Published September 1, 2001 | September 2001 issue
Federal Reserve policymakers have primarily focused on subordinated notes and debentures (SND) when discussing potential sources of market data. SND are fixed-income bank liabilities that are uninsured and unsecured. Rather than examining the yield on SND directly, analysts track the difference between the SND yield and yield on a security of a comparable maturity that has very little chance of default, such as a U.S. Treasury bond. Focusing on this "spread" should help highlight that part of the SND yield reflecting the chance that the issuer will default.
Supporters offer two reasons to focus on SND as a source of market data. First, the incentives facing SND holders are believed to be closely aligned with regulatory objectives. A recent Federal Reserve publication explained the point as follows:
Subordinated debt holders have an interest in discouraging excessive risk taking because their claims are both long-term and junior to all depositors and to any senior debt holders. Subordinated debt holders share in very limited ways in potential gains made by a company but are exposed to considerable risk if it encounters financial difficulty. In this respect, their risk preferences can resemble those of bank supervisors.1
SND supporters usually contrast the incentives of SND holders with those of equity holders who can benefit from the upside potential of bank risk taking (equity holders also have long-term, junior claims). Seeming ease of calculation and interpretation offers a second reason to focus on SND spreads.
In truth, use of SND signals faces many complications because of the limited number of issuers, the diversity of SND issues by banking institutions and the availability of reliable real-time data for supervisors. The choice of the risk-free benchmark rate has also become more problematic due to changes in the market for U.S. Treasury securities. These factors make it difficult to isolate that portion of the spread that reflects the riskiness of the institution.
Concerns about SND data led analysts to re-examine the ability of supervisors to harness equity data. Equity market instruments are issued by more firms. In addition, we have more confidence in the prices of equity securities because they are traded in more liquid and transparent markets compared to SND. But concerns about equity holders' ability to benefit from risk taking remained. Moreover, raw stock prices do not facilitate cross-institution comparisons nor do economists look to stock prices as direct measures of risk. So-called Merton models (named after the Nobel Prize-winning economist who first devised them) offer one method for addressing these concerns in a theoretically sound manner. Merton recognized that information embedded in equity prices, as well as in the movement of those prices, can allow valuation of the assets of the firm and the riskiness of those assets. More precisely, he observed that equity holders can be viewed as owning a call option that allows them to purchase the assets of the firm from other creditors. The models then use a very standard formula to value the option and determine the asset value and the volatility of assets. By comparing the value of assets to the value of liabilities and figuring out the chance that assets will fall below liabilities based on their volatility, analysts can determine the probability of a firm failing.
While computationally complex, this approach has become fairly standard among financial economists and policy analysts. Indeed, Federal Reserve economists have cited the output from such models to argue that deposit insurance provides a subsidy to banks.2 Certainly, questions about model assumptions and equity holder incentives remain. But, similar questions have also been raised about SND, particularly the degree to which supervisors and SND holders view the world similarly. After all, for the proper compensation an SND holder might be willing to finance bank risk taking that a supervisor would oppose.
Uninsured deposits offer another potential source of market data.3 These deposits take the form of certificates of deposit (CDs) with a principal of over $100,000. The basic research on jumbo CDs shows that their spreads also vary with the riskiness of the bank. Jumbo CDs of the vast majority of banks never trade in secondary markets, although the biggest banks issue jumbo CDs in very large amounts that do trade regularly. Rates on jumbo CDs of either type might be valuable market-based information, but the Federal Reserve does not have routine access to such data at the individual bank level.
Senior bonds also are a potential source of price signals. Prices paid for senior debt should reflect market perceptions of institutional risk taking. But use of senior debt prices would present a trade-off. Reviewing these prices would significantly expand the amount of bond market data available to supervisors. On the other hand, senior bond holders may not be as sensitive to risk as SND holders, due to their more protected position; this could reduce the information value of senior debt.
Supervisors could also review nonprice market data. Data on shifts in the composition of a bank's funding may be valuable. Banks appear to decrease their use of uninsured deposits and increase their use of insured funding as they approach failure. Likewise, the decision of a bank to abstain from issuing SND appears to provide information on its riskiness. In the equity markets, share purchases and sales by insiders may be informative. As is the case with jumbo CDs, these types of potential signals have received little attention to date, primarily because the Federal Reserve has not developed the apparatus to collect and distribute such nonprice data for use in banking supervision.
1 Lisa M. DeFerrari and David E. Palmer, "Supervision of Large Complex Banking Organizations," Federal Reserve Bulletin, February 2001, pp. 46-57.
2 For example, see Myron L. Kwast and S. Wayne Passmore, "The Subsidy Provided by the Federal Safety Net: Theory, Measurement, and Containment," Board of Governors Finance and Economics Discussion Paper Series, December 1997/58, p. 5, footnote 7.
3 Although not discussed at the conference, the overnight borrowing market could potentially offer yet another source of market data. See Craig H. Furfine, "Banks as Monitors of Other Banks: Evidence from the Overnight Federal Funds Market," The Journal of Business, January 2001, pp. 33-57.