In the M&M model, the trade-off between financial risk and the cost
of funds is unitary; if more debt is added to the financial mix, the cost of
debt rises and the desired rate of return on equity rises, so that the
weighted cost of the financial mix remains constant. Let us briefly recount
the three propositions of M&M in their seminal presentation of the cost of
capital and valuation. First, M&M hold for a class of similar (homoge-
neous) firms—those firms that are perfect substitutes of each other, such as
the industry concept—the cost of capital is a constant, ρo. The constant is
determined by dividing the expected return per share by the stock price.
This constant is the market rate of capitalization for firms in a particular
class of firms. Thus, the average cost of capital is independent of capital
structure.
Second, the expected stock yield is equal to the return on a pure equity
firm (the return on assets ρo) plus a premium for financial risk proportional
to the return on assets less the interest rate. Our earlier Lerner-Carleton de-
rivation is a variation of the M&M Proposition II. M&M argued that the
firm must earn a return on investments exceeding ρo. M&M’s Proposition
III holds that if the firm earns at least ρoon its investments, the project is
acceptable regardless of the securities issued to finance the investment.
M&M presented empirical evidence in their 1958 study, using the 40-firm
electric utilities study of Allen (1954) and the 42-firm oil companies sample
of Smith (1955). Both Allen and Smith provided data on the average values
of debt and preferred stocks and market values of securities, such that
M&M could calculate the debt-to-total value of securities ratio, d. M&M
regressed the net returns, x, defined as the sum of interest, preferred divi-
dends, and net income, as a function of ratio d. The Allen electric utilities
sample covered 1947–1948 and the Smith sample of oil companies was for
the year 1953. The M&M regressions were:
Electric utilities: x= 5.3 + .006d
(.008)
Oil companies: x= 8.5 + .006d
(.024)M&M held that the regression results supported their Proposition I. The
calculated t-statistics, found by the ratio of the regression slope, b, divided
by its standard error (in parentheses), should be 1.96 (or 1.645 at the 10
percent level) to be statistically significant. The calculated t-statistic on the
electric utilities sample is 0.75, far less than 1.645. The calculated t-statistic
for the oil companies sample is 0.25. Thus, there is no statistical significance
between net returns and the debt-to-assets ratio in the initial M&M study.
We take a detailed look at hypothesis testing in Chapter 9. M&M used