Editor's note: These remarks were delivered to the Bank Administration Institute Treasury Management Conference in New York, N.Y., Sept. 27, 2004.
My thanks to the BAI for the invitation to speak at the annual Treasury Management Conference here in New York City. There are many experts here, both on the program and in the room, on the functional details of risk management and, as an economist, I recognize the benefits of specialization and comparative advantage. Thus, I will not review more technical topics that others will cover. Indeed, I might enjoy having the intricacies of Basel II explained to me again since it is possible that the 100 or so briefings I have received to date have not completely stuck.
In my position, I have a responsibility to think broadly about public policies associated with bank risk taking and risk management and, in that vein, I have long been interested in the distribution of risk bearing in banking, broadly defined to include the multitude of activities now carried out under the auspices of a financial holding company. In my view, the public should not bear "too much" of the risk from banking activities and positions. I am worried, particularly, about the shifting of risk which occurs with regard to the nation's largest and in some ways most important banks and will focus my discussion on that topic. At least some of you likely will consider a talk on this issue overly academic and only remotely related to practical concerns. But the worries of central bankers and other policymakers can affect you; indeed, policy responses to limit risk shifting could be as important as the next revolution in finance or funding.
While I have hinted at my views—and these are only mine, I'm not speaking for others in the Federal Reserve—let me summarize them a bit more formally: Risk taking in banking is unusual in the U.S. market economy in that it is assumed by both public and private parties. Despite efforts at reform, I am concerned that banking policy continues to shift too much of bank risk taking to the public. In particular, not enough has been done to limit the risk shifting which results from the so-called too-big-to-fail (TBTF) status of the nation's largest and most important banks. This risk shifting is of concern because it is costly; it wastes society's resources. As a result, we have seen and will continue to witness regulatory and legal efforts to push risk taking back onto banking creditors.
Risk shifting to the public
Let me get into some details. The public underwrites or assumes some of the cost of bank risk taking in three well-known ways which make up the traditional banking safety net: They are deposit insurance, Federal Reserve discount window lending and payment system finality. Since these policies and their effects are well recognized, I will not dwell on them.
There is, however, an open question as to whether the traditional bank safety net transfers too much risk to the public. Whatever answer I might give to that question, it is true that these mechanisms have received considerable review over the years. The discount window, for example, was recently the subject of major modification, deposit insurance reform has been an FDIC priority for years, and the Federal Reserve has given payment system policies fairly regular tuneups. In contrast, TBTF has received relatively little attention, beyond what amounts to public "shoulder shrugging."
In this context, I am using TBTF to refer to the class of institutions whose nominally uninsured creditors believe they have a reasonable chance of transferring some of their loss to the public should their bank fail. That is, these creditors do not expect to bear the full magnitude of their potential loss. In the extreme, they may expect to bear virtually no loss, with the public picking up the full cost. Typically, TBTF applies to the largest institutions in the country, but there may also be cases where banks that specialize in certain types of activities, say, payment processing in certain financial markets, have TBTF status.
In the United States, the classic realized case of TBTF is of course Continental Illinois. This event occurred 20 years ago, but it illustrates the concerns well. Overseas, we have had a significant number of large banks whose uninsured creditors, and indeed equity holders, have received bailouts and where creditors apparently expect such bailouts to occur in the future.
Why risk shifting is costly
I have expressed concern about the transfer of risk but have not yet explained the basis for my view. When documenting concerns about TBTF, many observers point to the bill that taxpayers might face if public funds are used to provide direct bailouts to uninsured creditors of large banks. But these bailouts are not costs in an economic sense; rather they are transfers from one group (taxpayers) to another (uninsured creditors). Instead, the real cost of TBTF results from distortion of the risk/reward trade-off and related risk management decisions. In this sense, TBTF status may already be an integral part of the financial environment.
At its heart, risk management is the balancing of risks and rewards. For risk management to work effectively, risk and reward must be measured and priced correctly. But the shifting of risk to the public virtually guarantees that the cost of risk taking will be understated from the point of view of bank managers, and therefore banks will take on too much risk.
Let's look more closely at the mispricing of risk by bank creditors. Uninsured creditors at TBTF banks by definition do not expect to bear the full loss if their bank fails. Given this perception, we should not expect such creditors to become fully informed as to the risk taking of their bank nor should we expect their pricing and quantity decisions to reflect that risk taking. Instead, we would expect them to price their funds too low and provide too much funding relative to what they would do if they were fully at risk of loss.
Now let's turn to the bank response to this mispricing. Banks respond to unduly low prices and overly generous funding by taking on more risk than they otherwise would. After all, the reward for risk taking is inflated because the cost of it has been artificially reduced. I must emphasize that none of this is to say that bankers are acting illegally, or inappropriately, for that matter. They are instead behaving like all market participants, responding to the signals that the market presents.
It is difficult to determine precisely what banks do in the face of mispriced funds that they would otherwise not have done. They may, for example, make too many loans to telecommunication firms, bet excessively on mortgages or expand their branch structure ill-advisedly. But if we assume that bankers are rational and respond to the price and quantity decisions of creditors, we must expect that excessive risk is the result.
Poor decisions by banks mean that society's resources are used in a suboptimal manner. Funds flow to projects that, in retrospect, should not have been funded to the degree they were. And that is the real cost of TBTF—significant resource misallocation for society as a whole.
A fair question is, "What is the dollar figure associated with this resource misallocation?" Or, "What is the lost output associated with TBTF?" To be honest, quantification in this area is extremely difficult and at an early stage, but let me briefly review some evidence suggesting that the problem is real and large. For those interested, this evidence is described in more detail in a recent book I co-authored with Ron Feldman on the topic called Too Big to Fail: The Hazards of Bank Bailouts, published this year by Brookings.
First, a number of models and observations suggest that policies which fall under the TBTF umbrella (e.g., blanket guarantees for large bank creditors) have been associated with banking crises, in both developed and developing countries. In these models, banks considered TBTF provide excessive credit and take on too much risk. As their solvency eventually comes into question, creditors begin to doubt that governments will make good on their implicit support and act on their concern by, for example, pulling funds from the bank/country. The outcome of this chain of events can be a banking/financial/foreign exchange crisis. These banking crises are associated with lost output estimated at around 15 percent of gross domestic product.
Second, a number of event studies suggest that uninsured creditors expect some support if their large bank gets in trouble. These banks control the vast majority of banking assets in virtually all developed countries. Thus, the distortion stemming from poor asset allocations by these banks could be quite large—this is essentially an argument relying upon the size and scale of the banks involved.
Third, credit rating agencies leave little doubt about their view of the TBTF issue. Here is Moody's take on it: "No major bank in any developed country has ever been allowed to default upon any rated or unrated deposit or debt obligation. ... Consequently, as a general rule, the historical credit risk associated with OECD banks approaches zero. ... The reason, of course, is official support popularly known as 'Too Big to Fail.'"
Because banking assets have become more concentrated and large banks have become more complex, I think that more creditors expect some form of
TBTF protection and that creditors have more strongly held beliefs (than formerly) that they will be protected. Underlying this statement is a view as to why policymakers are at times willing to shift risk from bank creditors to the public.
As you may know, the TBTF problem was supposed to be addressed through the FDICIA legislation passed in 1991. Without going into the details, I believe this legislation has not fixed the risk-shifting problem as just described because the legislation does not address the underlying explanation for the bailout of TBTF creditors. If the underlying concern is not addressed effectively, the rationale for TBTF bailouts won't be eliminated.
So why would policymakers potentially support public bailouts of the creditors of TBTF banks? In short, policymakers fear that the failure of one large bank will "spill over" to another bank, to financial markets more generally and potentially to the real economy with adverse consequences for employment and growth. Because creditors know that policymakers have this concern, they should, and do, rationally expect some protection from the government.
As suggested a moment ago, I think that policymakers' fear of spillovers has increased for several reasons:
- Banks that were already large are now even larger.
- More banks are of a size that, should they encounter difficulties, there could be spillovers of consequence.
- Some important activities, such as capital market transactions processing, are highly concentrated, making the banks active in these markets difficult to replace.
- Banking has become more complex on many levels, including: the convoluted legal structure of some financial holding companies, the sophisticated trading/hedging strategies some banks carry out and the massive, overall breadth of their operations. And I have not touched on the international nature of activities, which potentially complicates regulatory and legal issues and concerns. The significance of these complexities is that they make it more difficult for policymakers to determine both if spillovers might result from a bank's failure and how severe those spillovers may be. And when in doubt, policymakers have incentives to intervene and provide bailouts rather than to stand back and risk contagion.
The bottom line is that to reduce the expectations of TBTF coverage, policymakers have to become more knowledgeable about the potential for spillovers. They also have to become more comfortable that creditors can absorb at least some loss without spillovers of consequence. If reforms do not address concerns over spillovers, they will not forestall TBTF bailouts.
The shape of reform
I may be more outspoken on this subject than many colleagues, but I think the situation just described makes TBTF an important policy issue. Further, I think it noteworthy that significant questions have been raised about the size and scope of GSE [government-sponsored entity] risk taking, with explicit concern raised about their TBTF status. At a minimum, recent events suggest a general policy interest in reducing TBTF expectations. In fact, I would go further; I see a good chance that future efforts at reform will begin to focus on TBTF issues, recognizing that the presence or absence of large bank troubles will go a long way toward shaping the debate.
There are three general types of reform that policy could take, although in the end these reforms are complementary. Policy could try to reduce the uncertainty associated with potential spillovers from large bank failures, to limit the losses associated with large bank failures and/or to limit the exposures banks have to each other via payment systems. Let me provide some examples to clarify these reforms.
To reduce uncertainty, supervisors could undertake more detailed planning for and simulation of large bank failures, focusing on resolution techniques that involve losses to uninsured creditors. These exercises might help to determine both how large bank solvency holds up under certain stresses and the level of exposure that banks have to each other. I would suggest that consideration be given to publicizing the general results of such exercises, excluding specific names, of course. Further in this vein, the legal status of creditors and specific agreements, such as those calling for closeout netting, could be solidified.
The ultimate barrier to risk shifting to the public is bank capital. To the degree that Basel II reforms better align risk and capital, they should help limit the shifting of risk. Another approach to reducing losses is to improve prompt corrective action (PCA)—that is, to close failing banks while they still have positive capital. Currently, PCA is not likely to be wholly effective because it relies on book capital measures and thus leads to something other than prompt closure, although it may be an improvement over prior practice. We ought to consider, however, moving from book value measures of capital to alternatives that better capture the real buffer that the public has against loss.
In the payments area, I think there has been some success. More wholesale payments occur in a real-time settlement framework, and collateral is more commonly provided. That said, there are still payments and securities transactions that do not settle in real time and where additional collateral could be provided. Also, some reforms, in areas such as foreign exchange, for example, are still relatively new and need continued vigilance.
In total, what would these reforms and others we might think of produce? With these reforms, policymakers should come to feel more comfortable in putting uninsured creditors at risk of loss. And there are a number of ways that policymakers could communicate that evolution. In turn, with new knowledge of policymakers' attitudes and plans, uninsured creditors would start to better price bank risk taking, and banks would reduce the amount of excessive risk taking that occurs. With less risk taking, the chance of failure would diminish, which could make the necessity for TBTF coverage relatively remote. A virtuous cycle could ensue.
Over the last several months, I have had the opportunity to discuss concerns associated with TBTF banks with a number of audiences, including some involved in fashioning legislation. In these discussions, there has been at least some sentiment for prohibiting mergers of large banks in order to limit the potential risk transfer to the public.
This sentiment could become reality in a number of ways, ranging from reducing the current 10 percent cap on deposits that a bank can hold nationally to including TBTF concerns in the antitrust review conducted by bank regulators or by other entities, such as the Department of Justice, when a merger is proposed. I raise these possibilities not to endorse them but rather to give a sense of where this debate could head if other reforms, more targeted and potentially more constructive, do not advance.
I began this talk by arguing that reforms to address risk shifting to the public will have practical consequences. The implications of what I have said are probably clear, but it is best to be explicit about potentially major changes.
Perhaps most importantly, if creditors are put at greater risk of loss, some banks will at some point in the future have to pay more—perhaps considerably more—to raise funds than under the current regime. If and when this happens, some institutions will become less profitable than currently, all else equal, reflecting both higher costs and a lower level of risk taking. Bank management and shareholders conceivably might be worse off, but the public would be better off.
As the situation evolves, banks may have reason to enhance what is already diligent work to line up potential backup funding. Indeed, entire liquidity management processes may need improvement, particularly if less intraday exposure in payments results from reform. Higher costs may also result if reforms lead to increased capital or to increased collateral associated with certain transactions. Supervisors may also focus more of their attention on the potential for and prevention of spillovers in the wake of a problem at a major bank, and they may request assistance with such analysis.
Finally, once uninsured creditors are put at greater risk of loss, policymakers may be more enthusiastic about using market data to help discipline banks. Subordinated debt issuance requirements are one possibility that has been identified previously. Policymakers may also make more use of market-based measures of solvency and market valuations in general. Overall, the extent of the changes could, at the end of the day, be far-reaching.