UNIVERSITY Of

ILLINOIS LIBRARY

AI URBANA-CHAb/lPAlQN

STACKS

Digitized by the Internet Archive

in 2011 with funding from

University of Illinois Urbana-Champaign

http://www.archive.org/details/commentsonstigli632whit

Faculty Working Papers

College of Commerce and Business Administration

University of Illinois at Urbana-Champaign

FACULTY WORKING PAPERS College of Coimnerce and Business Administration University of Illinois at Urbana-Champaign January 2, 1980

COMMENTS ON STIGLITZ'S RE-EXAMINATION OF THE MODIGLIANI-MILLER THEOREM

David T. Whitford, Lecturer, Department of Finance

#632

^

Summary;

The purpose of this paper is to clarify two misstatements in Joseph Stiglitz's landmark work, "A Re-Examination of the Modigliani- Miller Theorem." Although these revisions are essential for the original paper to stand alone, they do not alter the thrust of Stiglitz's thesis or his conclusions.

Comments on Stiglitz's Re-Examination of the Modigliani-Miller Theorem

The purpose of this paper is to isolate two apparent misstatements in Joseph Stiglitz's landmark paper, "A Re-Examination of the Modigliani- Miller Theorem." Its intent is not to discredit but clarify. Insertion of the suggested changes does not alter Stiglitz's conclusions. However, these inclusions are necessary for the paper to stand by itself.

Since the appearance of Modigliani and Miller's (M&M) classic paper, legitimate criticisms of their remarkable conclusions have focused on the validity of their assumptions. Stiglitz enumerates five limitations of the M&M proof. These are:

"1. It depend (s) on the existence of risk classes.

2. The use of risk classes seem(s) to imply objective rather than subjective probability distributions over the possible outcomes.

3. It (is) based on partial equilibrixjun rather than general equilib- rium analysis.

4. It (is) not clear whether the theorem held only for competitive markets.

5. Except under special circimistances, it (is) not clear how the possibility of firm bankruptcy affected the validity of the theorem." [p. 784]

Through a general equilibrium, state preference model, Stiglitz demonstrates that M&M's Proposition I "... holds under much more gen- eral conditions than those assumed in their original study. The validity of the theorem does not depend on the existence of risk classes, on compet-

-2-

itiveness of the capital market, or on the agreement of individuals about the probability distribution of outcomes." [Stiglitz, p. 784]

Two of M&M's assumptions are critical. These require that all market participants (firms and individuals) can borrow at the same market rate of interest, and that there is no risk of bankruptcy. Under a more stringent set of conditions, the MfiiM results will still be valid even though the exis- tence of limitations on individual borrowing and/or the possibility of bank- ruptcy is present. In detailing these more stringent conditions, Stiglitz has erred.

Limitations on Individual Borrowing

Stiglitz correctly refutes the objection that individuals cannot borrow at the same rate of interest as firms by emphasizing that individuals need not "actually borrow from the market but only change their holdings of bonds." Therefore he continues, "a problem can arise then only if an individual has no bonds in his portfolio." [p. 787] In equation (9) Stiglitz stipulates an inequality that constrains all individuals to be lenders.

(9) Z a^.B. > w^ - Z a^.V. Cfor all j) 1 i

where a. = the percentage of the i„ firm's equity owned by the

i individual, E-l/E. , J * i' 1 '

V = the value of firm i,

E. = the value of the i firm's equity,

B. = the value of the i firm's bonds, and 1 '

w-* = the value of the j individual's wealth.

Clearly then, V. H E. + B., 'ill

-3-

If we assume an environment of 3 individuals (j = a, b, c) and two firms (i = 1, 2) with market values:

V^ = E^ + B^,

V2 = E2 + B2,

where E , B. > 0 (i = 1, 2),

constraint (9) yields:

for j = a w - a,E^ - ct-E^ £. 2(a^B^ + a^B-),

for j = b w^ - a^E^ - O2E2 f. 2(a^B^ + a2B2) , and

for j = c w^ - a,E. - a^E„ < 2(a^B. + at^,,) .

xj. wz— ± 1. zz

These inequalities make very little economic sense. Moreover, it is not intuitively clear why this set describes the necessary and sufficient conditions for all individuals to be lenders. Equation (9') is a direct approach that requires all individuals to be lenders.

(9') w^ - S a^E. > 0 (for all i)

This inequality forces the wealth of each individual, w-' , minus the sum of his equity holdings in each firm, to be positive. Since all wealth is defined in terms of E. or B-' , the positive difference defined in (9') must be B-^ . Thus if (9') holds for every (j ) individual, everyone will own bonds (i.e. will be a lender). We can recast (9') in terms of V. rather than E. and offset the additional B. terms on righthand side of (9') as follows:

-4-

(9") w^ - I a^V. > - E a^B^,

.XI . X i

X X

which expands to

\P - I a^E. - l a^B. > - E ct^B..

XX .XX .XX

XX X

Because the B. terms cancel, (9') and (9'') are identical. Interest- ingly, (9'') may be transposed to give (9''').

(9'") E a^B. > - [w^ - E a^V.]

.XX .XX

X X

(9) E a^B. > [\^ - E a^V.]

X X — . XX X X

Equations (9''') and (9) are strikingly similar. It is possible that the difference results from a typographical error.

To check the suitability of the two formulations we can extend our simple financial environment consisting of two firms (i =1, 2), three individuals (j = a, b, c), and market values:

V^ = E^ + B^ and V^ = E2 + B2,

where E., B. > 0 (i = 1, 2). Also assume that: x' X

a^ = 1/2, a^ = 1/2, a^ = 1/2, a^ = 1/2, a^ = 0, a = 0.

If individual a owns 1/2 of B. and individual b owns the remaining 1/2 B^, this implies:

w^ = 1/2 E^ + 1/2 E2 + 1/2 B^,

w^ = 1/2 E^ + 1/2 E2 + 1/2 B^, and

w*^ = B2.

-5-

Plugging these values into Stiglitz's equation (9) infers:

3 1/2 B + B- >. 0 for j = a (correct),

1/2 B + B- >. 0 f or j = b (correct) , and

B„ ^ 0 for j = c (incorrect).

However, if these values are applied to equation (9*"), they yield:

1/2 B > 0 for j = a (correct),

1/2 B > 0 for j = b (correct), and

B- > 0 for j = c (correct) .

On the other hand, if only individual c own bonds, while individuals a and b own only equity (in the previous proportions), the following w^ would be appropriate:

w^ = 1/2 E^ + 1/2 E2,

w^ = 1/2 E + 1/2 E2, and

w*^ = B^ + B2.

Stiglitz's eqtiation (9) infers:

B + B^ >_ 0 for j = a (incorrect),

B^ + ^2 ^ 0 for j = b (incorrect), and

B + B„ <_ 0 for j = c (incorrect).

-6-

while (9 ' " ) gives : i

0 > 0 for j = a (correct) ,

0 > 0 for j = b (correct), and

B. + B„ > 0 for j = c (correct) .

Thus it appears (9'") is the appropriate constraining condition to insure that all individuals hold some bonds in a portion of their portfolio.

Mean Variance Analysis and the Separation Theorem in the Presence of Bankruptcy

In section III, Stiglitz establishes that even if a firm issues enough bonds to go bankrupt, as long as all individuals agree on the probability distribution of returns for each firm, the value of that "risky" firm is independent of its debt equity ratio.

Most treatments of mean-variance analysis focus on rates of return. Indeed, those familiar with the Sharpe-Lintner capital asset pricing model (CAPM) should recognize Stiglitz' s assumption as a restatement of the

homogeneous expectations or "idealized" uncertainty requirement. Fama

4 details several alternative versions of the CAPM.

(1) e(r^) = r* + S^[e(r^) - r*] ,

[e(r ) - r*] cov(r ,r ) (1») = r* + ? ^ "^

2 *

m

(1") = r* + A cov(r.,r ),

where r. = the rate of return on the ith risky asset (or portfolio),

-7-

r* = the rate of return on a risk-free asset (i.e. the risk-free rate),

r = the rate of return on the market portfolio of risky assets,

cov(r.,r ) = the covariance of return on the i asset with the market return, i' m '

2 a (r ) = the variance of the market return, m '

B. = cov(r.,rJ/a2(r^),

e(r^) - r*

A = J = the market price of risk, and

m e = the expectations operator.

Instead of focusing on rates of return, Stiglitz (referring to Lintner) places his emphasis on stock and flow variables (i.e. market values and "operating" dollar returns or operating income. Although one set of variables (rates of return) is the mirror image of the other (value and dollar returns) and vice versa, the transition between variables is not always smooth. Nevertheless, it should be evident that if Stiglitz 's presentation is correct, it should collapse into the CAPM.

We will reconstruct Stiglitz 's equations (10) through (12), based upon Lintner's equations (29) through (29'') because they served as Stiglitz 's point of embarkation.

(29) V. = (R. - W.)/r*,

_ v

(29') = [R^ - (X/T) Z R^ ]/r*,

J

E(R. - r*V.) (29") = [R - i l R ]/r*,

IJ

where V. = aggregate market value of firm i,

R. = the expected level of operating income for firm i.

-8-

R. . = the covariance of the operating incomes of firms i and j , and r* = the risk-free rate. Stiglitz's equation (10) is a restatement of (29'); however, a more precise version is (10').

(10) E^ + B^ = Jx - k Z e(X^ - X^)(X. - X.)W r*

r n

(10') V. = E. + B. = ,1 1 1

X^ - k E e[(X^ - X^)(X - X )]^ / r*, L j=l

where E., B., B., e and r* are as previously defined and

X. = the expected level of operating income for firm i.

Since X. = Lintner's R., we can rewrite (10') as follows; 1 1

(10

'•) V. = ix. - k E X..I / r*. ^ l^ j=l ^']

Comparing (10") with (29') it should be clear that k = X/T. Stiglitz defines k as:

k = r* [E (X. - X.) / ZZe(X. - X.)(X. - X.)].

However, since E(X. - X.) would equal zero, k would equal zero. It

i should be evident that this was not Stiglitz 's intent. If this were true,

the market value of every firm would equal the value of its gross dollar

returns capitalized at the risk-free rate. By writing Stiglitz 's variables

into Lintner's structure, we can derive a new k.

E(X. - r*V.) .1 1

k = ^--

V

EEX. .

-9-

Z(X. - r*V.)

.1 X

1

EEe[(X. - X.)(X. - X.)]

the excess dollar retiirns (flows) of the market (excess above the risk-free rate times the aggregate market value

of all firms comprising the market)

the variance of the dollar market return

Substituting this new k into (10') yields a close surrogate for Lintner's original equation (29).

(2) V^ = (X^ - n^) I r*,

2:(X - r*V ) E e[(X - X ) (X - X )]

where fi. = "^

1

VL e [(X. - X.)(X. - X.)] ij ^ ^ ^

Multiplying Q.. by 1/SV ZV, / 1/ZV ZV. does not change its value;

^ i J y i "-j ^

however, it transforms dollar flows (X, and X.) into rates of return (r . and r . ) • This procedure yields :

[e(r ) - r*] Z e[(X. - X.)(r. - 7.)]

"-j_ ~ J •

ZZ e [(r - 7 )(r -7 )] ij 1 1 J J

A closer inspection of ^. shows that

and

Z e [(X. - X^)(r - r )] = cov(X.,r^),

J

ZZ e [(r - r )(r - r )] = cov(r ,r ) ij 1 1 J J - "^ "»

m

Inserting these terms into (2) gives

-10-

[e(r ) - r*]cov(X, ,r )

By dividing both sides of (2') by V. and cross-multiplying, we obtain

[e(r ) - r*]cov(r ,r ) . . V m 1 m

r* = e(r.) -

m

or e(r.) = r* + 6.[e(r ) - r*] , — X 1 m

cov(r ,r ) where

m which is of course identical to (1). Thus, the reformulation of k must

be correct.

Conclusions

This paper has isolated two revisions required for exactness and rigor. The first is a reformulation of Stiglitz's equation (9). The second restates a variable, k, utilized in equations (10) through (12). Although the corrections are essential for the original paper to stand alone, they do not alter the thrust of Stiglitz's thesis or his conclusions.

-11-

FOOTNOTES

Proposition I states "... the market value of any firm is independent of its capital structure and is given by capitalizing its expected return at the rate p, appropriate to its risk class" . . . "the average cost of capital to any firm is completely independent of its capital structure and is equal to the capitalization rate of a pure equity stream of its class." [Modigliani & Miller, p. 268]

2 The superscript refers to individuals; the subscript pertains to firms.

3 This remark refers to the constraint's suitability and accuracy in

describing the given financial environment. For example. If each individual

in our hypothetical environment owns bonds, the constraint must describe

a feasible condition for every individual. On the other hand, if the

environment has an individual who owns no bonds, the constraint must

describe an infeasible condition for at least one individual.

4

Admittedly a precise statement of equations (1-1'') would include com- plementary random error terms. Fortunately for our purposes, the assump- tions of the model (see Fama, pp. 37-AO) obviate this requirement.

Conventionally, "operating income" is income before financial charges. Stiglitz uses the term gross returns, i.e. the dollar return before paying bondholders but after paying all non-capital factors of production.

As a general rule rates of return are represented as lowercase letters; values and dollar flows are uppercase letters.

This, of course, assumes that X. is a normally distributed random variable.

-12-

REFERENCES

E. Fama, "Risk, Return and Equilibrixnn: Some Clarifying Comments," Journal of Finance, March 1968, 23, 29-40.

J. Lintner, "The Valuation of Risk Assets and the Selection of Risky Investments in Stock Portfolios and Capital Budgets," Review of Economics and Statistics, February 1965, 47, 13-37.

F. Modigliani and M. H. Miller, "The Cost of Capital, Corporation Finance, and the Theory of Investment," American Economic Review, June 1958, 48, 261-297.

J. E. Stiglitz, "A Re-Examination of the Modigliani-Miller Theorem," American Economic Review, December 1969, 59, 784-793.

M/E/83