Some corporate decisions increase stockholder wealth while reducing the wealth of bondholders. When wealth transfers are large enough, stock prices can rise from decisions that reduce firm value. Yet, rational bondholders understand that actions taken after issuance will tend to increase stockholder wealth, and they forecast the value effects of future decision when bonds are sold. In an efficient market, the bond price at issuance reflects an unbiased forecast of the effects of such future actions. Thus, on average, bondholders will not suffer losses, although the firm (and hence its stockholders) must bear the costs of nonoptimal decisions motivated by wealth transfers from debtholders. Therefore, effective control of this bondholder-stockholder conflict can increase firm value. Bond covenants that constrain activities such as asset sales or dividend payments are examples of voluntary contracts that can reduce the costs generated when stockholders of a levered firm follow a policy that deviates from maximization of the firm's value. The cost-reducing benefits of covenants accrue to the firm's owners through the higher price the bonds command when issued. Furthermore, if covenants lower the costs that bondholders incur in monitoring managers, these cost reductions also are passed to stockholders through higher bond prices at issuance. Therefore, in structuring an optimal debt contract, the firm's managers face a trade-off between increased proceeds from the debt issue and reduced flexibility with respect to future policy choices. The constraints imposed through covenants are frequently specified in terms of accounting numbers. Debt covenants that employ accounting numbers are conventionally divided into (1) affirmative covenants, which require firms to maintain specified levels of accounting-based ratios, and (2) negative covenants, which limit certain investment and financing activities unless specified accounting-based conditions are met. For example, negative covenants restrict the payment of dividends, the disposition of assets, the issuance of additional debt, and merger activity; affirmative covenants specify minimum working capital and net worth requirements. Although standard covenants in debt issues require that accounting numbers be consistent with generally accepted accounting principles (GAAP), they normally do not prohibit managers from switching between accepted methods. In some bank-loan agreements, firms are required only to provide unaudited, internally generated financial statements, but the contract also requires the firm to maintain substantially the same set of GAAP. If a change becomes necessary, the bank must be notified in writing prior to the change with the reasons detailed (see Zimmerman 1975). Since different accounting techniques imply different accounting numbers, firms have incentives to relax onerous constraints through the choice of accounting techniques. Academic accountants have devoted substantial effort to obtain empirical evidence on the importance of debt agreements in determining accounting policy. (Watts and Zimmerman [1986] and Christie [1990] provide reviews.) Initial studies adopted indirect methods to account for the effect of debt covenants on accounting decision by using the firm's debt-equity ratio as an explanatory variable in cross-sectional regressions. This ratio is a proxy for closeness to covenant constraints, as well as for the expected costs should a breach occur. Duke and Hunt (1990) and Press and Weintrop (1990) offer evidence to support the use of the debt-equity ratio as a proxy for the closeness to debt covenant constraints. Christie (1990) documents significant support for this debt hypothesis by aggregating cross-sectional studies of accounting choice, generally concluding that the larger the firm's debt-equity ratio, the more likely the firm's managers are to select accounting procedures that shift reported earnings to the current period from future periods. Researchers have generally interpreted support for this debt hypothesis as evidence that managers act opportunistically. However, Watts and Zimmerman (1990) question whether the documented association is misinterpreted by researchers. Rather than reflecting managerial opportunism, the evidence may reflect the association among firms' investment opportunity sets, financial policies, and their efficient set of accounting methods. Even in theory, it is difficult to distinguish between opportunism and contracting efficiency as determinants of accounting policy choice. Given positive contracting costs, there will be a positive efficient amount of opportunism. Distinguishing between opportunism and efficiency is difficult in empirical work also. For example, a significant relation between accounting policy choice and leverage could indicate that managers of firms with high leverage act opportunistically in selecting accounting techniques to reduce costs imposed by constraints in debt covenants. Alternately, it could indicate that corporations for which a particular set of accounting techniques is efficient also tend to be those firms for which high leverage is efficient. Firms examined in these cross-sectional studies are not necessarily close to their debt covenant constraints at the date examined. When firms are not close to debt covenant constraints, managerial opportunism is a less plausible explanation for the documented association between leverage and accounting choice. Yet, since it is costly for firms to switch back and forth between accounting procedures, firms that switch accounting methods to delay default are likely to continue to employ income-increasing accounting procedures, even if default is no longer likely (see Sweeney 1992). A firm's current accounting policies thus should depend on its historical choices and the time series of variables hypothesized to influence accounting policy. Therefore, cross-sectional studies do not provide the most direct or most powerful tests of the relation between accounting choice and debt contracts. Recent studies overcome a number of limitations inherent in cross-sectional analyses by examining accounting-based defaults in debt covenants. Careful examination of the default process, its causes and cures, provides evidence on important aspects of the lending process. In this article, I review this developing literature to provide a richer understanding of the costs of leverage. These costs have important implications. For accountants, they offer potential explanations of a firm's accounting policy choice; for financial economists, they enrich our understanding of the firm's optimal capital structure.
