American Finance Association

Optimal Investment and Financing Policy

Author(s): M. J. Gordon
Reviewed work(s):
Source: The Journal of Finance, Vol. 18, No. 2 (May, 1963), pp. 264-272
Published by: Blackwell Publishing for the American Finance Association
Stable URL: http://www.jstor.org/stable/2977907 .
Accessed: 25/11/2011 23:53

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
 MANAGEMENT OF CORPORATE CAPITAL

OPTIMAL INVESTMENT AND FINANCING POLICY*

M. J. GORDONt

IN TWO PAPERS1 AND IN a recent book2 I have presented theory and

evidence which lead to the conclusion that a corporation's share
price (or its cost of capital) is not independent of the dividend rate.
As you may know, MM (Modigliani and Miller) have the opposite
view, and they argued their position at some length in a recent
paper.3 Moreover, the tone of their paper made it clear that they
saw no reasonable basis on which their conclusion could be ques-
tioned. Since they were so sure of their conclusion, it would seem
advisable for me to review carefully my thinking on the subject,
and this meeting appears to be a good time and place to do so.
I
Let us begin by examining MM's fundamental proof that the
price of a share is independent of its dividend. They defined the
value of a share at t - 0 as the present value of (1) the dividend it
will pay at the end of the first period, Di, plus (2) the ex-dividend
price of the share at the end of the period, Pi:

pO=.1 +1 k [D1+Pi] (1)

They then asked what would happen if the corporation, say, raised
its dividend but kept its investment for the period constant by
selling the additional number of shares needed to offset the funds
* This paper and the following papers by Ezra Solomon, James E. Walter, and John
Lintner, with discussions by Herbert Dougall, Merton Miller, and Robert F. Vandell,
were presented at a meeting of the American Finance Association in Pittsburgh, Pa.,
on December 29, 1962. The program was under the chairmanship of J. Fred Weston.
t Professor of business economics, University of Rochester.
1. "Dividends, Earnings and Stock Prices," Review of Economics and Statistics, May,
1959, pp. 99-105; "The Savings, Investment and Valuation of the Corporation," ibid.,
February, 1962, pp. 37-51.
2. The Investment Financing and Valuation of the Corporation (Homewood, Ill.:
R. D. Irwin, 1962).
3. "Dividend Policy, Growth, and the Valuation of Shares," Journal of Business,
October, 1961, pp. 411-33.
264
 OptimalInvestment and Financing Policy 265
lost by the dividend increase. They demonstrated that the ex-divi-
dend price of the stock at the end of the period would go down by
exactly the same amount as the increase in the dividend. Since the
sum D1 + Pi remains the same, Po is unchanged by the change in
the dividend.
I will not review their proof of the theorem in detail because I
find nothing wrong with it under the assumption they made that the
future is certain. However, after proving the theorem a number of
times under different conditions, they withdrew the assumption of
certainty and made the dramatic announcement,"our first step, alas,
must be to jettison the fundamental valuation equation."4 Under
uncertainty, they continued, it is not "at all clear what meaning can
be attached to the discount factor... ."5 The implication which they
made explicit in discussing my work is that under uncertainty we
cannot represent investors as using discount rates to arrive at the
present value of an expectation of future receipts.
It would seem that all is lost. But no! On the very next page we
are told that their "fundamental conclusion need not be modified
merely because of the presence of uncertainty about the future
course of profits, investment, or dividends. ...6 By virtue of the
postulates of "imputedrationality" and "symmetricmarket rational-
ity," it remains true that "dividend policy is irrelevant for the de-
termination of market prices."
Their paper continued with a discussion of market imperfections,
in which they note that the most important one, the capital gains
tax, should create a preference for low payout rates. They concede
that it may nevertheless be true that high payout rates sell at a
premium, but they found ". . . only one way to account for it, namely
as a result of systematic irrationality on the part of the investing
public." They concluded with the hope that ". . . investors, however
naive they may be when they enter the market, do sometimes learn
from experience; and perhaps, occasionally even from reading arti-
cles such as this."8
It would seem that under uncertainty they might have been less
sure of their conclusion for two reasons. First, under uncertainty
an invester need not be indifferentas to the distribution of the one-
period gain on a share between the dividend and price appreciation.
Since price appreciation is highly uncertain, an investor may prefer
4. Miller and Modigliani, op. cit., p. 426.
5. Ibid., p. 427. 7. Ibid., p. 429.
6. Ibid., p. 428. 8. Ibid., p. 432.
266 The Journal of Finance
the expectation of a $5 dividend and a $50 price to a zero dividend
and a $55 price without being irrational. Second, the expectation
of a stock issue at t - 1 may have a depressing influence on the
price at t- 0. What MM did was both change the dividend and
change the number of new shares issued. Can we be so sure that
the price of a share will not change when these two events take
place?
II
Let us turn now to the proof of the MM position on the dividend
rate that I presented in my RES paper and book. The reasons for
presenting this proof will be evident shortly. Consider a corporation
that earned Yo in the period ending at t 0 and paid it all out in
dividends. Further, assume that the corporation is expected to con-
tinue paying all earnings in dividends and to engage in no outside
financing. Under these assumptions the company is expected to
earn and pay Yo in every future period. If the rate of return on
investment that investors require on the share is k, we may repre-
sent the valuation of the share as follows:
P + .
(2)
YO+=i -'F +YO
( + k)l (+ k)2 ( + k)3 (+ k)t
We may also say that k is the discount rate that equates the divi-
dend expectation of Yo in perpetuity with the Price Po.
Next, let the corporationannounce at t 0 that it will retain and
invest Y1 Y
Yo during t 1 and that it expects to earn a rate of
return of k - Yo/Po on the investment. In each subsequent period
it will pay all earnings out in dividends. Share price is now given by
the expression
P 0 Yo+ kYo Yo+ kyo+ + Yo+ kYo
(1 + k)' (1 + k)2 (1 + k)3 (1 + k)t(3
Notice that the numeratorof the first term on the right side is zero.
It is the dividend and not the earnings in the period, since the in-
vestor is correctly represented as using the dividend expectation in
arriving at Po. If he were represented as looking at the earnings
expectation, then as Bodenhorn9noted, he would be double-counting
the first period's earnings.
It is evident that, as a result of the corporation's decision, the
investor gives up Yoat the end of t 1 and receives, in its place, kYo
9. Diran Bodenhorn, "On the Problem of Capital Budgeting," Journal of Finance,
December, 1959, pp. 473-92.
 OptimalInvestment and Financing Policy 267
in perpetuity. The distribution of dividends over time has been
changed. It is also evident that k Yo in perpetuity discounted at k
is exactly equal to Yo. Hence Po is unchanged, and the change in
the distribution over time of the dividends had no influence on share
price. In general, the corporation can be expected to retain and in-
vest any fractionof the income in any periodwithout share price being
changed as a consequence, so long as r, the return on investment, is
equal to k. If r > k for any investment, Po will be increased, but the
reason is the profitability of investment and not the change in the
time distribution of dividends.
Assume now that when the corporation makes the announcement
which changes the dividend expectation from the one given by equa-
tion (2) to the one given by equation (3), investors raise the dis-
count rate from k to k'. For the moment let us not wonder why the
discount rate is raised from k to k', i.e., why the rate of return in-
vestors require on the share is raised as a consequence of the above
change in the dividend expectation. If this takes place, equation (3)
becomes

PO (1+k')' (l+k' )2 (1 + Y)3 (1 + )4

It is clear that with k' > k, Po' < Po.

Let us review what happened. The dividend policy changed: the
near dividend was reduced, and the distant dividends were raised.
This caused a rise in the discount rate, and the result was a fall in
the price of the share. I, therefore, say that the change in dividend
policy changed the share's price.
In response to this argument, MM stated that I fell into "the
typical confounding of dividend policy with investment policy."'0
I don't understand their reasoning. It is well known that when the
rate of return on investment is set equal to the discount rate, chang-
ing the level of investment has no influence on share price. By this
means, I neutralized the profitability of investment. It seems to me
perfectly clear that I did not confound investment and dividend
policy; I changed the discount rate. Share price changed with the
dividend rate in the above example because the discount rate was
changed. The issue, therefore, is whether the behavior of investors
under uncertainty is correctly represented by a model in which the
discount rate that equates a dividend expectation with its price is a
function of the dividend rate.
10. Miller and Modigliani, op. cit., p. 425.
268 The Journal of Finance
I cannot categorically state that k is a function of the rate of
growth in the dividend, i.e., the dividend rate, but I can present
some theoretical considerationsand empirical evidence in support of
the theorem. It seems plausible that (1) investors have an aversion
to risk or uncertainty, and (2), given the riskiness of a corporation,
the uncertainty of a dividend it is expected to pay increases with the
time in the future of the dividend. It follows from these two propo-
sitions that an investor may be represented as discounting the divi-
dend expected in period t at a rate of kt, with kt not independentof t.
Furthermore,if aversion to risk is large enough and/or risk increases
rapidly enough with time, kt increases with t.
It is therefore possible, though not certain, that investor behavior
is correctly approximatedby the statement that, in arriving at the
value of a dividend expectation, they discount it at the rates kt, t =
1, 2 ... , with kt > kt-1. In this event the single discount rate we use
in stock value models is an increasing function of the rate of growth
in the dividend. In short, dividend policy influences share price. To
illustrate the conclusion, let us rewrite equation (2):

P= __
( 01+ k)+ (I + kt 2 ) +( *
We now look on the k of equation (2) as an average of the kt of
equation (4) such that if the entire dividend expectation is dis-
counted at this single rate, it results in the same share price. The
discount rate k is an average of the kt with Yo, the weight assigned
to each item.
Once again let the corporation retain YV= Yo and invest it to
earn kYo per period in perpetuity. Using the sequence of discount
rates kt, the same as that appearingin equation (4), the valuation of
the new dividend expectation becomes

P/ = O Yo+ kYo Yo+ kYo Y0= kYo+ 5

? (1 +kl)l l ++k2)2 ( +k-) . (1 +kt)t
The shareholder gives up Yo and gets kYo in perpetuity, but the
latter is now discounted at the rates kt, t = 2-o, and it can be
shown that kYo so discounted is less than Yo. Hence Po' < Po, and
dividend policy influences share price. It also can be shown that
k', the new average of the same kt, is greater than k. In general,
reducing the near dividends and raising the distant dividends (low-
ering the dividend rate) changes the weights of the kt and raises
their average.
 OptimalInvestment and Financing Policy 269
III
To summarizethe theoretical part of my argument, I started with
two assumptions: (1) aversion to risk and (2) increase in the un-
certainty of a receipt with its time in the future. From these assump-
tions I proceeded by deductive argument to the proposition that
the single discount rate an investor is represented as using to value
a share's dividend expectation is an increasing function of the rate
of growth in the dividend. The consequence of the theorem is that
dividend policy per se influences the value of a share. The assump-
tions have enough intuitive merit, I believe, that the theorem may
in fact be true.
Before proceeding to the empirical evidence, I would like to com-
ment briefly on two other criticisms MM directed at my argument.
First, they differentiated between my "purely subjective discount
rate and the objective market-givenyields" and stated: "To attempt
to derive valuation formulas from these purely subjective discount
factors involves, of course, an error. . .. My assumptions and em-
pirical results may be questioned, but where is the error? Does the
theorem fail to follow from the assumptions? Why, as they suggest,
is it logically impossible for an investor to arrive at the value of a
share by estimating its future dividends and discounting the series
at a rate appropriateto its uncertainty?
The following MM criticism of my argument I find even more
confusing. They stated: "Indeed if they [investors] valued shares
accordingto the Gordonapproachand thus paid a premiumfor higher
payout ratios, then holders of the low payout shares would actually
realize consistently higher returnson their investment over any stated
interval of time."" Under this reasoning two shares cannot sell at
different yields regardless of how much they differ in risk because
the holders of the higher-yield share would "actually realize con-
sistently higher returns over any stated interval of time." Do MM
deny that investors have an aversion to risk?
To test the theorem empirically, I proceeded as follows. The valu-
ation of a share may be represented by the expression

Po = fDt e-ktdt, (6)

where Dt is the dividend expected in period t and k is an operator

on the Dt that reduces them to their present value to the investor.
11. Ibid., p. 424. 12. Ibid., p. 425.
270 The Journal of Finance
Equation (6) is a perfectly general statement that is not open to
question. However, to use the equation in empirical work, we must
specify how investors arrive at DA from observable variables. For
this, I assumed that investors expect a corporation will: (1) retain
the fraction b of its income in each future period; (2) earn a rate
of return,r, on the common equity investment in each future period;
(3) maintain the existing debt-equity ratio; and (4) undertake no
new outside equity financing. Under the above assumpions the cur-
rent dividend is Do _ (1 - b) lfo, and its rate of growth is br. Fur-
ther, the entire dividend expectation is represented by these two
variables, and equation (6) is equal to
(1-b)Yo 7
Po~k- br
The above four assumptions may be criticized as being too great
a simplification of reality. I have admitted their limitations, and I
welcome improvement,but I know of no other empirical model that
contains as rich and accurate a statement of the dividend expectation
provided by a share. Most empirical work, including the published
work of MM, represents the investor as expecting that the corpora-
tion will pay all earnings in dividends and engage in no outside
financing. They, therefore, also ignore the influence of the profita-
bility of investment on share price. This model incorporates a pre-
diction of the corporation's investment and rate of return on the
investment in each future period. The expected investment in period
t is the fraction b of the period's income plus the leverage on the
retention that maintains the corporation'sexisting debt-equity ratio.
Further, the influence of this retention and borrowing on the divi-
dend expectation is incorporatedin the model.
The interesting thing about the model as it stands is that it is
consistent with the MM position and should provoke no objection.
To see this, let us make their assumption that k is independent of
b and, to neutralize the profitability of investment, let r - k. In this
model, dividend policy is represented by b the retention rate, so
that, if we take the derivative of Po with respect to b, we establish
the relation between share price and the dividend rate. We find that
SP18bO= 0. The value of a share is independent of the dividend
rate-exactly what MM argue.
One can use this model in empirical work under the assumption
that k is independent of br. I did and obtained poor results. Since
I found good theoreticalgrounds for believing that k is an increasing
 OptimalInvestment and Financing Policy 271
function of br, it would seem reasonable to explore the hypothesis,
and that is what I did. If k is an increasing function of br, we can
write equation (7) as
Po= Ao[ (1-b) Yo] [1 + br]a2. (8)
In this expression, Ao represents the influence of all variables other
than the current dividend, (1 - b) Yo, and its rate of growth, br.
When b - 8, Po is the multiple Ao of Yo. As br increases, the divi-
dend, (1 - b) Yo, falls and br rises, the former lowering price and
the latter raising price. Whether Po rises or falls with b depends on
r, the profitability of investment, and on a2. The expression a? may
be looked on as how much investors are willing to pay for growth.
Its value depends on how fast the kt rise with t, that is, on how fast
uncertainty increases with time and on the degree of investor aver-
sion to risk.
It should be noted that equation (8) is not merely a stock value
model. Given the investor's valuation of a share, Ao and a2, and,
given the profitability of investment, r, the model may be used to
find the retention rate (equal to the investment rate under our as-
sumptions) that maximizes the value of a share. Extensions of the
model developed elsewhere'3allow its use to find the investment and
the financing, retention, debt, and new equity that maximize share
price.
The empirical results I obtained with the above model have been
published in detail,"4and all I will say here is that they are very
good. Although the results compare favorably with earlier work,
they are not good enough to settle the question. MM"5and Beni-
shay'6 have pointed out that my independent variables are not free
of error, and the consequenceis that the parameter estimates have a
downward basis. Kolin'7 has reported that his empirical work re-
vealed no relation between dividend policy and share price. As things
stand, I would say that the influence of dividend policy on share
price is a question that requires further study. The axiomatic basis
13. M. J. Gordon, The Investment, Financing and Valuation of the Corporation
(Homewood, Ill.: R. D. Irwin, 1962).
14. Ibid.
15. Franco Modigliani and Merton Miller, "The Cost of Capital Corporation Finance,
and Theory of Investment: Reply," American Economic Review, September, 1959, pp.
655-69.
16. Haskel Benishay, "Variability in Earnings-Price Ratios: Reply," American Eco-
nomic Review, March, 1962 pp. 209-16.
17. Marshal Kolin, The Relative Price of Corporate Equity (Boston: Harvard Busi-
ness School).
272 The Journal of Finance
of the MM position is certainly not so powerful as to force the ac-
ceptance of their conclusions.
IV
I should like to close with a brief comment on the two major
camps that are emerging with respect to the theory of corporation
finance. In both camps optimal policy is taken as the policy that
maximizes the value of the corporation.Although corporationsmay
not make investment and financingdecisions with only this objective
in mind, managements are certainly not indifferent to the prices at
which their corporations' securities sell. Hence the policy question
posed has practical significance.
In one camp, where we find MM, it is argued that a corpora-
tion's cost of capital is a constant-i.e., independent of the method
and level of financing. Optimal policy is the investment that equates
the marginal return on investment with this cost of capital. The
inescapable conclusion is that financing policy is not a problem. The
opposite position is that a corporation's cost of capital varies with
the method and level of financing. My judgment is that the theoreti-
cal and empirical evidence we have favors this position.
However, regardless of which view prevails, the battle should be
lively and productive. For a long time the position that cost of capital
is a constant was held almost exclusively by economists, who were
sophisticated in methods of theoretical and econometric analysis
but knew little of finance. By contrast, the position that the cost
of capital is a variable was held by finance men, who were familiar
with their subject but not with advanced methods of theoretical
and empirical research. People in each group talked only to those
who agreed with them, and in consequence not much was said. The
situation has changed, it will change further, and the promise is that
the lively debate and active research in progress will advance our
knowledge on the subject.