What is the conclusion of the article?

IV. Summary and Conclusions This paper develops a model that synthesizes the modern balancing theory of optimal capital structure. This model incorporates positive personal taxes on equity and on bond income, expected costs of financial distress (bankruptcy costs and agency costs), and positive non-debt tax shields. We show that optimal firm leverage is related inversely to expected costs of financial distress and to the (exogenously set) amount of non-debt tax shields. A simulation analysis dem- onstrates that if costs of financial distress are significant, optimal firm leverage is related inversely to the variability of firm earnings. The empirical section investigates the cross-sectional behavior of 20-year average firm leverage ratios for 851 firms covering 25 two-digit SIC industries. Several important results emerge. First, there exists strong industry influences across these firm leverage ratios. The cross-sectional regressions on industry dummy variables explain 54% of variation in firm leverage ratios. Excluding from the regression all regulated firms during the sample period, such as trucking, telephone, electric and gas utilities, and airlines, still yields an R2 of 25%. Our search for the specific economic sources of these strong industry influ- ences on firm leverage ratios yields some noteworthy results. The volatility of firm earnings is an important, inverse determinant of firm leverage. It helps explain both inter- and intra-industry variations in firm leverage ratios. The intensity of R&D and advertising expenditures is also related inversely to leverage. Both of these results are consistent with the formal balancing model of optimal leverage.

A somewhat puzzling finding is the strong direct relation between firm leverage and the relative amount of non-debt tax shields. This contradicts the theory that focuses on the substitutability between non-debt and debt tax shields. A possible explanation is that non-debt tax shields are an instrumental variable for the securability of the firm's assets, with more securable assets leading to higher leverage ratios. A fundamental problem with the cross-sectional regressions is misspecifi- cation, which suggests a "missing variable" explanation for the perverse result on non-debt tax shields. The danger is that excluded variables are correlated with included variables, which can cause misleading inferences to be drawn from the regression results. Nonetheless, the strong finding of intra-industry similar- ities in firm leverage ratios and of persistent inter-industry differences, together with the highly significant inverse relation between firm leverage and earnings volatility, tends to support the modern balancing theory of optimal capital structure.