Monitoring and Controlling Bank Risk: Does Risky Debt Serve any Purpose?

We examine whether mandating banks to issue subordinated debt would serve to enhance market monitoring and control risk taking. To evaluate whether subordinated debt enhances risk monitoring, we extract the credit-spread curve for each banking firm in our sample and examine whether changes in credit spreads reflect changes in bank risk variables, after controlling for changes in market and liquidity variables. We find that they do not. Our result is robust to firm type, examination rating, size, leverage and profitability, as well as to different model specifications. To evaluate whether subordinated debt controls risk taking, we examine whether issuing subordinated debt changes the risk-taking behavior of a bank. We find that it does not. We conclude that a mandatory subordinated debt requirement for banks is unlikely to provide the purported benefits of enhancing risk monitoring or controlling risk-taking. (Bank Risk Changes; Estimation of Credit-Spread Curves) Since the mid 1980s economists have debated the merits of regulations that would require banks to issue a minimum level of subordinated notes and debentures (SND). Proponents of this regulation suggest that SND will enhance market discipline and curb excessive risk taking in two ways: through risk monitoring and through preventative influence. The benefits of monitoring are realized if investors accurately understand changes in a firm’s risk condition and incorporate their assessment promptly into the prices of risky debt issued by the firm. If they do, then changes in credit spreads will provide useful information to regulators and assist in supervision. This is referred to as indirect market discipline. The direct effect results from the increased costs of funding that result from investors being able to accurately reprice debt, should the bank adopt riskier strategies. Due to market risk monitoring, banks with SND may be less likely to adopt risky strategies in the first place. This is the preventative influence role of SND. The purpose of this paper is to examine whether SND issued by banks and bank holding companies (BHCs) (together referred to as banking firms) enhance risk monitoring and/or have preventative influence. To evaluate risk monitoring, we examine whether changes in firm-specific risks get reflected in changes in credit spreads of SND issued by banking firms, after controlling for economy wide factors and liquidity factors. To evaluate preventative influence, we examine changes in risk characteristics of banks and BHCs after they first issued SND. We focus on banking firms because policymakers are actively considering the use of subordinated debt as a regulatory tool. A consultative paper issued by the Basel Committee on Banking Supervision (1999) proposes new risk-based capital standards (Basel II) that seeks to improve the incentive effects of capital regulation through increased granularity in risk measurement, improved supervision, and increased market discipline. Mandatory subordinated debt requirement appears to be the cornerstone of Basel II’s market discipline provisions. In addition, the U.S Shadow Regulatory Committee has come out strongly in favor of mandatory SND as a mechanism for realizing enhanced market discipline of banks. Finally, the Gramm-Leach-Bliley Act of 1999 mandated a joint Federal Reserve and U. S. Treasury study of bank subordinated debt requirements. This legislation also requires all large banking firms to have at least one issue of subordinated debt outstanding at all times. A lengthy literature exists that addresses the question of whether market prices of liabilities respond to individual bank risk taking. To date the results of empirical studies have been mixed. Studies done prior to 1992 fail to find a significant relationship between firm risk and yields on subordinated debt.1 More recent studies, however, do indicate that risk is being appropriately priced. For example, Flannery and Sorescu (1996) find that for banks over the 1983-1991 Examples include Avery, Belton and Goldberg (1988), Gorton and Santomero (1990), who find no effects, Cramer and Rogowski (1985) and Goldberg, Jooyd-Davies (1985) who obtain mixed conclusions, and Baer and Brewer (1986), and Hannan and Hanweck (1988). For excellent reviews of this literature see Flannery (1998) and Bliss (2000).