Gary H. Stern - President, 1985-2009
Published September 1, 2008 | September 2008 issue
In our Bank's 2007 Annual Report, I expressed concern about the recent expansion of the safety net for large financial firms and, particularly, its potential to dull the market forces that would otherwise constrain excessive risk-taking. Although the Annual Report essay came out just a few months ago, the financial safety net has expanded since its release. The too big to fail (TBTF) problem has worsened yet again.
At the same time, however, there has been progress in beginning to develop a policy framework to address TBTF and to enhance market discipline. Policymakers have begun to focus more explicitly on minimizing the fallout, or “spillovers,” from a financial firm's impairment as they consider how to improve financial stability and to reduce the incentives that TBTF firms have to take on excessive risk.
Naturally, I view these latter developments quite positively. In our 2004 book on TBTF, Ron Feldman and I emphasized that “policymakers should give highest priority to reforms limiting the chance that one bank's failure will threaten the solvency of other banks.”1 We came to that conclusion with the following logic:
Policymakers extend support to weak but systemically important financial firms in order to contain spillovers;
Limiting spillovers from failures can reduce the principal rationale for extraordinary government support;
Creditor expectations of government support will diminish (and market discipline will increase) when policymakers have less reason to provide such support.
Recent comments from Secretary of the Treasury Henry Paulson echo this argument (and we have seen it elsewhere as well 2):
In an optimal system, market discipline effectively constrains risk because the regulatory structure is strong enough that a financial institution can fail without threatening the overall system. For market discipline to constrain risk effectively, financial institutions must be allowed to fail. Under optimal financial regulatory and financial system infrastructures, such a failure would not threaten the overall system.
However, today two concerns underpin expectations of regulatory intervention to prevent a failure. They are that an institution may be too interconnected to fail or too big to fail. We must take steps to reduce the perception that this is so—and that requires that we reduce the likelihood that it is so.3
Having agreement on a policy framework is a necessary but not sufficient base for reform. Government agencies charged with addressing instability and related TBTF concerns, and private sector groups and firms critical to that effort, require recommendations on how to implement the tenets of the framework. We have long had a list of reforms to address TBTF,4 but heretofore we have not prioritized those proposals. So, where would we have policymakers start? We would begin the effort to manage TBTF with an approach we call systemic focused supervision (SFS).
In general, SFS attempts to focus supervision and regulation efforts on spillover reduction, and it consists of three pillars: early identification, enhanced prompt corrective action (PCA) and stability-related communication. In particular, SFS uses the information-gathering and analytical skills of supervisors to better understand how one firm's impairment might spread to other firms or markets; it relies on the enforcement capabilities of regulation (combined with market information) to close firms before they incur losses that could bring down their peers; and it extends central bank communication techniques to financial-stability-related efforts.
This program builds on the strength and current direction of supervision and regulation to focus across firms and the interconnections in the banking and financial system as a whole, rather than concentrating on supervisory assessment of single firms. Combined, these efforts constitute important actions in a long-term effort to limit the spillovers from the failure or impairment of a systemically important financial institution. I'll now describe what I see as the basics of the three components.
Early identification.5 This is a process to identify and to respond, where appropriate, to the material exposures among large financial institutions and between these institutions and capital markets. By “material,” I mean sufficiently significant such that problems at one financial institution could substantially impair other institutions and/or normal market functions.
Early identification could take many forms. Supervisors might begin by examining the performance of a number of large financial institutions hypothetically subjected to a series of shocks. The shocks could include large losses to a given type of loan or security on the firm's balance sheet, or a significant drop in the availability of funding. The results of this simulation would provide policymakers with a sense of which stresses lead to significant problems at the firms. A second step is to determine how the material difficulties of one of these large institutions would affect the others. At a minimum, this would involve determining how much the failing institution owes the others at the end of the business day, what form the exposure takes, how much the exposure varies over time and so on.
The goals of the exercise I just described are (1) to give policymakers a sense of which events are not likely to severely impair the financial institution, thus permitting them to avoid providing support, and (2) to identify those exposures that might bring down the firm, and thus are deserving of closer policy scrutiny and, most importantly, an effective and timely response.
As part of this effort, supervisors should also consider how they will make assessments of spillover potential at the time a financial institution experiences serious difficulty. Supervisors must determine what type of information they will need in short order from financial institutions during a period of turmoil and what information they can actually get in short order, and then develop a plan to address whatever information gaps are identified. Closing these gaps means that policymakers can make informed judgments at the time of failure and, where possible, identify and resolve those issues that would otherwise lead to provision of extraordinary support.
Enhanced prompt corrective action.6 The Federal Deposit Insurance Corporation Improvement Act of 1991 implemented PCA. Like many so-called structured early intervention and resolution (SEIR) regimes, PCA works by requiring supervisors to take specified actions against a bank as its capital falls below specified levels. A bank whose capital declines below a given level, for example, could have its ability to pay dividends constrained. In the extreme, chartering authorities could shut down banks that have capital levels below a trigger and that cannot raise additional capital.
Closing banks while they still have positive capital, or at most a small loss, can reduce spillovers in a fairly direct way. If a bank's failure does not impose large losses, by definition it cannot directly threaten the viability of other depository institutions that have exposure to it. Thus, the PCA regime offers an important tool to manage systemic risk.
However, many observers, including some of the most zealous advocates of using a SEIR regime in the United States, view PCA as inadequate for at least two related reasons. First, capital measures for banks can reflect a “rearview mirror” or historical assessment of the bank's assets, particularly for valuations of bank loans. Such assessments may, at times, prove excessively generous.
Second, bank capital measures reflect judgments of firm management on the value of the bank assets and liabilities, albeit judgments heavily informed by a sophisticated body of accounting guidance. While management judgments on the valuation of assets and liabilities typically raise little concern, at times these judgments may overstate the capital of the firm. Under the capital measures reported by banks, therefore, a bank subject to supervisory oversight and which appears to have positive capital can actually have large losses upon failure.
Using PCA triggers based on more forward-looking measures of bank solvency outside the control of bank management could help address these concerns. Data generated directly from financial markets offer one source of forward-looking measures of a bank's condition;7 market participants do not always get their forecasts right, but they do appear to incorporate assessments of the future prospects of firms in their pricing decisions. Moreover, management of firms cannot readily control these measures. This suggests that an enhanced PCA regime relying on both book value capital and market measures of risk—such as subordinated debt spreads, prices of credit default swaps and/or equity values—would be an improvement over the current regime. In fact, the original proposals for SEIR in the United States used market measures of bank net worth to provoke supervisory action. In practice this means that some combination of market signals and accounting measures of insolvency could lead to the timely closure of a bank.
Just as management cannot directly control market measures of bank conditions, supervisors also must accept rather than control what the measures have to say. Use of market measures therefore could guard against supervisory forbearance that occurs when supervisors do not take appropriate remedial action against a financial institution as its condition worsens. Reported capital measures that do not reflect the true condition of the financial institution might facilitate forbearance. But market measures would not—even when political pressure, or the hope that a weak financial institution will resolve itself, supports forbearance.8
Communication. The first two pillars of SFS seek to increase market discipline by reducing the motivation policymakers have for protecting creditors. But creditors will not know about efforts to limit spillovers, and therefore will not change their expectations of support, absent explicit communication by policymakers about these efforts. What form might that communication take?
I have suggested that this communication should possess several attributes.9 First, it should be released routinely, like the semiannual Humphrey-Hawkins testimony, to facilitate the focus of interested parties. Second, it should disclose information on stability-related activity at an early stage, even if it is work-in-progress. Such a strategy would provide creditors with a richer sense of the changes under way. Finally, the communication should explicitly link the activity under way to the goal of reducing spillovers, thus raising the feasibility and prudence of putting creditors at greater risk of loss.
Professor John Taylor of Stanford University has also offered some valuable suggestions for stability-related communication, particularly communication following acts that seemingly expand the safety net.10 He recommends that regulatory agencies conduct and publish after-the-fact reviews that cover, in part, the counterfactual. (That is, in retrospect, what would have happened if safety net support had not been offered?) Among other things, such analysis would help agencies determine where they should focus their efforts to limit spillovers in the future.
Rather than discussing additional detail on what the communication might include, I would simply note that agencies in the United States have numerous examples from which to borrow. Many central banks and treasury departments around the world provide stability-related communication. Improving upon what we might take from other countries and adapting it for the United States strikes me as a low-cost initial step.
I see several advantages to SFS as policymakers consider longer-term efforts to respond to and learn from recent market turbulence. The supervisory role I envision begins with information collection and analysis, activities that I believe supervisors have performed well. As noted above, in my proposal supervisors would identify in advance the potential consequences of the failure of one bank for others. 11A defining feature of bank supervisors, after all, is the private information they collect on financial institutions. Moreover, supervisors have the relatively unique ability to look across multiple firms.
Supervisors have already recognized the substantial benefits they can provide by taking a cross-financial institution, spillover/stability-focused role. The Federal Reserve Bank of New York's effort to improve derivative processing and overall infrastructure provides a prominent example.12 The fact that SFS builds on growing supervisory practices provides another reason to begin with this reform. Likewise, our communication recommendation builds on efforts already under way.13
The disruptions that have plagued the financial and credit markets for over a year have led the Federal Reserve and Treasury to justifiably take extraordinary actions. These actions, however, potentially impose costs by expanding the scope of the federal safety net, so we must determine a long-run plan to address these costs and the more substantial TBTF problem they create. We offered a framework to address this problem in 2004 and believe it continues to hold promise. In this essay, I have suggested that policymakers now look to systemic focused supervision as an important step in limiting spillovers and in changing the expectations of creditors in a constructive way.
1 See Gary H. Stern and Ron J. Feldman's 2004 book, Too Big To Fail:
The Hazards of Bank Bailouts (Washington, D.C.: Brookings Institution
Press, p. 147).
2 Most recently, see Federal Reserve Chairman Ben Bernanke's speech, “Reducing Systemic Risk,” at the Federal Reserve Bank of Kansas City's Annual Economic Symposium, Jackson Hole, Wyo., Aug. 22, 2008.
4 To be sure, other policymakers have already given serious thought to considering how government and the private sector should respond to recent credit and financial market turbulence. Domestically, the President's Working Group (PWG) has issued a list of high priority reforms as has the Financial Stability Forum (FSF), and work has begun on implementing some of the recommendations.
5 I base this discussion on pages 112-116 of Too Big To Fail.
6 I base this discussion on pages 124-128 of Too Big To Fail.
7 The Federal Reserve Bank of Minneapolis has also been a long-time supporter of increasing the use of market data in the supervisory process. For example, see the September 2001 Region with a special discussion of “Using Market Data in the Supervisory Process.”
8 Incorporating market signals into PCA certainly has its downsides. Market signals can reflect factors not directly related to the health of a financial institution, for example, although I think supervisors have tools to at least partially mitigate these imperfections (e.g., extracting the credit risk component from market signals or averaging across market signals or over time for a given market signal to filter out “noise” in the data). Moreover, like other reforms that increase market discipline, enhanced PCA may force firms into resolution, and potentially into the sale of troubled assets, precisely when the financial system seems most vulnerable. On balance, however, allowing financial firms to build up losses before they fail does not seem conducive to maintaining financial stability.
9 See Gary H. Stern's column, “Addressing the Trade-offs: Market Discipline, Stability and Communication” in the December 2007 Region.
10 See John B. Taylor's presentation, “Toward a New Framework for Exceptional Access,” at the policy workshop, “The Future Role of Central Banking Policy: Urgent and Precedent-Setting Next Steps,” July 22, 2008.
11 Such analytical tasks, while challenging, seem at least as tractable as other charges supervisors carry out, such as determining the optimal levels of minimum capital or liquidity.