Notes from the Vault
Larry D. Wall
A critical element of a credible policy for resolving large financial firms is a timely decision to put the firm into resolution.1 As I explained in an earlier Notes from the Vault, the chances of a successful resolution are substantially better if resolution starts while the firm still has positive capital. However, that post also indicated that the structure of the existing mechanisms, which rely on accounting values and supervisory judgment, tend to work in favor of postponing resolution.
As an alternative to relying on accounting data and supervisory judgment, many economists have proposed the use of market data such as stock or bond prices to trigger bank resolution or recapitalization.2 For example, banks could be forced to recapitalize or be resolved if their capital ratio as measured in market value terms dropped below some threshold or their bond yields rose above another threshold. However, a variety of potential problems have been raised regarding the use of market data. Whether these problems with market data would be more severe than problems with existing triggering mechanisms is unclear.
This Notes from the Vault will discuss ways in which we can gain experience with market-based measures as triggers while mitigating potential problems with these measures.
The case for market-based triggers
A fundamental question is, why can't supervisors use the same resolution trigger that is applied to judge nonfinancial firm insolvency, the inability to meet obligations as they come due? This trigger may not be perfect, but it works well enough for the rest of the economy.3
The problem with relying on the inability to honor obligations for banks is that this occurs most frequently because depositors start withdrawing their money—in other words, a "bank run" occurs. Although some academics have accepted bank runs as an essential mechanism for closing insolvent banks, a common concern is that such runs can be "irrational" and lead to widespread runs, often called "panics."4 Such panics not only put banks under stress but can also impair the ability of the entire banking system to provide services that are essential to the operation of the real economy. Concerns about such runs and panics have caused most developed countries to adopt deposit insurance to discourage runs and create lender-of-last-resort facilities to provide liquidity to banks experiencing a run.
Yet if we are unwilling to rely on the inability to meet obligations as a trigger for the resolution of failing banks, we must create some other mechanism.5 In practice, that other mechanism has typically relied on the bank supervisors. For example, the Federal Deposit Insurance Corporation (FDIC) is authorized to force resolution on any one of a number of grounds (See FDIC, pages 213–214). The problem with relying on the supervisors, however, is that they may choose to forbear from forcing insolvent banks into resolution. For example, in the 1980s the thrift supervisors devised a variety of mechanisms to allow market value insolvent savings and loans to continue operating, and bank supervisors allowed the large money center banks to defer recognizing losses on their loans to Latin American countries.
In response to concerns about supervisory forbearance, Section 131 of the Federal Deposit Insurance Corporation Improvement Act of 1991 mandates prompt corrective action (PCA). PCA requires the supervisors to impose progressively stricter supervision as a bank's capital declines, with resolution made almost mandatory when a bank's tangible equity capital becomes less than 2 percent of total assets.
The problem with PCA as implemented is that it relies on accounting measures of capital and bank managers have substantial discretion over the measurement of accounting capital. In particular, as a bank becomes financially distressed, its managers may become very reluctant to fully recognize the losses in their portfolio. This reluctance is evidenced by the large gap between bank accounting and market values in the 2007–09 crisis, as has been pointed out by both the Federal Reserve Bank of San Francisco's Fred Furlong, a group vice president in the economic research department, and Bank of England Executive Director Andrew G. Haldane.
As a result, the effectiveness of PCA depends to a very large degree on the willingness of bank supervisors to force bank managers to recognize losses in their banks' financial statements. In other words, although PCA was intended to restrict supervisors' ability to forbear, ultimately PCA as implemented would be effective only to the extent supervisors did not forbear in forcing banks to recognize their losses.
In contrast to the bank supervisors, market participants with their own funds at stake have little incentive to exercise forbearance. A trigger based on information obtained from financial markets may provide a mechanism for enforcing timely resolution of large failing banks.6
Concerns about market-based resolution triggers
Although market triggers are less susceptible to forbearance, the use of market-based triggers—such as equity market capitalization falling below some threshold, for supervisory action, including bank resolution—has drawn a number of criticisms. One of the concerns raised about market-based triggers is that the supervisors know more about a bank's condition than the market. Bank supervisors are inside the largest banks and can obtain detailed information about the banks' portfolios not ordinarily available to market participants.
This is a strong argument against relying solely on market-based triggers to force resolution. If confidential information available to the supervisors indicates that a bank is critically undercapitalized, the supervisors should be able to force the bank into resolution without waiting for the market. The market likely will eventually learn what the supervisors know, but bank management has a strong incentive to delay disclosing the adverse confidential information for as long as possible.
However, the argument is less compelling when the market has, instead, mistakenly identified a bank as failing. The late Professor Stephen D. Smith and I argued that if market participants incorrectly believe a bank is approaching insolvency, the bank's managers will have a strong incentive to release sufficient information about its portfolio to correct that misunderstanding. Thus, we should expect that supervisors' information advantage over financial market participants is temporary. If market participants continued to believe the bank was becoming insolvent after the bank has released more information, that would send a strong signal the bank is indeed failing.
Thus, we should expect that supervisors' information advantage over financial market participants is temporary. If market participants continued to believe the bank was becoming insolvent after the bank has released more information, that would send a strong signal the bank is indeed failing.
Another argument against relying on market prices is that they are noisy indicators of a bank's true condition. This is a potentially more serious problem because market prices incorporate more than investors' views about the value and riskiness of a bank's portfolio. For example, a bank's stock or bonds may incorporate a time-varying liquidity risk premium, which is a discount demanded by the investor to reflect the difficulty a potential buyer may have in subsequently selling the obligation if that became necessary. A sudden exogenous jump in the liquidity premium on a bank's stock (or bonds, depending on the trigger) could result in a nontrivial drop in its market price even if the bank's financial condition had not significantly deteriorated.
A second concern is that the value of some assets could be subject to manipulation. For example, someone may try to force a bank's stock price down by engaging in short sales in an attempt to force the price below market value and cause the bank to go into resolution. A third concern is that of multiple equilibria. In this case, the possibility exists that the market price could take on two different equilibrium values. The first equilibrium is that market participants believe the bank will survive, in this case the stock price remains above the threshold and the bank will not be put into resolution. The other equilibrium occurs when investors do not believe the bank will survive. In this equilibrium, the low stock price triggers resolution. These concerns about market manipulation are discussed in the context of using contingent capital bonds to recapitalize the bank by Federal Reserve Bank of Cleveland economist Edward Simpson Prescott (2012).
Thus, there are several potential problems with market measures so that the use of one of these measures could result in premature triggering of resolution. On the other hand, we cannot say how significant these would be in practice as market risk measures have not been used to trigger supervisory action.
We would be in a dilemma if we had to rely exclusively on the supervisors or exclusively on the market to trigger resolution. We have considerable experience suggesting that bank resolution is likely to be delayed if we rely primarily on the supervisor to trigger resolution. However, there are plausible arguments suggesting that reliance on market mechanisms may result in a healthy bank being forced into resolution.
But we are not bound to rely exclusively on one or the other as a trigger, we could use both. The market-based trigger could be set so that it would not be triggered so long as the supervisors enforced timely resolution. In effect, the market-based resolution trigger would ordinarily be a fail-safe mechanism, activated only after the supervisors had clearly engaged in forbearance.
One way to make the market trigger a fail-safe mechanism would be to set a pricing trigger at a point where the bank should already have been put into resolution. For example, Federal Reserve Bank of Chicago economist Douglas D. Evanoff and I suggested that subordinated bonds be used as an automatic trigger for supervisory intervention only after the bond yields had increased to junk bond levels. This would allow bond-specific liquidity premiums to rise substantially without necessarily triggering resolution.
A second way to make the trigger a fail-safe mechanism would be to require a persistent signal from the market before triggering resolution. For example, the trigger could be based on the weighted average price over some interval as recommended by Andrew Haldane. The requirement that a signal be persistent would likely create problems for those trying to manipulate market prices to force a good bank into resolution.
Arguably, these fail-safe requirements would not be sufficient to deal with the problem of multiple price equilibria. One way of dealing with this concern would be to make the market trigger one in which the supervisors could either resolve the bank or explain why they thought the bank was still viable.
The weakness of a resolve or explain regime is that it might allow for some forbearance on the supervisors' part. The difference from the existing regime is that the existence of supervisory forbearance would be transparent to the political system and not hidden behind the bank's misleading financial statements. This transparency could be increased by establishing a requirement that the Government Accountability Office (GAO) review all cases where the supervisors did not resolve a bank after that bank had tripped the market-based trigger.
A variety of proposals have been made to limit supervisory forbearance through the use of market-based triggering mechanisms. However, the market-based mechanisms have been shown to be subject to several theoretical problems. Whether and how significant the problems would be in practice is as yet untested.
This post proposes the use of market mechanisms as fail-safe triggers for supervisory action rather than relying on them as the primary trigger. Structured as fail-safe mechanisms, market triggers would force action only after the supervisors should have acted.
One could correctly argue that using market prices as fail-safe mechanisms does not guarantee an end to supervisory forbearance. Moreover, it is reasonable to believe that such a use of market mechanisms will nevertheless reduce forbearance in the short run and possibly pave the way to greater reliance on the market in the long run.
The short-run impact of adding market mechanisms would be to change the political setting for the supervisors. Supervisors who choose to engage in forbearance would be taking the risk that the market mechanism will later signal that the bank should be put into resolution. If such a signal is observed, the supervisors may then put the bank into resolution but face questioning as to why they engaged in forbearance. Alternatively, under a "comply or explain" approach, the supervisors may continue to forbear. However, in this case they would face public scrutiny in the form of a GAO investigation and likely a follow-up congressional hearing.7
Another short-run impact may be to alter bank management's behavior. Under the current regime that relies on the supervisors' triggering action, the chief executive officer of a bank facing financial problems may be reluctant to cut dividends or issue new stock, fearing that doing so would anger his shareholders. However, under a regime with fail-safe market triggers that makes forbearance less likely, a CEO who delayed raising capital would run a greater risk that his bank would be put into resolution, not only costing him his job but also all of his holdings of his bank's stock and likely most of the value in his bank pension plan.
Finally, in the longer run we would get a chance to observe what problems, if any, market-based mechanisms would have in practice. Given that information, we can then adopt the mechanism as the primary resolution trigger, modify the mechanism to address observed problems, or retain some version of the market trigger only as a fail-safe device.
Larry D. Wall is the executive director of the Center for Financial Innovation and Stability at the Atlanta Fed. The author thanks Paula Tkac for helpful comments on the paper. The views expressed here are the author's and not necessarily those of the Federal Reserve Bank of Atlanta or the Federal Reserve System. If you wish to comment on this post, please e-mail firstname.lastname@example.org.
1 Some have argued that a bank that is too big to fail is simply too big, with the implication that the problem of resolving large financial banks could be eliminated if we broke up the largest ones. Although there may be scope for breaking up the largest banks, there are reasons to be skeptical that breaking up the banks is the solution to too big to fail. The problem is that the largest institutional customers need banks that have the liquidity to make payments of hundreds of millions or even billions of dollars. Many of these customers also need the ability to obtain very large loans on short notice. Any bank that is large enough to serve these institutional customers is likely large enough to be considered "systemic" in at least some circumstances.
2 Examples of such a proposal include my 1989 proposal for puttable subordinated debt, and Federal Reserve Bank of Chicago economist Douglas D. Evanoff's and my proposal for using subordinated debt spreads. Another proposal to use a market-based trigger comes from University of Florida Professor Mark J. Flannery's recommendation that equity prices be used to trigger the conversion of contingent capital bonds.
3 One can view such a trigger as a crude measure of whether a firm has sufficient positive net value. If the firm has sufficient value, it can obtain a loan to meet maturing obligations if necessary. If no one is willing to lend to the firm, that sends a fairly strong signal that the firm is perceived to have insufficient equity to be viable given its industry and management.
5 The alternative would be to allow insolvent banks to continue operation with potentially very large losses to the taxpayers and the social cost of allowing inefficient managers to continue controlling the failed bank.
6 Market-based triggers would not be as valuable for small banks, nor would they be feasible for most small banks. Market triggers would be less valuable for smaller banks, as the losses from forbearance are smaller. Moreover, market triggers would not be feasible for most small banks as these banks do not have liquid markets for the stocks or bonds.
7 A possible reason why the supervisors may prefer forbearance would be because they are not confident in their tools and procedures for resolving a large bank. Such a lack of confidence may be justified as I argued in "Too Big to Fail: No Simple Solutions." In this case, the existence of a fail-safe market mechanism may be helpful in motivating policy initiatives to correct that lack of confidence. That is, the threat of having to publicly show forbearance at some point in the future may supply additional motivation for the supervisors to work with Congress to correct the weaknesses in their resolution tools and procedures.