The profit function can be valued exactly when such
assumptions are made. Exact valuation formulas for the contingent claims can
be derived in two distinct ways. In one approach, it is assumed that the current
value of the profit function follows a continuous random walk through time.
Using standard continuous-trading arbitrage arguments due to Merton (1973)
and Black and Scholes (1973), the valuation formula emerges as the solution
to a partial differential equation. The alternate, discrete-time approach due to
Rubinstein (1976) and Brennan (1979) assumes that there is a representative
investor whose preferences determine market prices.
Two special cases are relevant here: either a normal or a lognormal distri-
butional assumption for sales revenue X, because these are the predominant
assumptions in the CVP literature (see, for example, Hilliard and Leitch, 1975,
748 J.K. Cheung J. Heaney

Figure 3 The newsboy problem

Profit $T

Sales revenue $X

F is the fixed cost, v is the unit variable cost, P is the unit selling price, and Y is the output in dollars.

1976; Lau and Lau, 1976). Consider the case where ln{X) is normally distributed.
Under certain circumstances,''^ the value of CiY) in (14) can be given by the
Black-Scholes option pricing formula. In this case, decision rule (14) takes the
form
V

p (14')

10 The capital market must be arbitrage free and sufficiently complete so that a portfolio can
be constructed from marketable securities whose value equals the value of X at any time
f, 0 < f < 1.
A Contingent-Claim Integration of Cost-Volume-ftofit Analysis 749

where N is the cumulative normal distribution and o is the rate of change in the
market value of the natural log of the cash flow X.

Relationship with Magee's and Shih's results


A previous section that claimed the present approach to CVP analysis leads to
a more general result than that in Magee (1975) and Shih (1979). This section
provides the proofs via the newsboy problem.

Consistency with Magee's result


Magee (1975) studied the CVP relationship in the framework of capital market
equilibrium. Specifically, rather than using the variance of profit as a measure
of risk, he used the covariance of profit with the retum on the market portfolio.
His approach provides a decision rule for the newsboy problem as shown in
(20) below. We wiU show that when Magee's assumptions are imposed on (14),
(20) belov/ will be the result. Hence, the two approaches are consistent with one
another.
Magee's approach is based on that of Ekem and Wilson (1974), who assume
that individual investors are mean-variance maximizers. There are basically two
justifications for assuming mean-variance decision makers: either investor pref-
erences are quadratic, or asset prices are joint normally distributed.'' Because
the end-of-period cash iow in the newsboy problem is truncated above, its
probability distribution is not normally distributed. Consequently, the Ekem and
Wilson approach will apply only if the representative investor's utility function
is qeadratic. This fact will first be used to derive the value of a call option on
the sales revenue in the newsboy problem, and then it will be shown that sub-
stituting the result so obtained and Magee's other assumptions into (14) leads
to Magee's decision rule.
First note that, for a representative investor, the value of any random end-
of-period cash flow Z is given by (see, for example, Rubinstein, 1976):

^ EiU'iW)]il+r)

where
Z - end-of-period cash flow being valued,
W = end-of-period per capita wealth,
U' = marginal utility of the representative investor, and,
r = one-period riskless rate of interest.
Specifically, when Z in (15) is given by the function max [0, X—K], then XQ
is the value of a call option on X, with an exercise price K. We seek to determine

11 For generalizations of these requiremeHts, see Chamberlain (1983).


750 J.K. Cheung J. Heaney

the value of such a call option when the representative investor's utility function
is quadratic. To do this, let U(W) have the following form:

U(W) = W- cW^/2 (16)


where c is a constant. Applying (16) in (15), substituting Z with max [0, X-K]
and using the definition of the expectation of a function of two random variables
in the numerator result in the following:
i-oo po
dX K)f(X, W)
C{K) = ^ ^^ (17)
ll-cE{W)](l+r) ^ ^
where/(X, W) is the joint density function for X and W.
We next use another of Magee's (1975, p. 262) assumptions:

(18)
where Eir^lX) is the conditional expected value of r^, and 1+ r^ = W/WQ, WQ
being the initial per capita wealth. Equation (18) represents a restriction on the
joint distribution between cash flow X and the rettim on the market portfolio
r^}^ This reduces (17) to (see Appendix B for details):

C(K) = : j i - I y (X- K)fxiX)dX - bF ^{X - K)[X - E(X)]fx{X)dx


(19)
where F = cWo/[l - cE{W)l Finally, applying the decision rule (14) to (19)
gives the following result:
V f°° r°°
- = / fx(X)dX-bP iX - E(X)YxiX)dX (20)
P JK JK

which is Magee's (1975) decision rule as stated in his equation 17 (p. 262).
Thus, Magee's result can be derived as a special case of (14).'-'

Consistency with Shih's result


In regard to the newsboy problem, Shih (1979, p. 692) has derived the following
decision rule:
V
fN{D)dD = - (21)
/ P
Jo
where
/iv = a nonnal density,
£> = the random demand,
Q = the production decision variable,

12 The parameter b in (18) is given hy b = Cov (r^, X)/Var (X) = fe-Var {rm)/Var(X).
13 Note that implementation of (20) requires knowledge of b, which requires knowledge of
A Contingent-Claim Integration of Cost-Volume-Profit Analysis 751

V — the unit variable cost, and


p = the unit selling price.
The assumptions underlying (21) are that demand is distributed normally, with
mean fia and variance o^, and that the decision maker is risk neutral. The
purpose is to show that imposing these assumptions on (14) will lead to (21).
For purposes of comparison, oote that, due to the symmetry of//>/ about jio,
(21) is equivalent to

fN{D)dD = - (22)
P
Now substitute Z = {D ~\ID)I<SD in (22). Then Shih's first-order condition (21)
can be rewritten as
V

P
where W{-} is the standard normal cumulative density. With the substitutions of
Q = YIP^ {!/) = E{X)IP and Op = Oo/P- (23) becomes

=- (24)

Finally it will be shown that (24) can be derived from (14) by assuming normality
for D and risk neutrality for investor preferences.
First, recall the definitions of two variables, D = X/p and Q — Y/p, where X
and Y are, respectively, sales and production in dollars. Assuming normality for
X, an exact form of the partial derivative in (14) can be obtained from Brennan
(1979)." This results in:

dY ^ " 1 00
Consequently, according to our decision rule (14), the optimal output in dollars
F can be found by solving

=- (26)

But in a risk-neutral market, e''X(j = E{X). Thus (26) is equivalent to (24).


Hence Shih's decision rule for the newsboy problem is also a special case of
our (14).

14 The exact form of a call option whose underlying asset is distributed normally with a variance
CJg, whose time to maturity is unity, and whose the exercise price is K, is

{e'Xfj— K\
Oo
_ (I
) + e Oofi i -
J I
where N{-} is the standard normal cumulative density and «{-} is the standard normal density.
See Brennan (1979) for details.
752 J.K. Cheung J. Heaney

In this case, because investors are risk neutral, no assumptions need be made
about the joint distribution of X with macroeconomic variables. However, given
that X is distributed normally, (25) will hold even if investors are risk averse.
Thus, (26) is more general than (24). To see this, note that e''Xo ^ E{X) when
investors are risk averse, except for the case where px = 0. But it seems that (26)
requires no assumption about the joint distribution of cash flow with aggregate
economic variables, an unlikely result if investors are risk averse. It turns out that
(26) indeed requires such an assumption, for the call-option fonnula of Brennan
(1979) from which (25) is obtained requires that cash flow and aggregate wealth
have a bivariate normal distribution when investors are risk averse.'^ Thus, given
the joint normal distribution of X with aggregate wealth, investors' risk attitudes
determine Xo. This is the way in which the market's risk attitude enters (26).
Suppose that a change in the current value of aggregate wealth is expected
to affect future cash flows. In the environment of (26), this means that in the
most general case a change in the current value of aggregate wealth affects the
joint distribution of X and wealth, thus changing Jo, cr, and possibly r, and
hence affecting the decision variable Y. Thus, the task of modeling the effects
of change in aggregate wealth on the decision variable F is quite complex in
the case of (26). The next section develops an approach that deals more directly
with the effects of changes in aggregate wealth on the CVP decision.

Applying the Banz-Miller approach to the newsboy problem


Under the state-contingent claim approach of Banz and Miller (1978), macro-
economic variables play a role in the firm's CVP analysis. This section shows
how their approach can be applied to CVP analysis. Via the newsboy problem as
an illustration, a numerical example will be provided using the state-contingent
prices given in Banz and Miller. Furthermore, it will also be shown that the
Banz-Miller approach to capital budgeting is consistent with the result stated in
(14). •
Consider the typical newsboy problem in which the firm has to decide on the
quantity of production at time t, for sale at time t+\, and where productions
unsold at time t+1 will be worthless. Under the state-contingent claim approach,
the quantity of goods that could be sold at any time ?, Qf, would depend on
the state of the world at that time. Suppose that the state of the world can be
represented by the values of a set of macroeconomic variables 6,. Then we can
write gf = Qf{Qi), « = 1, ..., N, where it is assumed that there are N possible
states of the world at any time t. Similarly, let g f be the number of units of
goods the firm decides to produce at time f, which it tries to sell at time f + 1.
Then the quantity produced also depends on the state of the world at time t.
That is, Q't = QfiQk), k - I, ..., N, where the world is in state k at time t.
In such a case, the number of units actually sold at time /-i-l, 2f4.i(%), would

15 Brennan (1978) assumes in addition that the representative investor has constant absolute risk
aversion (i.e., has an exponential utility function).
A Contingent-Claim Integration of Cost-Volume-Profit Analysis 753

depetid on both the state of nature at time t+l and the production decision at
time f. Hence, the following relationship holds between units actually sold at
time t+l and units produced at time t:

Qt^i(%) = ef^i(%) for states where ft^i(%) < gf (9^)


= ef(8,) for states where Qly{%j) > efC^)
Consequently, (27) leads to the following modified form of the profit function
(12) at time t + 1, given that Q^(Qk} units were produced at time t:

(28)
- -F(9,) - F if X(9;) > F(9,)
P
where X(8y) = p • Qt+iiQj) is the market value of sales revenue at time t+l, and
Y(Qic} = P • Qf(fik) is the market value of the Q^{6k) units produced in state k
at time t}^ Equation (28) should be compared with (12).
Following Banz and Miller (1978), we introduce one-period state prices Ve^^et
for the value of $1 to be delivered in state Bj at time t+l, given that state 8^
occurs at time t. Given these state prices, the value of firm's production decision
at time t is

The newsboy problem in this case is solved by choosing QfiQk) in (28) so as


to maximize PVt(Tt.n) in (29). The following is a numerical example of the
newsboy problem using the state-contingent prices given in Baoz and Miller.
As in Banz and Miller (1978), assume that there are three equally probable
states of the world: 6i = depression, 02 = normal, and 93 = boom.'^ Assume
that the firm's sales projections are given by the vector X = [195, 200, 205],
where $195 of sales revenue is expected in a depression state and so on, that
the firm's fixed costs are $30, and that its unit variable cost is 45 percent of
unit selling price (i.e., v/p = 0.45). Using these assumptions in (28), the firm
caB work out its profits at time r -i-1 for each state j = 1 , 2 , and 3, conditional
on each current state k = I, 2, and 3. This results in a 3 x 3 matrix where each
current element is Tt+i (0,, 0*). The firm then determines the values of these state-
contingent profits using one-period state-contingent prices in a manner specified
by (29). This leads to the present value of each production decision conditional

16 For simplicity, (27) assumes that both X and Y are stationary random processes that depend
only on the state that occurs, not on time.
1? Banz and Miller do not specify the joint distribution of the firm's cash flow with the state
variable (returns on the market portfolio). Instead, they discount the expected cash flow condi-
tional on each state. That is, T in (29) is expected T given 0,. See Breeden and Litzenberger
(1978) for justification for this.
754 J.K. Cheung J. Heaney

on each of the three states of nature. The following one-period three-state prices
are available from Banz and Miller (1978, p. 666):

"0.5251 0.2935 0.1735-


0.5398 0.2912 0.1672
-0.5544 0.2888 0.1612-
where, for example, $0.5398 is the current price of $1 to be received if a
depression state occurs at time t+l (i.e., J = 1), given that the world is presently
in a normal state (i.e., k = 2, etc.)
Using these state prices in the example leads to the following values for each
state-dependent production decision:
76.64 77.11 77.59
76.74 77.16 77.58
L75.38 75.75 76.13J
where the first column gives the values of production at the 195, 200, and 205
levels, respectively, if the world is currently in a depression state. The other two
columns are to be interpreted similarly. Based on this matrix, the firm would
choose the production level such that, given the current state of the world, the
value of its decision is maximized. These maximum values are 76.74 for column
1, 77.16 for column 2, and 77.59 for column 3. In the example, therefore, the
firm would set production equal to 200 if the current state is either depression
(column 1) or normal (column 2), and equal to 195 if the current state is boom
(column 3). This completes the illustration of how the newsboy problem can be
solved in the Banz-Miller context.
To see the connection between the put-call options approach as given by
(12)-(14) and the Banz and Miller approach as outlined in (28)-(29), substitute
Tt+i(dj, dt) from (28) into (29) and rearrange the results. We then have

(30)

all;-

where e^''* = XwLi "^Qjfit i^ the one-period riskless discount factor'^ for state k,
and j * is all the states at time t+l where X(Qj) > Y(Qk)- But the first term on
the RHS of (30) is the present value of X,+i, and the third term is the present
value of a call option on the cash flow Xt+i. Thus (30) can be written as

7(9,) F1 C(e,, 7(6)) (31)

18 Note that the riskless interest rate is now state dependent in contrast to the less realistic
assumption' of constant interest rate assumed earlier.
A Contingent-Claim Integration of Cost-Volume-Profit Analysis 755

Clearly, this result parallels that in (13). Hence, the put-call formulation of the
newsboy problem remains valid in the Banz-Miller framework. Note that in
(13) the state variable Qt is XQ itself and that the possible dependence of the
market values on the state variable is suppressed. It is perhaps worth remarking,
therefore, that using (31) as an objective function for the newsboy problem
transcends any distributional assumptions about By and X and any assumptions
about market risk attitudes.

Summary
CVP analysis has Ihe basic objective of highlighting the sensitivity of profits
with respect to changes in sales, cost stracture, or both. Much of recent research
shows that, by imposing a probability distribution on sales, the major source
of BBcertainty, much more can be said about profitability, such as the various
mathematical moments of profits, the probability of breaking even, etc. We
obsewe that although these approaches give added insights to the problem at
hand, they are short of readily leading to optimal decisions, particularly where
mutually exclusive projects are involved. This paper suggests that, in order to
help make decisions consistent with the maximization of shareholder wealth, the
profit function in the CVP relation should be valued in a way that is consistent
with the theory of capital budgeting.
This paper shows how the profit function in the CVP relation can be for-
mulated in a put-call option framework. It demonstrates why such an approach
is well suited for the CVP relation, given that the related profit function is
mostly piecewise linear, thus lending itself to be valued as modified puts, calls,
and T-bills. To the extent that the value of the underlying asset, sales revenue,
reflects the market's risk attitude, our approach is consistent with capital market
equilibrium. This is so because the options approach to valuation also depends
implicitly on market risk attitude. Magee (1975) has shown that, by assuming
that investors are mean-vaiiance maximizers, the CVP relation can be cast in a
market equilibrium framework. On the other hand, Shih (1979) studies the CVP
relation in a risk-neutral market context. This paper shows that both Magee's
and Shih's results can be derived as special cases of our formulation.
We next cast the traditional CVP analysis in the Banz and Miller (1978)
framework of capital budgeting, in which the firm's cash flow is viewed as
dependent on market-wide factors. Hence, we show how the traditional CVP
analysis can be extended to reflect directly the influence of macroeconomic
factors. To illustrate precisely how this works, a numerical example is provided
using the state prices presented in Banz and Miller. Finally, we show that our
option valuation point of view is retained in the Banz and Miller framework.
It should be emphasized that the arbitrage approach has only limited prac-
ticality at present due to the onerous assumptions underlying the solution. We
view the CVP problem essentially as a problem of valuing an option on the
firm's sales revenue. Unfortunately, the standard Black and Scholes method of
756 J.K. Cheung J. Heaney

valuing options is not directly applicable in this case because the underlying
asset here, the firm's sales revenue, is not traded. Thus we require additional as-
sumptions beyond those required for the ordinary option valuation. To maintain
the spirit of Black and Scholes in valuing the option in question by arbitrage, we
must assume not only that trading of securities is continuous but also that the
market is sufficiently complete so that the firm's sales revenue can be duplicated
by existing securities.
In sum, the arbitrage approach to the CVP analysis requires assuming that:
a Markets are frictionless.
b Securities are continuously traded.
c A constant, risk-free rate exists, and
d Security markets are sufficiently complete such that one could construct a
portfolio to replicate the firm's sales revenue.
Of these four assumptions, the first three are standard in the Black and Scholes
world. However, the problem of identifying the replicating portfolio under (d),
even assuming the (d) is satisfied, would appear to be a formidable problem.'^
The arbitrage approach to the CVP analysis necessarily requires some heroic
assumptions, but then so do the possible alternative approaches. As an alternative
to the arbitrage approach, one can construct models of general equilibrium. But,
inter alia, such models generally require specifying:
i the joint distribution of the firm's cash flow with market-wide factors, and
ii the representative investor's attitude toward risk.
For example, Magee (1975) assumes under (i) that the expected return of the
market portfolio conditional on the firm's cash flow is linear in the cash flow—
see (18)—and under (ii) that investors have quadratic utility functions—see
(16). Shih (1979), on the other hand, avoids making an assumption about (i),
but assumes that investors are risk neutral under (ii). Clearly, neither of these
sets of assumptions is likely to be realized in practice.
Use of the Banz and Miller (1978) approach provides a multiperiod frame-
work with stochastic interest rates. However, the additional realism is bought at
the expense of additional assumptions. In particular, Banz and Miller assume:
a The state space can be partitioned into a small number of equally probable
states.
b The Black and Scholes model can be applied sequentially to value a call
option on the market portfolio to determine state-contingent prices,
c Random returns on the market portfolio can be described by a stationary
transition-probability matrix,
d An asset's value can be determined by discounting its expected cash flow
conditional on each state of nature, which includes ranges of return on the
market portfolio.

19 Nevertheless, such an assumption plays a key role in financial theory (see, for example, Ross,
1978) and is beginning to emerge in the applied literature (for example, see Trigeorgis and
Mason, 1987).
A Contingent-Claim Integration of Cost-Volume-Profit Analysis 757

Without passing judgment on the realism of these assumptions, we note that


according to Breeden and Litzenberger (1978, p. 648), using the Black and
Scholes formula to value options on aggregate consumption, or aggregate wealth,
requires the existence of a representative investor whose utility function exhibits
constant relative risk aversion. Thus, the general equilibrium underlying the Banz
and Miller model appears to be just as selective as that of some other approaches.
Nevertheless, despite these problems with implementing the theory in prac-
tice, we believe that decisions based on CVT analysis, which affect shareholder
wealth, should be made in a capital budgeting framework. This paper has offered
a theoretical formulation that provides some insight into how this integration may
ultimately be achieved.

Appendix A: Value of a Truncated T-Bil!


Consider the valuation of a modified call option OABC as shown in Figure 4.
Clearly, this option is worth
PViOABC) = QiC(Ki) + (02 - e,)C(i^2) (A.I)
where C(/) is the price of a call option with exercise / . We refer to a cash
flow such as OK2BN as a truncated T-bill. To value such cash flow, apply
(A.I), and iet ^1 tend toward K2 and 62 tend toward zero. This results in
01 = (BK2/K2-K1), 82 = 0 and

PV(MK2BN}= Urn
K2 —

From (A.2), factoring out BK2 and applying the definition of a derivative lead
to the following result:

'^^^ (A.3)
dK
L

Intuitively, if capital markets are risk neutral, (A.3) can be interpreted as

PV(MK2BN) = BK2ie^' Prob {X > K2)] (A.4)


But risk neutrality need not be the case.

Appendix B: Derivation of equation (19)


Rewrite (17) as
/•oo /-oo

/ dX dW{X-K)fiW\X)fx{X)
JK i-00
~cj dxj dWW{X-K)f{W\X)fx{X)
758 J.K. Cheung J. Heaney

Figure 4 Valuation of a truncated T-bill

Value of
underlying asset x
MKfilSl is a truncated T-bill, valued as OABC with if, tending toward if^ and 9^ tending toward zero.

where/(W^ I J ) is the conditional distribution of W given X, fx(X) the marginal


distribution of X, and r the riskless rate. Performing the integration over W
yields the following:

/ dX(X-K)fx(.X) -c K)E{W\X)fx(X)
JK JK
[I - cE(W)]{l + r) (B.2)
dX{X -
JK

where E{W \X) is the conditional expected value of per capita wealth.
Next, we seek an expression for E{W\X) in (B.2). Given Magee's assumption
(p. 262) that
(B.3)
A Contingent-Claim Integration of Cost-Volume-Profit Analysis 759

where E{rm\X) is the conditional expected value of r^ and 1 4- r^ = W/Wo, Wo


being the initial per capita wealth, we have

E{W\X) = WQ + aWi) + bW^X (B.4)

and
E(W) = Wo + aWo + bWaEiX) (B.5)

Thus,
E{W\X) = E{W) + bW^X - EiX)] (B.6)

Substituting (B.6) into (B.2) then gives

dX{X-K)[l-cE{W)]fx{X)
K
/•oo
- I dX(X - K)cbWolX - E(X)]fxiX)
C(K) = ^—^ ^ (B.7)
^ ^ [l-c£(W)](l+r)
Finally, letting P = CWQ/[1 - cE(W)], collecting terms in [1 - cE{W)] in the
numerator and simplifying reduce (B.7) to
I / /.oo ^oo \
C(K) = I (X- K)fxiX)dX -bP I {X- K)[X - E{X)Yx{X)dX
l+r \JK JK J
(B.8)
which is (19) in the text.

References
Adar, Z., A. Bamea, and B. Lev, "A Comprehensive Cost-Volume-Profit Analysis
Under Certainty," The Accounting Review (January 1977) pp. 137-149.
Banz, R. and M. Miller, "Prices for State-Contingent Claims: Some Estimates and
Applications," Journal of Business (October 1978) pp. 653-672.
Bieniian, H., Jr. and S. Smidt, The Capital Budgeting Decision, 7th ed. (New York:
Macmillan, 1988).
Black, F. and M. Scholes, "The Pricing of Options and Corporate Liabilities," Journal
of Political Economy (May-June 1973) pp. 637- 659.
Brealey, R. and S. Myers, Principles of Corporate Finance, 2nd ed. (New York:
McGraw Hill, 1984).
Breeden, D., and R. Litzenberger, "Prices of State-Contingent Claims Implicit in
Option Prices," Journal of Business (October 1978) pp. 621-651.
Brennan, M.J., "The Pricing of Contingent Claims in Discrete Time Models," Journal
of Finance (March 1979) pp. 53-68.
Chamberlain, G., "A Characterization of the Distributions That Imply Mean Variance
Utility Functions," Journal of Economic Theory (1983) pp. 185-201.
Cheung, J., "On the Nature of Deferred Income Tax," Contemporary Accounting
Research (Spring 1989) pp. 625-641.
Copeiand, T. and J. F. Weston, Financial Theory and Corporate Policy, 3rd ed. (Read-
ing, MA: Addison-Wesley, 1988).
760 J.K. Cheung J. Heaney

Cox, J.C. and M. Rubinstein, Options Markets (Englewood Cliffs, NJ: Prentice-Hall,
1985).
Deakin, E. and M. Maher, Cost Accounting, 2nd ed. (Homewood, IL: Irwin, 1987).
Debru, G., The Theory of Value (New Haven: Yale University Press, 1959).
Ekem, S., and R. Wilson, "On the Theory of the Firm in an Economy with Incomplete
Markets," Bell Journal of Economics and Management Science (Spring 1974) pp.
171-80.
Gehr, A. Jr., "Risk-Adjusted Capital Budgeting Using Arbitrage," Financial Manage-
ment (Winter 1981) pp. 14-19.
Hirshleifer, J., Investment, Interest, and Capital (Englewood Cliffs, NJ: Prentice-Hall,
1970).
Hilliard, J. and R. Leitch, "Cost-Volume-Profit Analysis under Uncertainty: A Lognor-
mal Approach," The Accounting Review (January 1975) pp. 69-80.
, "CVP Analysis under Uncertainty: A Log Normal Approach—A Reply," The
Accounting Review (January 1976) pp. 168-171.
Lau, A. and H. Lau, "C-V-P Analysis Under Uncertainty—^A Log Normal Approach:
A Comment," The Accounting Review (January 1976) pp. 163- 167.
Lintner, J., "The Valuation of Risky Assets and the Selection of Risky Investments
in Stock Portfolios and Capital Budgets," Review of Economics and Statistics
(February 1965), pp. 13-37.
Magee, R., "Cost-Volume-Profit Analysis, Uncertainty and Capital Market Equilib-
rium," Journal of Accounting Research (Autumn 1975), pp. 257-266.
Majd, S. and S. Myers, "Tax Asymmetries and Corporate Tax Reform," in M. Feld-
stein (ed.). The Effects of Taxation on Capital Accumulation (Chicago: University of
Chicago Press, 1987) pp. 343-373.
Manes, R., "A New Dimension to Breakeven Analysis," Journal of Accounting Re-
search (Spring 1966) pp. 87-100.
Merton, R.C., "The Theory of Rational Option Pricing," Bell Journal of Economics
and Management Science (Spring 1973) pp. 141-183.
Radner, R., "A Note on Unanimity of Stockholders' Preferences among Alternate
Production Plans: A Reformulation of the Ekem-Wilson Model," Bell Journal of
Economics and Management Science (Spring 1974) pp. 181-184.
Reinhardt, U., "Break-Even Analysis for Lockheed's TriStar: An Application of Finan-
cial Theory," Journal of Finance (September 1973) pp. 821-838.
Ross, S., "A Simple Approach to the Valuation of Risky Streams," Journal of Business
(July 1978), pp. 453^75.
Rubinstein, M.E., "The Valuation of Uncertain Income Streams and the Pricing of
Options," Bell Journal of Economics (Autumn 1976) pp. 407-425.
Shih, W., "A General Decision Model for Cost Volume Profit Analysis under Uncer-
tainty," The Accounting Review (October 1979) pp. 687-706.
Trigeorgis, L. and S. Mason, "Valuing Managerial Flexibility," Midland Corporate
Finance Journal (Spring 1987), pp. 14—21.

Vous aimerez peut-être aussi