Leverage and Taxes: Evidence from the Real Estate Industry

Abstract

Finance theory has long viewed corporate income taxes as a potentially important determinant of corporate financing decisions and capital structures. But finance academics have been unable to provide convincing empirical evidence of a material effect of taxes on corporate leverage, in part because of difficulties in constructing an effective proxy for marginal corporate tax rates, and hence for the tax benefits of debt, for large samples of individual companies.The authors address this by analyzing leverage decisions in an industry whose publicly traded entities are organized either as taxable corporations, or as real estate investment trusts (REITs) that effectively avoid entity level taxation. This enables them to measure the relative tax benefits of debt with greater precision while controlling for important nontax characteristics that affect debt usage. The tax hypothesis predicts that for real estate firms with similar asset portfolios, taxable firms should have more debt than their nontaxable counterparts. Both the nontaxable and the taxable real estate firms in our sample routinely have more than twice the leverage of industrial firms, which suggests that factors other than taxes are contributing to their use of debt. But among real estate firms, tax status appears to play a much weaker role. Taxable firms have significantly more leverage only after 2000, when restrictions on REITs were removed through new regulations that made their operations much more like those of taxable real estate firms. Our findings also depend on real estate characteristics—most notably, only residential real estate firms demonstrated differences that are consistent with the tax hypothesis.Taken together, the authors’ findings suggest that although taxes do seem to matter, their role is clearly secondary relative to factors such as the nature of the firm’s assets. A generous interpretation of our evidence puts the effect of taxes between one‐third and one‐half of that implied by prior research.