Capital structure is irrelevant in an ideal world because agency and financial distress costs, for example, will adjust to keep the total cost of debt the same. In reality, the benefits of debt, such as reduced agency costs and improved debtholder oversight, have to be offset against the costs of debt, notably an increase in the cost of financial distress. The optimal capital structure depends on balancing the marginal costs and benefits. The authors improve this process by quantifying the cost of debt and by studying how this cost relates to the characteristics of a company.

The authors use corporate financial statement data from the Compustat database for 1980–2007 (126,611 company-year observations in total). They reduce the sample by removing companies with data containing unexpected negative values or outliers or with missing data. Companies in which substantial merger and acquisition activity occurred and companies from heavily regulated sectors are also removed, which leaves a research sample of 97,942 company-year observations. They also analyze a subsample of 12,833 companies that are nondistressed (determined on the basis of Z-scores) and unconstrained by fixed-adjustment costs (because the company initiated a large long-term debt or equity issuance in the year and is likely to have reached equilibrium). The authors repeat the analyses for a further subsample that focuses on the period when changes resulting from the Tax Reform Act of 1986 were occurring.

They estimate the marginal benefit of debt (i.e., tax benefits reduced at higher levels of debt when the contemporary taxable profits are insufficient and the benefits have to be carried backward or forward). They use changing marginal benefit curves (primarily as a result of tax changes) to obtain different points on the marginal cost curve (assuming equilibrium) from which the marginal cost curve is then estimated. The authors find that the marginal cost of debt increases with the amount of debt used, which illustrates the increased cost of financial distress at higher debt levels.

By including control variables, they show that high collateral lowers the cost curve. A surprising result is that larger companies behave as if they have a higher cost of debt than smaller companies because larger companies use less debt. Previous research shows mixed results in this area. It appears that size has an influence on both the marginal cost and the marginal benefit of debt, so results on the use of debt may vary in different settings.

Companies with growth opportunities (i.e., low book-to-market value) have a higher cost of debt than those with a higher book-to-market value. This finding is consistent with previous research that shows that debt overhang, such as debt servicing and debt covenants, may inhibit the exercise of future growth opportunities.

Companies with intangible assets, however, experience a lower cost of debt, possibly because in this context intangible assets act like collateral and can thus be used to support debt claims. Companies with higher cash flows have a higher cost of debt than companies with lower cash flows, which is consistent with the pecking-order theory. Finally, companies that pay higher dividends face a higher cost of debt than those that pay lower dividends, possibly because dividends are rarely omitted and, therefore, leave fewer funds to service the debt.

The cost functions and benefit functions enable the authors to determine the net benefit of using debt. The average of net benefits for the full sample of companies equals 3.5 percent of book value. In this sample, they find that the net costs of overleveraged companies are greater than those of underleveraged companies. The findings also allow the authors to estimate the cost function and benefit function for any company and to determine the company’s optimal capital structure.