MODIGLIANI AND MILLER: THEORY OF INVESTMENT 273 pe can no longer be identified with the "average cost of capital which is pk=X,/V. The difference between Pe and the"true"average cost of apital, as we shall see, is a matter of some relevance in connection with investment planning within the firm( Section If). For the description of market behavior however, which is our immediate concern here, the dis- tinction is not essential. To simplify presentation, therefore, and to pre serve continuity with the terminology in the standard literature we shall continue in this section to refer to pr" as the average cost of capital, though strictly speaking this identification is correct only in the absence of taxes Efects of a Plurality of Bonds and Interest Rates. In existing capita markets we find not one, but a whole family of interest rates varying with maturity, with the technical provisions of the loan and, what is most relevant for present purposes, with the financial condition of the borrower.6 Economic theory and market experience both suggest that the yields demanded by lenders tend to increase with the debt-equity ratio of the borrowing firm(or individual) If so, and if we can assume as a first approximation that this yield curve, r=r(D/S), whatever its precise form, is the same for all borrowers, then we can readily extend our propositions to the case of a rising supply curve for borrowed funds. I Proposition I is actually unaffected in form and interpretation by the fact that the rate of interest may rise with leverage; while the average cost of borrowed funds will tend to increase as debt rises, the average cost of funds from all sources will still be independent of leverage(apart from the tax effect). This conclusion follows directly from the ability of those who engage in arbitrage to undo the leverage in any financial structure by acquiring an appropriately mixed portfolio of bonds and stocks. Because of this ability, the ratio of earnings(before interest charges) to market value--i e,, the average cost of capital from all 16 We shall not consider here the extension of the analysis to encompass the time structure of interest rates, Although some of the problems posed by the time structure can be handled witl in our comparative statics framework, an adequate discussion would require a separate paper. IT We can also develop a theory of bond valuation along lines essentially parallel to those fol- lowed for the case of shares. We conjecture that the curve of bond yields as a function of lever- age will turn out to be a nonlinear one in contrast to the linear function of leverage developed r common shares. However, we would also expect that the rate of increase in the yield on hew issues would not be substantial in practice. This relatively slow rise would refect the fact compensation for their increased ris tive to p that they become self-defeating by giving rise to a situation in which even normal fluctuations ly force the company into bankruptcy. The difficulty of borrowing more, therefore, tends to show up in the usual case not so much in higher rates as in the form th bility to obtain new borrowed funds, at least from the institutional investors who normally set the standards in the market for bonds