Blackwell Publishing American Finance Association
Blackwell Publishing American Finance Association
Blackwell Publishing American Finance Association
Market Value and Systematic Risk Author(s): Stuart M. Turnbull Reviewed work(s): Source: The Journal of Finance, Vol. 32, No. 4 (Sep., 1977), pp. 1125-1142 Published by: Blackwell Publishing for the American Finance Association Stable URL: http://www.jstor.org/stable/2326518 . Accessed: 03/05/2012 10:35
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].
Blackwell Publishing and American Finance Association are collaborating with JSTOR to digitize, preserve and extend access to The Journal of Finance.
http://www.jstor.org
THE JOURNAL
OF FINANCE
* SEPTEMBER
1977
ONE OF THE CENTRAL ISSUES in the theory of finance is the relationship between expected risk and expected return required by individuals investing in assets. The Capital Asset Pricing Model (CAPM)' provides such a theoretical relationship under conditions of market equilibrium. Explicitly, the model states that the expected one-period return for an asset is a linear function of its systematic risk, which is a measure of the responsiveness of the asset's return to changes in the return on the market as a whole. Proposed applications of the CAPM have ranged from strategies for security selection, measurement of investment performance, establishing a structure of managerial fees, to estimating the cost of capital. Such applications usually involve estimating the systematic risk of a firm's equity. Generally, it is assumed that systematic risk is constant over some arbitrary time period and can be estimated using time series data. In the time series testing of the validity of the CAPM to equity securities, though elaborate portfolio construction procedures are used, it is implicitly assumed that systematic risk of equity is constant.2 Yet, there is an increasing amount of empirical evidence by Blume [1971], White [1972], and others indicating that systematic risk of equity is non-stationary. Given the theoretical and practical importance of systematic risk, it is pertinent to enquire about its determinants. In a survey of recent empirical work, Myers [1975] tentatively suggests that systematic risk of equity appears to be related to four variables: financial leverage, earnings variability, growth, and an accounting or cyclical beta, which is a measure of the covariance between swings in the firm's earnings and swings in the general economy. However, most of the empirical studies suffer from the deficiency of lacking any theoretical framework, relying upon intuition for their structural development. To describe the determinants of systematic risk, it is necessary to consider the general question of describing the determinants of market value. Black [1969] drawing upon some unpublished work of Treynor, has argued that the value of an uncertain cash flow depends upon various economic variables, such as the level of * Dept. of Political Economy University of Toronto. An earlier version of this paper,was presented at the European Finance Association meetings, September 1975. The Author is grateful to the London Graduate School of Business Studies for research support. The helpful comments of Richard A. Brealey, Stewart C. Myers, Stephen A. Ross and the referee Robert H. Litzenberger are gratefully acknowledged. Responsibility for any errors rests with the author. 1. A good survey article is given in Jensen [1972]. 2. It should be noted that this implies that the systematic risk of the firm is a function of the firm's leverage ratio, even in perfect securities markets without corporate taxes, which contradicts the Modigliani and Miller [1958] capital structure irrelevance theorem.
1125
1126
GNP, plant capacity and inventory levels within the relevant industry. From this formulation, a general deterministic partial differential equation describing the present value of the uncertain cash flow in terms of its relation to the economic variables, is developed. The major limitation of this work is the assumption of an exogenous risk premium. Brennan [1973] derives an expression for this risk premium by assuming the validity of the CAPM. By positing that the terminal value of the cash flow is a linear function of a firm specific component and a general economic index, a solution to the general differential equation is obtained. However, if the economic variables affecting the value of the uncertain cash flow change stochastically over time, this might affect the investment opportunity set. As Fama [1970] has shown, changes in the investment opportunity set can affect the equilibrium structure of returns and the CAPM will no longer provide an adequate description of the return mechanism.3 Without restrictions on the preference structure of investors' utility functions, the generality of the derived results are jeopardized.4 Another potential problem with Brennan's work is that of limited liability; there is nothing inhibiting the market value of the firm from being negative. The objectives of this study are to derive an expression for the market value of a firm and to describe the mechanism, within a continuous time context, of how the factors that determine the firm's market value affect its systematic risk. The market value of a firm is determined by positing that the firm's cash flows depend upon a firm specific and a set of general economic variables. A general partial differential equation describing the market value of the firm is derived, explicit account being taken of the effects of a changing investment opportunity set upon the structure of returns. A general expression is obtained for the systematic risk of the firm. An explicit solution is derived for the systematic risk and market value of the firm. The present value of the firm depends upon the current expected values of the components of the cash flows, each component being discounted, in general, with a different risk premium. This offers a theoretical and practical solution to the capital budgeting problem of determining the value of uncertain cash flows, as these risk premia can be estimated for a wide class of applications. From the explicit solution, the determinants of systematic risk can be identified. It is shown that systematic risk depends upon the duration of the project, and is a non-increasing function of duration. By considering a simple growth model, it is possible to derive an expression for systematic risk in terms of the growth rate variable for the firm's cash flows. It is demonstrated that systematic risk does not increase as the growth rate increases, a result which is contrary to the usual view of how growth affects systematic risk.5 The dependence of the firm's systematic risk upon the duration and growth rate of its cash flows, implies that for capital
3. For a further discussion of the ramifications of a changing opportunity set upon the structure of returns, see Merton [1970]. 4. If the economic index is the market portfolio, then Brennan's results hold. While Brennan shows that the index is related to the market portfolio, it is important to note that at the very outset the index can only be the market portfolio, for the analysis to be generally valid. 5. See, for example, Lerner and Carleton 11965].
1127
budgeting decisions it is not, in general, possible to use the firm's capitalization rate as a cut-off point for accepting investment projects. The theoretical framework developed in this study to describe the determinants of systematic risk, can be used to explain some of the deficiencies of recent empirical studies. The model also has important implications for the empirical testing of the CAPM and implies that at least a two-variable model should be used. In Section II the framework of the model, the derivation of the partial difference equation, and a general expression for systematic risk, are presented. In Section III an explicit solution to the partial differential equation is given, and the nature and implications of the solution discussed. A general discussion of systematic risk and an explicit form describing its determinants, is given in Section IV. The dependence of systematic risk upon duration and growth is demonstrated in Section V, and the implications for capital budgeting decisions are discussed in Section VI. In Section VII the developed model is used to explain some of the deficiencies of recent empirical studies. The general implications of the model with regard to empirical testing of the Capital Asset Pricing Model are derived in Section VIII. Conclusions are presented in Section IX.
II.
VALUATION OF UNCERTAIN CASH FLOWS AND SYSTEMATICRISK
To explain the determinants of the systematic risk of a firm, a general approach is developed to the valuation of uncertain cash flows. By considering the specific nature of a firm's cash flows and the equilibrium expected rate of return required by individuals to invest in the shares of the company, a second order partial differential equation, describing the equilibrium market value of a firm, is derived. Capital Market Structure It is assumed that the capital market is structured as follows: Assumption 1: All assets have limited liability; Assumption 2: There are no transaction costs, taxes, or problems with indivisibilities of assets; Assumption 3: There are a sufficient number of investors with comparable wealth levels, so that each investor believes that it is possible to buy or sell as much of an asset at the market price; Assumption 4: The capital market is always in equilibrium; Assumption 5: There exists an exchange market for borrowing and lending at the same rate of interest; Assumption 6: Short-sales of all assets, with full use of the proceeds is allowed; Assumption 7: Trading in assets takes place continuously in time; Assumption 8: The term structure of interest rates is "flat" and known with certainty; Assumption 9: The dynamics for the value of the firm, V, through time can be described by a diffusion type of stochastic process with stochastic differential equation dV =adt+adZ, (1)
1128
The Journal of Finance where a is the instantaneous conditional expected percentage change in value per unit time; a2 iS the instantaneous conditional variance per unit time; and dZ is a standard Gauss-Weiner process such that E(dZ) = 0 and var(dZ) = dt. In general, a and a will change stochastically over time. It is assumed that the firm pays no dividend. Equation (1) implies that price changes are serially independent. If a and a were assumed to be constant, then (1) would imply that prices are stationarily and log-normally distributed.
Cash Flow Structure It is assumed that a firm undertakes a single project, which is realized at time T. The cash flow for the jth firm is denoted by CJT where the tilde signifies that at time t (t < T) the cash flow is a random variable. The equilibrium market value of all claims to this cash flow, at time t, is denoted by Vt(CjT). The cash flow of a firm can, in general, be decomposed into a firm specific component and a set of economic components. For example, the expected value of a cash flow for a manufacturer of durable goods will depend upon the difference between the expected unit selling price minus the expected unit cost of production. The expected selling price may depend upon such economic factors as net income and the availability of credit; if the economy is in a recession, or if there is an increase in the amount of credit rationing, then this may have an adverse effect upon demand and thus the value of the cash flow.6 It is assumed that investors form their expectations about future cash flows conditional on a set of economic indices for the economy as a whole and a firm specific component; that is,7 A.
m
E[CT|t]=E[:aT
(2)
where aTT is a random variable representing that part of the cash flow, at time T, that is intrinsic to the firm; 'kT is a random variable representing the kth economic variable at time T; pt is the information set containing the current values of the firm specific component, atT, and the set of economic variables { Ikt; E( ) is the expectation operator; and {bk} are firm specific constants, as are the power exponents, which are defined in the interval 0< y, < 1, i=0, 1, ...,m. It is assumed that the { Ik,t are independent,8 that is, cov(Ikt,Ijt)= 0, for all k and j, k # j; the firm specific component is independent of the set of economic variables, and independent across firms, cov(]atT,katT) =0, for all j and k, j # k. Equation (2) states that the expected value of the cash flow conditional upon the current
6. For example, suppose the cash flow for the jth firm can be written as CT= PT - CJTlwhere PT is a random variable representing the unit selling price; and -T a random variable representing the unit cost of production for the firm. If the equilibrium selling price is determined by a set of economic variables, then CJT=(bo -CT)+k= IbkIkTb The cash flow is decomposed {'kT' such that PT-bo+ E.IbkikT, into a firm specific component, (b0- CT). PlUSa set of economic components, l bklkT 7. The jth suffix has been dropped for purposes of clarity. 8. This is not a crucial assumption; it is a simple matter to relax this assumption and consider the general case of the economic variables being correlated.
1129
information set can be decomposed into a linear summation of different components: the conditional expected value of a firm specific component, E[[a| lo 10j and a function of the conditional expected values of the different economic components k =lbkE[I kTI j. Given assumptionsabout the dynamicsof Jz and Ik,} over time, then current values of a,T and {Ik,} can, in general, be related to their respective expected values at time T. B. Partial Differential Equation At any instant in time the market value of all claims to the future cash flow depends upon the set of economic variables {Ik,4 and the firm specific component a,t. Hence, the value of the uncertain cash flow can be written
Vt(CT)
=
t(Iz,T),
where I is a (k x 1) vector of the economic variables (Ik,}. It is assumed that the dynamics of the economic variables can be described by a diffusion-type Itb process with stochastic differential equation
dI
hk (Ik, )dt+ gk (Ikt )dZ'
k=
(4)
A (a,T)dt + B(a,T)dZa,
(5)
where hk is the instantaneous conditional expected change in value of the kth economic variable per unit time; g2 is the instantaneous conditional variance per
unit time, k =
,...,
m; A is the instantaneous
of the firm specific variable per unit time; B2 is the instantaneous conditional variance per unit time; and dZa, and dZ1k, k = l... m are standard Gauss-Wiener processes. It should be noted that by assumption dZa is independent of dZIk. Equations (4) and (5) imply that for a particular variable unanticipated changes are serially independent.9 Given (3), then by lto's lemma,'0 and using (4) and (5). gives
dy - (+?
hkIk,
+ Aa,Tda
I
a2 2
22a
12
+ 2
tT
gk
)dt
( BatT aa dZa
k1=
gk
k, a k dk
(6)
9. If the economic variables are future prices of commodities, then a sufficient condition for (4) to be an accurate description would be for commodity markets to be efficient-see Samuelson [1965]. 10. For a statement of Ito's lemma, see Kushner [1971].
1130
Equation (6) describes the stochastic changes in the market value of the uncertain cash flow caused by changes in the economic variables { Ikt and the firm specific component a,T. But the dynamics of the firm are described by (1) and thus by comparison with (6), implies that +I ay=-+EhkIk, and ay
m
av
1 4 +Aa,Taav + -B
2 2
a,T22
a2y
1 + 2
a 2y
(7)
aydZ=
BatT
4 dZ aa
gk'kt dlik
k-1 aIk(8
(8)
Equation (7) is a second order partial differential equation for the market value of the firm. The right hand side of the equation depends upon the parameters describing the dynamics of the different variables and how the value of the firm is related to changes in the underlying variables. The left-hand side of the equation contains the equilibrium expected rate of return. Equation (7) is similar to that derived by Black [1969]. Equation (8) shows how random changes in the firm specific component and the economic variables produce random changes in the market value of the firm. This is intuitively reasonable, for it would be expected that unanticipated changes in the operations of the firm and the economic variables that influence the future prospects of the firm, might affect its value. The systematic risk for the firm is defined by
8v=cov
M /var M
where M is the market portfolio of risky assets. Substituting for dv and using (8), gives
3v=
[a,ria
cov(BdZa,dj)
k4 E
dZ
M)
/var(
()
'M)/varM)
(10
and
l (=covi
d_kt I
dM
7vrdM'~ M/var
k=l.
m,
(11)
where /Pta can be interpreted as the responsiveness between changes in the firm specific component and the market portfolio, and /,tk as the responsiveness between changes in the economic variable Ik and thP market portfolio. Hence, the
Market Value and Systematic Risk systematic risk of the firm can be written in the form =
(
tT
1131
aa
8ta
k lI
aIk
/3tIk)'
(12)
which implies that systematic risk is composed of two parts that reflect its relationship with changes in the firm specific component and the economic variables. Given that over time changes in these variables occur, then such changes will, in general, affect the systematic risk of the firm implying that it will not be constant over time. Before the partial differential describing the market value of the firm can be solved, it is necessary to determine how stochastic changes in the firm specific component and the economics variables affect the equilibrium expected rate of return. It is assumed that investors allocate wealth between present and further consumption so as to maximize the expected utility of lifetime consumption and terminal wealth, that is
MaxEo Uf
where U is a strictly concave von Neumann-Morgenstern utility function for consumption; B is a strictly concave "bequest" or utility of terminal wealth function; and C(S) is the instantaneous consumption flow at time S. At each point in time, the investor's consumption-investment decision will depend upon current wealth, W, time t, and the economic variables, I, used to form expectations about future values. Unanticipated changes in the economic variables might affect the value of assets and thus the investment opportunity set. As it is assumed that firm specific characteristics are independent across firms, and independent of the economic indices, their effects upon the risk of a portfolio can be eliminated by diversification. Hence, the consumption-investment decision will not depend upon the firm specific characteristics of individual cash flows. Given the assumptions about the dynamics of the securities and the economic variable,11 as Merton [1970, 19731 has shown, the equilibrium instantaneous expected rate of return can be written in the form
m
ar=,J(a A-
r) +
E: k= I
( 13)
where r is the instantaneous risk-free of interest; aM is the instantaneous expected = rate of return on the market portfolio of risk assets; a- a,W/UMM, and denotes the risk of the security; a M is the instantaneous covariance of the jth systematic security with the market portfolio of risky assets; a,Ik is the instantaneous covariance of the jth security with the kth economic variable; and Ylk is a preference term12 reflecting investors' attempts to hedge against changes in the
11. See (1) and (4). - JkWIJw, 12. This is defined as yIk=-Hk/Q, k= . m, where Hk= /J w,' the prefix 'i' denoting the identity of the ith investor-see Merton [1973]. J4
and Q=j,-
1132
investment opportunity set caused by unanticipated changes in the kth economic The left-hand side of the above equation denotes the equilibrium instantaneous risk premium for the jth security. The first term on the right-hand side represents the risk premium generated by the security's covariance with the market, and is identical to that given by the continuous time analog of the Capital Asset Pricing Model.'4 The second set of terms reflects the investors' attempts to hedge against changes in the investment opportunity set caused by unanticipated changes in the economic variables. If the economic variables do not change stochastically over time, then the terms {Ylk} w-ould not be present and thus the continuous time analog to the CAPM would attain. Substituting (13) into (7) and using (12) to eliminate the systematic risk term
variable.'3 gives'5
a m
yr=
- + E1 h
I a 2y
al
+(Aa,:-Y2a,T/8a)
a2y
=YIk0MIk*
tial equation which describes the market value of a firm. It depends upon the level which pertain to their {hg}, of the economic variables (Ik,), as well as {h} and dynamics; the level of the firm specific component, ,T, and A and B, which describe its dynamics; { Ylk)} a set of terms representing investors' preferences for hedging against changes in the investment opportunity set; /8a and { /PI, which respectively reflect the responsiveness of the firm specific component and the economic variables to changes in the return on the market portfolio; and Y2,which is a type of general market risk premium. Any solution to (14) must satisfy the terminal boundary condition
m
y(I, T, aTT
) = (
TT )
bk (Ik T I
)(1l5
k=
which follows as a logical consequence of (2) describing the specification of how expectations are formed. Also, any solution must be non-negative, given the assumption of limited liability. This has direct implications for the specification of the equations (4) and (5) describing the dynamics of the firm specific component and the economic variables. For example, if these variables were normally distributed, then there would be a finite probability of the firm having a negative value. It is assumed, for simplicity, that the dynamics of the firm specific component and
13. While at the micro level of the individual little can be stated about the sign of Yk' at the macro level it might be possible to fix a sign to YIk depending upon the economic variable-see Merton [1973]. 14. See Merton [1970].
gk2(ay/aIk)dt,
15. To obtain (16) the following identity is used: a.,kdt=cov(dy/y,dIk)=cov(adZ,dik)=(l/y) using (4) and (8).
1133
the economic variables are described by log-normal distributions; that is, in equations (4) and (5), A, B, {hk} and { gk) are constants. This implies that the distributions of the firm specific and the economic variables will be stationary and that f3a and { /3,k) will be constant over time.
III.
EXPLICIT SOLUTION
Given the complexity of (14) and the nature of the boundary conditions, it does not appear to be possible to obtain an explicit solution to the partial differential equation. However, if the terms {Y,1 which reflect investors' preferences for hedging against changes in the investment orportunity set, are zero then an explicit solution can be derived. The work of Fama and MacBeth [1974] would suggest that assuming the prefern,nceterms to be zero might be a good empirical approximation. As shown by Merton [1973], the preference terms will be identically zero if investors have logarithmic utility functions. Assuming that the preference terms are zero,'6 a solution to equation (14) which satisfies the concave boundary conditions (15) is given by
t,aIT)
=
y(I,,
{ a1exp[A,(Tm
+ E bk{Iktexp[h,k(Tk5 1
t)}Yk
xexpf
(T-t)[-y
Yk +
g2 lyk(l
Yk)]
exp[-r(T-t)].
(16)
where y= aM - r, the instantaneous expected excess rate of return on the market portfolio. The current expected values of the firm specific component and the economic variables, are given by E [aTyoI= {a,Texp(T- t)[A + BI(yo- 1)])Yo 12 (yk- 1)])1, k= l.,m, and EtIk70jf, using (4) and (5). Ik, exp(T- t)[h Thus the market value of the firm depends upon the current expected values of the components of the cash flow, which are discounted with different risk premia. For the case in which the terminal cash flow is a linear function of the firm specific component and the economic variables (y, = 1, i =0, ,..., k), then (16) can be written in the form
y (I, t, atT)=(
bkE(Ik Tk,)exp[
(T t)/3I(M-r)]
exp[-r(T-t)].
(17)
16. The importance of this assumption lies only in the fact that it facilitates an explicit solution. In general, if the terms { Ylk) are not zero, a numerical solution to (14) can be derived.
1134
If the firm specific component and the economic variables do not change stochastically over time, implying that /la' BI, { PI3k), and { glk} are zero, then (17) degenerates to the intuitive result that the terminal cash flows, which are known with certainty,'7 are discounted at the risk-free rate of interest. The terms inside the {) bracket of (17) represent the certainty equivalent value of the uncertain cash flows and this is discounted at the risk-free rate of interest. Equation (17) also gives the market value of the firm for the case when the economic variables {Ik } are statistically correlated.18 The insensitivity of the solution to the correlation arises from the linear nature of the boundary conditions. The risk premia used to discount the different components of the cash flow are typically of the structural form /3(a - r), suggesting the interpretation of an excess return. The term (aM- r) is the instantaneous expected excess return on the market portfolio of risky assets, and the beta coefficient measures the responsiveness of a specific variable with the market portfolio. While in general it might be difficult to estimate the firm specific /3a, if there are a large number of firms such that the actions of a particular firm have negligible effect upon the market portfolio, then it would be expected that /8a would be approximately zero; that is, the firm specific component is uncorrelated with the market. The economic variables { /Ik} are not firm specific and, at least in theory, they can be estimated. This implies that given the estimates of the different risk premia, the market value of the firm can be determined. Equation (20) provides a theoretical, and for a wide class of applications, a practical solution to the capital budgeting problem of determining the market value of uncertain cash flows. Using (17), it is possible to examine some of the deficiencies of the traditional procedure of determining the present value of a firm by discounting its uncertain cash flows components using a single capitalization rate. Use of a single capitalization rate implicitly assumes that all cash flows are equally risky and that the risk arises from one common source. If these conditions are not met, then the relatively riskless components of the cash flow will be undervalued and the relatively risky components overvalued, which, in general, will result in non-optimal investment decisions. The market value of the firm given by (17) is derived assuming that the firm has a single project, which is realized at time T. For a firm with n projects, which are realized at times TJ, 1,...,n j= its market value is given by
n
V=
E(aTT
m
I ,)exp
-TJ -t)
[r
+ #BaJ(t-)
+E
k bJkE(IkT|
I)exp{ -(TJ-t)[r+fII(aM-r)]}],
(18)
1135
For the general case in which the terminal cash flow is a concave function of the firm specific and the economic variables, as described by (15), the general solution is given by (16). All of the risk premia are now affected by the exponent form of the component cash flows and are of the form /3a(0M-r)yo and /38k(aM - r)Yk, k= l,...,m. IV. RISK SYSTEMATIc
The systematic risk of the firm can be expressed in terms of elasticities by using equation (12); that is,
t'=ea 8a
+
Id
k = I
elIk
Ik'
(1 9)
where ea= (a,T/y)(ay/ aa), the relative sensitivity of the market value of the firm to changes in the firm specific component; and elk =(Ikt1y)(v/vaIk), the relative sensitivity of the market value of the firm to changes in the Ikt economic variable, k = 1, ... ,m. Thus, the systematic risk of the firm is a weighted average of the beta coefficient of the firm specific component and the beta coefficients of the economic variables. Over time, changes in the value of the firm and its sensitivity to the underlying variables will, in general, affect the values of the elasticities and hence systematic risk would not be expected to be stationary. If an expression for the market value of a firm is available, which explains the structural dependence of the firm specific component and the economic variables, then an explicit form for the systematic risk can be derived. For the case in which the terminal cash flow is a linear function of the firm specific component and the economic variables, then using (17) the systematic risk of the firm is given by
flvAt,
A( {
aE(aTT
t)expH(t-t)/3a(aM-r)]
E 8,kbk E(IkT1,) k= I
exp
(T
t),13k((M-r)]
Bexp[-r(t-t)].
(20)
The above expression demonstrates the importance of the beta coefficients for the firm specific component and the economic variables, as well as the complex structure of the elasticities. The effects of leverage upon the systematic risk of equity can be demonstrated by using a result of Merton [1974], who shows that the systematic risk of equity can be related to the systematic risk of the firm: /E = ( Vft/fJ)I3v where /3E is the systematic risk of the firm's equity; /At is the systematic risk of the firm: V is the market value of the firm; f is the market value of equity; and fv = (af/ aV). Given the assumption of the irrelevance of capital structure to the value of the firm, then the systematic risk of the firm, A,, will be independent of capital structure. Hence, the systematic risk of equity can be expressed in the form
k = I
1136
The terms within the first set of brackets reflect the effects of leverage, and the terms in the second set of brackets are due to the systematic risk of the firm. These terms are not independent. As the value of the firm changes over time, this will, in general, affect the systematic risk of the firm and the leverage factor. Both will affect the systematic risk of equity. Equation (21) demonstrates that a priori systematic risk of equity would not be expected to be constant over time, a conclusion which is consistent with the empirical findings of Blume [1971] and White [1972].
V. MATURITY AND GROWTH
It is possible to examine how duration and growth affect the market value of the firm and its systematic risk. The greater the growth of the firm then, intuitively, the greater will be its systematic risk.20 Similarly, it is argued, again on an intuitive basis, that the longer the duration of a project, the greater will be the risk. The effect of maturity upon the systematic risk of the firm can easily be determined by differentiating the expression for systematic risk, (20), with respect to the terminal life of the project, T. For the case of only one economic variable (that is, m = 1), then at8v
8I
-|8a
/-,8a 8I
)(atm-
r) + (A - h 1
aT
where9, = [A
if A =h,
-
r- /3a(aM-
r-31j (aM-r)](T-
t). Clearly,
then2'
aT A
that is, the systematic risk of the firm does not increase as the duration of the project increases,22 a conclusion which is counter intuitive. The systematic risk of the firm is a weighted average of the beta coefficients of the individual components of the firm's cash flow. The weight of the larger beta coefficient decreases as duration increases, while the weight of the smaller beta coefficient increases, and consequently the systematic risk of the firm does not increase. The firm's systematic risk will be constant as duration is changed if /3a=1, which implies that the certainty equivalent value of the firm decreases exponentially over time. This result is equivalent to that derived by Robichek and Myers [1967, p. 84], who showed that a constant capitalization rate implies that the certainty equivalent value decreases at a constant rate over time. Hence, in general,
20. For example, this type of argument is used by Beaver, Kettler and Scholes [19701. 21. This assumption can be relaxed. The result still holds if
('8 _J-A )2(aM -r) + ( .81-wl8 )(A - h ) > O.
22. This conclusion is in agreement with a result derived by Myers [19751 using a discrete time framework.
1137
while the dispersion of the cash flow increases over tiime, the firm's systematic risk either remains constant or decreases. To examine the effects of growth23 upon the systematic risk of the firm, it is assumed that the cash flow of the firm depends upon a firm specific component alT and an economic variable I, and each component of the firm's expected cash flow grows at a constant rate g, so that a,T= afexp(gT) and bT= bfexp(gT). Hence, the market value of the firm, using (17), is given by MV(t)= afexp(gt)f
/8a(aM-
r)-g]dT
(amM-r)-g]dT
+ bfexp( gt)I,
af exp( gt)
r+/,
r + 8/(atM-r)-g
where MV(t) is the market value of the firm; and it is assumed, for expositional simplicity, that A1 O, and h IO, that is, there are random disturbances about the mean values of the cash flow components. The beta coefficient for the firm is simply the market value weighted average of the betas of all the firm's expected cash flows, and is given by
_ _ _
F
L
~~af
(a
bf
1,
MV(t)
r +/
a-r)-g /3 +
/1B
exp(gt).
(23)
While the systematic risk of the firm is a function of the growth variable, g, it is not immediately clear how growth affects it, as both the numerator and denominator decrease when the rate of growth increases. Differentiating %,f with respect to g gives
__F
=_ (aM
r)afbfI,( 8a
bfI,3a
- /3_)2
ag
[(aM-r)(af/31-+
(2
<
(25)
that is, systematic risk is a non-positive function of the rate of growth.24This is an interesting result because it is exactly contrary to the usual view of how growth affects systematic risk, and can be explained in a similar manner as for duration. Intuitively, it is argued that the more rapid the growth of the firm, then the greater is its capitalization risk.25The above result shows that as the growth rate in the
23. Growth is defined to be expansion of the firm's cash flows, as opposed to growth in future investment opportunities. 24. A similar conclusion has been reached independently by Myers [1975]. 25. See Durand [19571, and Lerner and Carleton [1965].
1138
firm's earnings increases, then systematic risk does not increase and may even decrease.
VI. IMPLICATIONS FOR CAPITAL BUDGETING
The dependence of a firm's systematic risk upon the maturity and growth rate of its cash flows has important implications for the capital budgeting decision.26 Using the CAPM, it is argued27 that if the expected rate of return of a project is greater than the capitalization rate for a firm's stock, then the project should be accepted, assuming that the project has the same systematic risk as the firm. The firm's equilibrium capitalization rate is estimated by measuring its systematic risk. It has been shown that in general a firm's systematic risk depends upon the maturity and growth rate of its cash flows, which implies that if the firm's capitalization rate is used as a hurdle rate for accepting investment projects, then these projects must have in general the same maturity and growth rate of the firm, if they are to have the same systematic risk. Furthermore, as the systematic risk of the firm is a weightedaverageof the 8a and { 8I ), then the systematicrisk of the projects will differ from that of the firm, unless the cash flows of the projects are strictly proportional to those of the firm. Consequently, in general, the use of the firm's capitalization rate as a hurdle rate for accepting or rejecting investment projects can lead to non-optimal decisions. VII.
RELATION TO EMPIRICAL STUDIES
In a survey of recent empirical work Myers [1975] tentatively identifies four variables as being important real determinants of systematic risk: leverage, an accounting beta, earnings variability, and growth. As this study provides a theoretical framework describing the determinants of systematic risk, it can be used to examine the empirical results. The empirical significance of leverage upon the systematic risk of equity is expected, given the theoretical arguments advanced by Merton [1974]. The effect of leverage upon the systematic risk of equity is reflected by the leverage factor described in (21).28 The systematic risk of the firm-see (19)- is expressed in terms of the variables fBa and { /3k4, which represent the responsiveness between the firm specific and the economic variables with the market portfolio. The nature of these variables is similar to the accounting beta empirically identified as a determinant of systematic risk. The accounting beta, which measures the covariance between swings in the firm's earnings and swings in the earnings of the general economy, can be justified if it is assumed that it acts as a form of proxy variable for the /Ba and { /3k), though the adequacy of such a variable is questionable. The use of a single variable to
26. A full exposition of these and other implications of using the CAPM for capital budgeting decisions is given in Myers and Turnbull [1977]. 27. See, for example, Rubinstein [1973]. 28. It should be noted that this factor is measured using market values; Black, Kettler, and Scholes [1970] used book values, though there was still a large, significant effect.
1139
represent a number of different factors /Ba and { 83k), does imply a possible misspecification in the empirical models. From the theoretical model, only the systematic component of earnings variability should be relevant-see (20). Empirically, even if only the systematic components are relevant, the significance of earnings variability might still occur, as variance is composed of systematic and unsystematic components. Unfortunately, no empirical studies have separated the variance of a firm's earnings into these
components.29
The empirical evidence supporting a positive relationship between growth and systematic risk is quite weak; for example, Beaver, Kettler, and Scholes [1970] found that though systematic risk and growth were significantly correlated during the period 1947-56, they were uncorrelated during the period 1957-65. From (23) it is seen that systematic risk is a function of growth, but it is not a positive function; that is, as the growth rate of the firm's earnings increases, its systematic risk does not increase and may even decrease. The growth measures used in the empirical studies would appear to be unreliable proxies for some other variable. It should also be noted that systematic risk is a non-linear function of growth, while empirical studies have assumed a linear function. This misspecification implies that, in general, the statistical estimates are unreliable, being biased and inefficient.
VIII. IMPLICATIONS FOR THE EMPIRICAL TESTING OF THE CAPITAL ASSET PRICING MODEL
Any empirical test of the Capital Asset Pricing Model (CAPM) is usually a joint test of the CAPM and the market model.30 For the case of when the investment opportunity set is changing stochastically over time, the CAPM still provides a description of the structure of returns if investors have logarithmic utility functions.31 However, application of the market model for empirical testing will imply a misspecification of the model, as the systematic risk of the firm will, in general, be changing stochastically over time. Using the continuous time analogue to the CAPM and (12), the structure of returns is given by =
aa\M
1 )|+
yIk
(am-r)
(26)
The variables on the right hand side of (26) are typically of the form of an elasticity multiplied by the expected excess return on the market. Both of these variables may change over time, and thus the market model will not, in general, provide an adequate description of the return generating mechanism. This can be made more apparent by writing the above equation in the form
m
a -r
=9at/0a
E
k= I
Okt8Ik'
(27)
29. See Myers [1975] for a more detailed discussion. 30. Originally proposed by Markowitz [1959] and extended by Sharpe [1963] and Fama [1973]. 31. For proof, see Merton [1973].
1140
where 6a and {Sk } are defined by comparison with (26). These terms can be interpreted as the market risk premium for the particular factor; over time there is no a priori reason why these terms should be constant. By assuming a multi-factor generating model and using an arbitrage argument, Ross [1973] has shown that the expected return must be of the same structural form to that of (27), irrespective of the explicit nature of investors' utility functions; that is, it is not necessary to assume that investors have logarithmic utility functions. Equation (26), which provides a theoretic explanation of the ad hoc expression used by Rosenberg and McKibben [1973], explicitly demonstrates the contributory role of how the real variables of the firm affect the equilibrium expected rate of return. For the case of when there is only one economic variable (m = 1), then it can be demonstrated that the economic variable and the market portfolio are almost perfectly correlated.32 But, using the results of King [1966],33who found that the percentage of variance (of returns) explained by the market factor was less than 40 per cent, implies that there must be at least a second economic variable (m > 2). This would provide an explanation of the results of Brennan [1971], who found that the security return generating process must be represented by at least a two-factor model. For the case of when there are two economic variables the ex-ante expected rate of return for a portfolio is of the form34
ap-r=
eIk
/8ll[eil(aM-r)1]
+/31[
e,(aM-r)],
(28)
is a weighted average of the firm elasticities for the economic variable, Ik. where In general, {e, will not be constant over time. While the structural form of (28) is e} consistent with the two-variable form postulated by Black, Jensen, and Scholes [1972] and might provide an explanation of their results, confirmation can only be obtained by empirical testing.
IX. CONCLUSION
A general partial differential equation describing the market value of a firm is derived, explicit account being taken of the effects of a changing investment opportunity set upon the structure of asset returns. From this formulation an expression is obtained for the systematic risk of a firm in terms of elasticities, which measure the relative sensitivity of the firm's market value to changes in the firm
32. The correlation coefficient is given by
p= I/ (Il +
>
NJ aa var.
' )/[var(
NJ-a
where NJ is the number of shares outstanding for the jth firm. 33. See also Meyers [19731. 34. For many individual firms the coefficient ,Ba would be expected to be small, if not zero. If portfolio data are used the firm specific components, which are assumed to be independent across firms, would be diversified away.
1141
specific and economic variables, and a set of coefficients measuring the responsiveness of changes in the firm specific and economic variables to changes in the value of the market portfolio of risky assets. An explicit solution is derived for the market value of the firm and its systematic risk. This provides a theoretical and practical solution to the capital budgeting problem of determining the value of uncertain cash flows, as the risk premia can be estimated for a wide class of applications. It is demonstrated that systematic risk is a non-increasing function of growth and maturity. This implies that for capital budgeting decisions the firm's capitalization rate cannot, in general, be used as a hurdle rate for accepting or rejecting investment projects. The developed theoretical model for systematic risk provides insight into some of the deficiencies of recent empirical studies and has important implications for the empirical testing of the CAPM.
REFERENCES 1. W. Beaver, P. Kettler, and M. Scholes. "The Association Between Market Determined and Accounting Determined Risk Measures," Accounting Review, Vol. 45 (October 1970). 2. F. Black. "Corporate Investment Decisions," Associates in Finance, Financial Note No. 2B (May 1969). 3. F. Black, M. C. Jensen, and H. Scholes. "The Capital Asset Pricing Model: Some Empirical Results," Studies in the Theory of Capital Markets. Edited by M. Jensen. New York: Praeger, 1972. 4. M. Blume. "On the Assessment of Risk," Journal of Finance, Vol. XXVI, No. 1 (March 1971). 5. M. J. Brennan. "Capital Asset Pricing and the Structure of Security Returns," Working Paper, University of British Columbia, May 1971. 6. . "An Approach to the Valuation of Uncertain Income Streams," Journal of Finance, Vol. XXVIII, No. 3 (June 1973). 7. D. Durand. "Growth Stocks and the Petersburg Paradox," Journal of Finance, Vol. XII, No. 4 (September 1957). 8. E. Fama. "Multiperiod Consumption-Investment Decisions," American Economic Review, Vol. LX, No. 1 (March 1970). . "A Note on the Market Model and the Two-Parameter Model," Journal of Finance, Vol. 9. XXVIII, No. 5 (December 1973). 10. E. Fama and J. D. MacBeth. "Tests of the Multiperiod Two-Parameter Model," Journal of Financial Economics, Vol. 1, No. 1 (May 1974). 11. R. S. Hamada. "Portfolio Analysis, Market Equilibrium and Corporate Finance," Journal of Finance, Vol. XXIV, No. 1 (March 1969). . "The Effect of the Firm's Capital Structure on the Systematic Risk of Common Stock," 12. Journal of Finance, Vol. XXVII, No. 2 (May 1972). 13. M. C. Jensen. "Capital Markets: Theory and Evidence," Bell Journal of Economics and Management Science, Vol. 3, No. 2 (Autumn 1972). 14. B. F. King. "Market and Industry Factors in Stock Price Behavior," Journal of Business, Vol. XXXIX (January 1966). 15. H. J. Kushner. Introduction to Stochastic Control, New York, Holt, Rinehart and Winston, Inc., 1971. 16. E. M. Lerner and W. T. Carleton. A Theory of Financial Analysis, New York: Harcourt, Brace, & World, Inc., 1965. 17. H. A. Markowitz. Portfolio Selection: Efficient Diversification of Investments. Cowles Foundation Monograph No. ly. New York; John Wiley & Sons, 1959. 18. R. C. Merton. "Dynamic General Equilibrium Model of the Asset Market and Its Applications to the Pricing of the Capital Structure of the Firm," Massachusetts Institute of Technology, Sloan School of Management, Working Paper No. 490-70 (December 1970).
1142
19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32.
. "An Intertemporal Capital Asset Pricing Model," Econometrica, Vol. 41, No. 5 (September 1973). . "On the Pricing of Corporate Debt: The Risk Structure of Interest Rates," Journal of Finance, Vol. XXIX, No. 2 (May 1974). S. L. Meyers. "A Re-Examination of Market and Industry Factors in Stock Price Behavior," Journal of Finance, Vol. XXVIII, No. 3 (June 1973). F. Modigliani and M. Miller. "The Cost of Capital, Corporation Finance, and the Theory of Investment," American Economic Review, Vol. XLVIII, 1958. S. C. Myers. "Procedures for Capital Budgeting Under Uncertainty," Industrial Management Review, Vol. 9 (Spring 1968). . "The Relation Between Real and Financial Measures of Risk and Return," London Graduate School of Business Studies, 1975. and S. M. Turnbull. "Capital Budgeting and the Capital Asset Pricing Model: Good News and Bad News," Journal of Finance, Vol. XXXII, No. 2 (May 1977). A. Robichek and S. C. Myers. Optimal Financing Decisions. New Jersey: Prentice-Hall, Inc., 1967. B. Rosenberg and W. McKibben. "The Prediction of Systematic and Specific Risk in Common Stocks," Journal of Financial and Quantitative Analysis, Vol. VIII (March 1973). S. A. Ross. "Return, Risk, and Arbitrage," Working Paper No. 17-73a, The Wharton School, University of Pennsylvania, 1973. M. E. Rubinstein. "A Mean-Variance Synthesis of Corporate Financial Theory," Journal of Finance, Vol. XXVIII, No. 1 (March 1973). P. A. Samuelson. "Proof that Properly Anticipated Prices Fluctuate Randomly," Sloan Management Review (Spring 1965). W. F. Sharpe. "A Simplified Model for Portfolio Analysis," Management Science, (January 1963). R. W. White. "On The Measurement of Systematic Risk," Unpublished Ph.D. Dissertation, Massachusetts Institute of Technology, 1972.