Industrial revenue bonds (IRBs) possess special tax characteristics which provide an opportunity to test the interest tax shield hypothesis of Modigliani and Miller [1963]. The announcement day excess returns for IRB issues reflect a positive significant market reaction. However, this excess return is unrelated to the tax shield on the debt. Nor is it a function of non-debt tax shields, a proxy for an interior optimal capital structure, or the risk of the issue. It is consistent, however, with the argument that lRBs provide a subsidy to issuing firms. Miller's [1977] equilibrium model predicts that the tax rate of the marginal buyer of debt will be equal to the corporate tax rate. The Miller hypothesis is tested by comparing the yield spread between IRBs and taxable corporate debt. The empirical estimation indicates a segmentation of the market for tax-exempt debt. For short-term issues, the implied tax rate is not significantly different from the corporate tax rate, consistent with Miller’s prediction. For 1ong·term issues, the implied tax rate is significantly less than the corporate tax rate and decreases with maturity. This is consistent with the excess supply of tax-exempts being purchased by individuals in increasingly lower tax brackets. / Ph. D.