I am very pleased to have this opportunity to appear before this committee to discuss one of the most important issues currently facing our country. The financial crisis will eventually end. The legislation that Congress adopts to prevent a similar event in the future is likely to be with us for 50 years.
The terms "too big to fail" and "systemically important" are virtually interchangeable. The reason that we might consider some financial institutions too big to fail (TBTF) is that their failure could produce substantial losses or other ill effects elsewhere in the economy--a systemic breakdown of some kind. Thus, if a firm is systemically important, it is also likely to be TBTF.
Understanding the virtual identity between these two terms is essential, because we should not be concerned about business failures unless they can have knock-on effects that could involve the whole economy or the whole financial system. There is real danger that policymakers will confuse efforts to prevent simple business failures with efforts to prevent systemic breakdowns. It is to the credit of the Obama administration that they have not claimed that the bankruptcy of General Motors would cause a systemic breakdown, even though GM's failure could cause widespread losses throughout the economy.
In this testimony, I will discuss the GM case frequently, as a way of testing whether we have adequate concepts for determining whether a financial firm is TBTF. If GM is not TBTF it raises questions whether any nonbank financial firm--no matter how large--is likely to be TBTF.
The discussion that follows will specifically address the four issues that Chairman Dodd outlined in his letter of invitation:
- Whether a new regulatory framework is desirable or feasible to prevent institutions from becoming "too big to fail" and posing the risk of systemic harm to the economy and financial system;
- Whether existing financial organizations considered "too big to fail" should be broken up;
- What requirements under a new regulatory framework are necessary to prevent or mitigate risks associated with institutions considered "too big to fail;" for example, new capital and disclosure requirements, as well as restrictions on size, affiliations, transactions, and leverage; and
- How to improve the current framework for resolving systemically important non-bank financial companies.
Is it desirable or feasible to develop a regulatory framework that will prevent firms from becoming TBTF or posing a risk of systemic harm?
A regulatory framework that will prevent companies from becoming TBTF--or causing systemic breakdowns if they fail--is only desirable or feasible if Congress can clearly define what it means by systemic harm or TBTF. If Congress cannot describe in operational terms where to draw the line between ordinary companies and companies that are TBTF--or if it cannot define what it means by "systemic harm"--it would not be good policy to give the power to do so to a regulatory agency. The standard, "I know it when I see it" may work when a systemic event is imminent, but not for empowering a regulatory agency to designate TBTF or systemically important firms in advance. If Congress does so, the likelihood of severe and adverse unintended consequences is quite high.
First, if a firm is designated in advance as TBTF (that is, as systemically important), it will have competitive advantages over other firms in the same industry and other firms with which it competes outside its industry. This is true because the TBTF designation confers important benefits. The most significant of these is probably a lower cost of funding, arising from the market's recognition that the risk of loss is significantly smaller in firms that the government will not allow to fail than it is in firms that might become bankrupt. Lower funding costs will translate inevitably--as it did in the case of Fannie Mae and Freddie Mac--into market dominance and consolidation. Market sectors in which TBTF firms are designated will come to be dominated and controlled by the large TBTF firms, and smaller firms will gradually be squeezed out. Ironically, this will also result in consolidation of risk in fewer and fewer entities, so that the likelihood of big firm collapses becomes greater and each collapse more disruptive. In some markets, status as TBTF has another advantage--the appearance of greater stability than competitors. In selling insurance, for example, firms that are designated as systemically important will be able to tell potential customers that they are more likely to survive and meet their obligations than firms that have not been so designated.
Accordingly, if there is to be a system of designating certain firms as systemically important, it is necessary to be able to state with some clarity what standards the agency must use to make that decision. Leaving the agency with discretion, without definitive standards, would be courting substantial unintended consequences. The natural tendency of a regulator would be to confer that designation broadly. Not only does this increase the regulator's size and power, but it also minimizes the likelihood--embarrassing for the regulator--that a systemic event will be caused by a firm outside the designated circle. Accordingly, the ability of Congress to define what it means by a TBTF firm would be important to maintain some degree of competitive vigor in markets that would otherwise be threatened by the designation of one or more large firms as systemically important and thus TBTF.
Second, apart from competitive considerations, it is necessary to consider the possibility that ordinary business failures might be prevented even though they would not have caused a systemic breakdown if they occurred. Again, the tendency of regulators in close cases will be to exercise whatever power they have to seize and bail out failing firms that might be TBTF. The incentives all fall in this direction. If a systemic breakdown does occur, the regulator will be blamed for failing to recognize the possibility, while if a firm is bailed out that would not in fact have caused a systemic breakdown, hardly anyone except those who are forced to finance it (a matter to be discussed later) will complain. This makes bailouts like AIG much more likely unless Congress provides clear guidelines on how a regulator is to identify a TBTF or systemically important firm.
The stakes for our competitive system are quite high in this case, because bailouts are not only costly, but they have a serious adverse effect on the quality of companies and managements that continue to exist. If firms are prevented from failing when they are not TBTF or otherwise systemically important, all other firms are weakened. This is because our competitive market system improves--and consumers are better served--through the "creative destruction" that occurs when bad managements and bad business models are allowed to fail. When that happens, the way is opened for better managements and business models to take their place. If failures are prevented when they should not be, the growth of the smaller but better managed and more innovative firms will be hindered. Overall, the quality and the efficiency of the firms in any market where this occurs will decline.
Finally, setting up a mechanism in which companies that should be allowed to fail are rescued from failure will introduce significant moral hazard into our financial system. This is true even if the shareholders of a rescued firm are wiped out in the process. Shareholders are not the group whose views we should be worried about when we consider moral hazard. Shareholders, like managements, benefit from risk-taking, which often produces high profits as well as high rates of failure. The class of investors we should be thinking about are creditors, who get no benefits whatever from risk-taking. They are the ones who are in the best position to exercise market discipline, and they do so by demanding higher rates of interest when they see greater risk-taking in a potential borrower. To the extent that the wariness of creditors is diminished by the sense that a company may be rescued by the government, there will be less market discipline by creditors and increased moral hazard. The more companies that are added to the list of firms that might be rescued, the greater the amount of moral hazard that has been introduced to the market. The administration's plan clearly provides for possible rescue, since it contemplates either a receivership (liquidation) or a conservatorship (generally a way to return a company to health and normal operations).
Accordingly, although it is exceedingly important for Congress to be clear about when a company may be designated as TBTF, it will be very difficult to do so. This is illustrated by the GM case. GM is one of the largest companies in the U.S.; its liquidation, if it occurs, could cause a massive loss of jobs not only at GM itself but at all the suppliers of tires, steel, fabrics, paints, and glass that go into making a car, all the dealers that sell the cars, all the banks that finance the dealers, and all the communities, localities, and states throughout the U.S. that depend for their revenues on the taxes paid by these firms and their employees. In other words, there would be very serious knock-on effects from a GM failure. Yet, very few people are suggesting that GM is TBTF in the same way that large financial institutions are said to be TBTF. What is the difference?
This question focuses necessary attention on two questions: what it means to be TBTF and the adequacy of the bankruptcy system to resolve large firm failures. If GM is not TBTF, why not? The widespread losses throughout the economy would certainly suggest a systemic effect, but if that is not what we mean by a systemic effect, what is it that we are attempting to prevent? On the other hand, if that is what we mean by a systemic effect, should the government then have the power to resolve all large companies--and not just financial firms--outside the bankruptcy system? The fact that GM may ultimately go into bankruptcy and be reorganized under Chapter 11 suggests that the bankruptcy system is adequate for large financial nonbank institutions, unless the propensity of nonbank financial institutions to create systemic breakdowns can be distinguished from that of operating companies like GM. Later in this testimony, I will argue that this distinction cannot be sustained.
The forgoing discussion highlights the difficulty of defining both a systemic event and a systemically important or TBTF firm, and also the importance of defining both with clarity. Great harm could come about if Congress--without establishing any standards--simply authorizes a regulatory agency to designate TBTF companies, and authorizes the same or another agency to rescue the companies that are so designated. My answer, then, to the Committee's first question is that--given the great uncertainty about (i) what is a systemic event, (ii) how to identify a firm that is TBTF, and (iii) what unintended consequences would occur if Congress were not clear about these points--it would be neither desirable nor feasible to set up a structure that attempts to prevent systemic harm to the economy by designating systemically important firms and providing for their resolution by a government agency rather than through the normal bankruptcy process.
Nevertheless, it would not be problematic to create a body within the executive branch that generally oversees developments in the market and has the responsibility of identifying systemic risk, wherever it might appear to be developing within the financial sector. The appropriate body to do this would be the President's Working Group (PWG), which consists of most of the major federal financial supervisors and thus has a built-in market-wide perspective. The PWG currently functions under an executive order, but Congress could give it a formal charter as a government agency with responsibility for spotting systemic risk as well as coordinating all financial regulatory activity in the Executive Branch.
Breaking Up Systemically Significant or TBTF Firms
There could be constitutional objections to a breakup--based on the takings and due process clauses--unless there are clear standards that justify it. I am not a constitutional lawyer, but a fear that a company might create a systemic breakdown if it fails does not seem adequate to take the going concern value of a large company away from its shareholders. As we know from antitrust law, firms can be broken up if they attempt to monopolize, and under certain other limited circumstances. But in those cases, there are standards for market dominance and for the requisite intent to use it in order to create a monopoly--and both are subject to rigorous evidentiary standards. As I pointed out above, there are no examples that define a systemic risk or why one company might cause it and another might not. Accordingly, providing authority for a government agency to break up companies that are deemed to be systemically risky could be subject to constitutional challenge.
In addition, as a matter of policy, breaking up large institutions would seem to create many more problems than it would solve. First, there is the question of breaking up successful companies. If companies have grown large because they are successful competitors, it would be perverse to penalize them for that, especially when we aren't very sure whether they would in fact cause a systemic breakdown if they failed. In addition, our economy is made up of large as well as small companies. Large companies generally need large financial institutions to meet their financing needs. This is true whether we are talking about banks, securities firms, insurance companies, finance companies, or others. Imagine a large oil company trying to insure itself against property or casualty losses with a batch of little insurance companies. The rates it would have to pay would be much higher, if it could get full coverage at all. Or imagine the same oil company trying to pay its employees worldwide without a large U.S. bank with worldwide operations, or the same company trying to place hundreds of millions of dollars in commercial paper each week through small securities firms without a global reach.
There are also international competitive factors. If other countries did not break up their large financial institutions, our large operating companies would probably move their business to the large foreign financial institutions that could meet their needs.
Leaving our large operating companies without an alternative source of funding could also be problematic, in the event that a portion of the financial markets becomes unavailable--either in general or for a specific large firm. The market for asset-backed securities closed down in the summer of 2007 and hasn't yet reopened. Firms that used to fund themselves through this market were then compelled to borrow from banks or to use commercial paper or other debt securities. This is one of the reasons that the banks have been reluctant to lend to new customers; they have been saving their cash for the inevitable withdrawals by customers that had been paying over many years for lines of credit that they could use when they needed emergency funds. The larger firms might not have been able to find sufficient financial resources if the largest banks or other financial institutions had been broken up.
The breakup of large financial firms would create very great risks for our economy, with few very benefits, especially when we really have no idea whether any particular firm that might be broken up actually posed a systemic risk or would have created a systemic breakdown if it had failed.
Are there regulatory actions we can take to mitigate or prevent systemic risk caused by TBTF companies?
For the reasons outlined below, it is my view that only the failure of a large commercial bank can create a systemic breakdown, and that nonbank financial firms--even large ones--are no more likely than GM to have this effect. For that reason, I would not designate any nonbank financial institution (other than a commercial bank) as systemically important, nor recommend safety and soundness supervision of any financial institutions other than those where market discipline has been impaired because they are backed by the government, explicitly or implicitly.
The track record of banking regulation is not good. In the last 20 years we have had two very serious banking crises, including the current one, when many banks failed and adversely affected the real economy. The amazing thing is that--despite this record of failure--the first instinct of many people in Washington it is to recommend that safety and soundness regulation be extended to virtually the entire financial system through the regulation and supervision of systemically important (or TBTF) firms. After the S&L debacle and the failure of almost 1600 commercial banks at the end of the 1980s and the beginning of the 1990s, Congress adopted the Federal Deposit Insurance Corporation Improvement Act of 1991 (FDICIA), a tough regulatory statute that many claimed would put an end to banking crises. Yet today we are in the midst of a banking crisis that some say could be as bad as that of the Great Depression, perhaps even worse. If banks were not backed by the government--through deposit insurance, a lender of last resort, and exclusive access to the payment system--their risk-taking would probably be better controlled by market discipline exerted by creditors. But given the government support they receive, and its effect in impairing market discipline, regulation and supervision of their safety and soundness is the only sensible policy.
Nevertheless, there are some reasonable steps that could be taken to improve bank regulation and to mitigate the possibility that the failure of a large bank might in the future have a significant adverse effect on other economic actors. For the reasons outlined above, I don't think that restrictions on size are workable, and they are likely to be counterproductive. The same thing is true of restrictions on affiliations and transactions, both of which will impose costs, impair innovation, and reduce competition. Since we have no idea whether any particular firm will cause a systemic breakdown if it fails, it does not seem reasonable to impose all these burdens on our financial system for very little demonstrable benefit. Restrictions on leverage can be effective, but I see them as an element of capital regulation, as discussed below.
A good example of the unintended consequences of imposing restriction on affiliations is what has happened because of the restrictions on affiliations between banks and commercial firms. As the Committee knows, the Bank Holding Company Act provides that a bank cannot be affiliated with any activity that is not "financial in nature." For many years the banking industry has used this to protect themselves against competition by organizations outside banking, most recently competition from Wal-Mart. They and others have argued that the separation of banking and commerce (actually, after the Gramm-Leach-Bliley Act was adopted in 1999, the principle became the separation of finance and commerce) was necessary to prevent the extension of the so-called federal "safety net" to commercial firms. That idea has now backfired on the banks, because by keeping commercial firms out of the business of investing in banks, they have made it very difficult for banks to raise the capital they need in the current financial crisis. We should not impose restrictions on affiliations unless there is strong evidence that a particular activity is harmful. All such restrictions turn out to be restrictions on competition and ultimately hurt consumers, who must pay higher prices and get poorer services. Because Wal-Mart was unable to compete with banks, many Wal-Mart customers pay more for banking services than they should, and many of them can't get banking services at all.
Nevertheless, capital requirements can be used effectively to limit bank risk-taking and growth, and this would be far preferable to other kinds of restrictions. It would make sense to raise bank capital requirements substantially. The only reason banks are able to keep such low capital ratios is that they have government backing. In addition, capital requirements should be raised as banks grow larger, which is in part the result of higher asset values that accompany a growing market. An increase of capital requirements with size would also have the salutary effect of dampening growth by making it more expensive, and it would provide a strong countercyclical brake on the development of asset bubbles. Higher capital requirements as banks grow larger would also induce them to think through whether all growth is healthy, and what lines of business are most suitable and profitable. In addition, as bank profits grow, capital requirements or reserves should also be increased in order to prepare banks for the inevitable time when growth will stop and the decline sets in. Before the current crisis, 10 percent risk-based capital was considered well-capitalized, but it is reasonably apparent now that this level was not high enough to withstand a serious downturn.
In addition, regulation should be used more effectively to enhance market discipline. Bank regulators are culturally reluctant to release information on the banks they supervise. This too often leaves market participants guessing about the risks the banks are taking--and wrongly assuming that the regulators are able to control these risks. To better inform the markets, the regulators, working with bank analysts, should develop a series of metrics or indicators of risk-taking that the banks should be required to publish regularly--say, once every month. This would enable the markets to make more informed judgments about bank risk-taking and enhance the effectiveness of market discipline. Rather than fighting market discipline, bank regulators should harness it in this way to supplement their own examination work.
Finally for larger commercial banks, especially the ones that might create systemic risk if they failed, it would be a good idea to require the issuance of a form of tradable subordinated debt that could not by law be bailed out. The holders of this debt would have a strong interest in better disclosure by banks and could develop their own indicators of risk-taking. As the market perceived that a bank was taking greater risk, the price of these securities would fall and its yield would rise. The spread of that yield over Treasuries would provide a continuing strong signal to a bank's supervisor that the market foresees trouble ahead if the risk-taking continues. Using this data, the supervisor could clamp down on activities that might result in major losses and instability at a later time.
Can we improve the current framework for resolving systemically important nonbank financial firms?
The current framework for resolving all nonbank financial institutions is the bankruptcy system. Based on the available evidence, there is no reason to think that it is inadequate for performing this task or that these institutions need a government-administered resolution system. Because of the special functions of banks, a special system for resolving failed banks is necessary, but as discussed below banks are very different from other financial institutions. The creation of a government-run system will increase the likelihood of bailouts of financial institutions and prove exceedingly costly to the financial industry or to the taxpayers, who are likely to end up paying the costs.
The underlying reason for the administration's proposal for a special system of resolution for nonbank financial institutions is the notion that the failure of a large financial firm can create a systemic breakdown. Thus, although many people look at the administration's resolution plan as a means to liquidate systemically important or TBTF firms in an orderly way, it is more likely to be a mechanism for bailing out these firms so that they will not cause a systemic breakdown. The Fed's bailout of AIG is the paradigm for this kind of bailout, which sought to prevent market disruption by using taxpayer funds to prevent losses to counterparties and creditors.
As support for its proposal, the administration cites the "disorderly" bailout of AIG and the market's panicked reaction to the failure of Lehman Brothers. On examination, these examples turn out to be misplaced. Academic studies after both events show that the market's reaction to both was far more muted than the administration suggests. Moreover, the absence of any recognizable systemic fallout from the Lehman bankruptcy--with the exception of a single money market mutual fund, no other firm has reported or shown any serious adverse effects--provides strong evidence that in normal market conditions the reaction to Lehman's failure would not have been any different from the reaction to the failure of any large company. These facts do not support the notion that a special resolution mechanism is necessary for any financial institutions other than banks.
The special character of banks. Although the phrase "shadow banking" is thrown around to imply a strong similarity between commercial banks and other financial institutions such as securities firms, hedge funds, finance companies or insurers, the similarity is illusory in most important respects. Anyone can lend; only banks can take deposits. Deposit-taking--not lending--is the essence of banking. By offering deposits that can be withdrawn on demand or used to pay others through an instruction such as a check, banks and other depository institutions have a special and highly sensitive role in our economy. If a bank should fail, its depositors are immediately deprived of the ready funds they expected to have available for such things as meeting payroll obligations, buying food, or paying rent. Banks also have deposits with one another, and small banks often have substantial deposits in larger banks in order to facilitate their participation in the payment system.
Because of fear that a bank will not be able to pay in full on demand, banks are also at risk of "runs"--panicky withdrawals of funds by depositors. Runs can be frightening experiences for the public and disruptive for the financial system. The unique attribute of banks--that their liabilities (deposits) may be withdrawn on demand--is the reason that banks are capable of creating a systemic event if they fail. If bank customers cannot have immediate access to their funds, or if a bank cannot make its scheduled payments to other banks, the others can also be in trouble, as can their customers. That is the basis for a true systemic event. The failure of a bank can leave its customers and other banks without the immediate funds they are expecting to use in their daily affairs. The failure of a large bank can cause other failures to cascade through the economy, theoretically creating a systemic event. I say "theoretically" because the failure of a large bank has never in modern times caused a systemic event. In every case where a large bank might have failed and caused a systemic breakdown, it has been rescued by the FDIC. The most recent such case--before the current crisis--was the rescue of Continental Illinois Bank in 1984.
The foregoing description of how a large bank's failure can cause a systemic breakdown raises a number of questions about whether and how a systemic breakdown can be caused by the failure of a nonbank financial institution. These financial institutions--securities firms, hedge funds, insurance companies, finance companies, and others--tend to borrow for a specific term or to borrow on a collateralized basis. In this respect, they are just like GM. In common with all other large commercial borrowers, nonbank financial institutions also fund themselves with short-term commercial paper. Unless they are extremely good credits, this paper is collateralized. If they should fail, their creditors can recoup their losses by selling the collateral. Their failures, then, do not cause any immediate cash losses to their lenders or counterparties. Losses occur, to be sure, but in the same way that losses will occur if GM should file for bankruptcy--those who suffer them do not lose the immediate access to cash that they were expecting to use for their current obligations, and thus there is rarely any contagion in which the losses of one institution are passed on to others in the kind of cascade that can occur when a bank fails. It is for this reason that describing the operations of these nondepository institutions as "shadow banking" is so misleading. It ignores entirely the essence of banking--which is not simply lending--and how it differs from other kinds of financial activity.
Because of the unique effects that are produced by bank failures, the Fed and the FDIC have devised systems for reducing the chances that banks will not have the cash to meet their obligations. The Fed lends to healthy banks (or banks it considers healthy) through what is called the discount window--making cash available for withdrawals by worried customers--and the FDIC will normally close insolvent banks just before the weekend and open them as healthy, functioning new institutions on the following Monday. In both cases, the fears of depositors are allayed and runs seldom occur. The policy question facing Congress is whether it makes sense to extend FDIC bank resolution processes to other financial institutions. For the reasons outlined above, there is virtually no reason to do so for financial institutions other than banks.
Before proceeding further, it is necessary to correct some misunderstandings about the effectiveness of the FDIC, which has been presented by the administration and others as a paragon in the matter of resolving banks. The facts suggest a different picture, and should cause policymakers to pause before authorizing the FDIC or any other agency to take over the resolution of nonbank financial institutions. The FDIC and the other bank regulators function under a FDICIA requirement for prompt corrective action (PCA) when a bank begins to weaken. The objective of PCA is to give the FDIC and other supervisors the authority to close a bank before it actually becomes insolvent, thus saving both the creditors and the FDIC insurance fund from losses. It has not worked out that way. Thus far in 2009, there have been 32 reported bank failures for which the FDIC has reported its losses. In these cases, the losses on assets have ranged from 8 percent to 45 percent, with both an average and a weighted average of 28 percent. In 2008, there were 25 bank failures, with losses averaging 25 percent. There may be reasons for these extraordinary losses, including the difficulty of dealing with the primary federal or state regulator, but the consistency of the losses in the face of the PCA requirement casts some doubt on the notion that even the best federal resolution agency--dealing with failing insurance companies, securities firms, hedge funds and others--would be able to do a more efficient job than a bankruptcy court.
While the failures of the FDIC as a resolution agency are not well known, the weakness of the bankruptcy system as a way of resolving failing financial institutions has been exaggerated. The evidence suggests that Lehman's bankruptcy filing--as hurried as it was--has resulted in a more orderly resolution of the firm than AIG's rescue by the Fed. As reported by professors Kenneth Ayotte and David Skeel, things moved with dispatch after Lehman filed for bankruptcy under Chapter 11 of the code. Thus, as Ayotte and Skeel note:
"Lehman filed for Chapter 11 on September 15, 2008. Three days later, Lehman arranged a sale of its North American investment banking business to Barclays, and the sale was quickly approved by the court after a lengthy hearing . . . Its operations in Europe, the Middle East, and Asia were bought by Nomura, a large Japanese brokerage firm. By September 29, Lehman had agreed to sell its investment management business to two private equity firms."
Chapter 11 allows bankrupt debtors to remain in possession of their assets and continue operating while their creditors reach agreement on how best to divide up the firm's assets. It also permits firms to return to financial health if their creditors conclude that this is more likely to result in a greater recovery than a liquidation. In other words, Chapter 11 provides a kind of bailout mechanism, but one that is under the control of the creditors--the parties that have suffered the real losses. Neither the taxpayers nor any other unrelated party is required to put in any funds to work out the failed company.
There are many benefits of a bankruptcy that are not likely to come with a system of resolution by a government agency. These include certainty about the rights of the various classes of creditors; a well-understood and time-tested set of procedures; the immediate applicability of well-known stay provisions that prevent the disorderly seizure of collateral; equally well-known exemptions from stay provisions so that certain creditors holding short-term obligations of the failed company can immediately sell their collateral; and well worked out rules concerning when and under what circumstances preferential payments to certain creditors by the bankrupt firm have to be returned to the bankrupt estate.
Still, the examples of Lehman Brothers and AIG have had a significant impact on the public mind and a hold on the attitudes of policymakers. It is important to understand these cases, and the limited support they provide for setting up a system for resolving large nonbank financial institutions.
The market reactions after the failures of AIG and Lehman are not examples of systemic risk. Secretary Geithner has defended his proposal for a resolution authority by arguing that, if it had been in place, the rescue of AIG last fall would have been more "orderly" and the failure of Lehman Brothers would not have occurred. Both statements might be true, but would that have been the correct policy outcome? Recall that the underlying reason for the administration's plan to designate and specially regulate systemically important firms is that the failure of any such company would cause a systemic event--a breakdown in the financial system and perhaps the economy as a whole. If this is the test, it is now reasonably clear that neither AIG nor Lehman is an example of a large firm creating systemic risk or a systemic breakdown.
In a widely cited paper and a recent book, John Taylor of Stanford University concluded that the market meltdown and the freeze in interbank lending that followed the Lehman and AIG events in mid-September 2008 did not begin until the Treasury and Fed proposed the initial Troubled Asset Relief Program later in the same week, an action that implied that financial conditions were much worse than the markets had thought. Taylor's view, then, is that AIG and Lehman were not the cause of the meltdown that occurred later that week. Since neither firm was a bank or other depository institution, this analysis is highly plausible. Few of their creditors were expecting to be able to withdraw funds on demand to meet payrolls or other immediate expenses, and later events and data have cast doubt on whether the failure of Lehman or AIG (if it had not been bailed out) would have caused the losses that many have claimed.
In another analysis after the Lehman and AIG events, Ayotte and Skeel concluded that the evidence suggests "at a minimum, that the widespread belief that the Lehman Chapter 11 filing was the singular cause of the collapse in credit that followed is greatly overstated." They also show that that there was very little difference between the market's reaction to Lehman and to AIG, although the former went into bankruptcy and the latter was rescued.
Advocates of broader regulation frequently state that financial institutions are now "interconnected" in a way that they have not been in the past. This idea reflects a misunderstanding of the functions of financial institutions, all of which are intermediaries in one form or another between sources of funds and users of funds. In other words, they have always been interconnected in order to perform their intermediary functions. The right question is whether they are now interconnected in a way that makes them more vulnerable to the failure of one or more institutions than they have been in the past, and there is no evidence of this. The discussion below strongly suggests that there was no need to rescue AIG and that Lehman's failure was problematic only because the market was in an unprecedentedly fragile and panicky state in mid-September 2008.
This distinction is critically important. If the market disruption that followed Lehman's failure and AIG's rescue was not caused by these two events, then identifying systemically important firms and supervising them in some special way serves no purpose. Even if the failure of a systemically important firm could be prevented through regulation--a doubtful proposition in light of the current condition of the banking industry--that in itself would not prevent the development of a fragile market, or its breakdown in the aftermath of a serious shock. The weakness or failure of individual firms is not the source of the problem. In terms of a conventional systemic risk analysis, the chaos that followed was not the result of a cascade of losses flowing through the economy as a result of the failure of Lehman or the potential failure of AIG. In the discussion that follows, I show first that Lehman did not cause, and AIG would not have caused, losses to other firms that might have made them systemically important. I then show that both are examples of nonbank financial firms that can be successfully resolved--at no cost to the taxpayers--through the bankruptcy process rather than a government agency.
AIG should have been sent into bankruptcy. AIG's quarterly report on Form 10-Q for the quarter ended June 30, 2008--the last quarter before its bailout in September--shows that the $1 trillion company had borrowed, or had guaranteed subsidiary borrowings, in the amount of approximately $160 billion, of which approximately $45 billion was due in less than one year. Very little of this $45 billion was likely to be immediately due and payable, and thus, unlike a bank's failure, AIG's failure would not have created an immediate cash loss to any significant group of lenders or counterparties. Considering that the international financial markets have been estimated at more than $12 trillion, the $45 billion due within a year would not have shaken the system. Although losses would eventually have occurred to all those who had lent money to or were otherwise counterparties of AIG, these losses would have occurred over time and been worked out in a normal bankruptcy proceeding, after the sale of its profitable insurance subsidiaries.
Many of the media stories about AIG have focused on the AIG Financial Products subsidiary and the obligations that this group assumed through credit default swaps (CDSs). However, it is highly questionable whether there would have been a significant market reaction if AIG had been allowed to default on its CDS obligations in September 2008. CDSs--although they are not insurance--operate like insurance; they pay off when there is an actual loss on the underlying obligation that is protected by the CDS. It is much the same as when a homeowners' insurance company goes out of business before there has been a fire or other loss to the home. In that case, the homeowner must go out and find another insurance company, but he has not lost anything except the premium he has paid. If AIG had been allowed to default, there would have been little if any near-term loss to the parties that had bought protection; they would simply have been required to go back into the CDS market and buy new protection. The premiums for the new protection might have been more expensive than what they were paying AIG, but even if that were true, many of them had received collateral from AIG that could have been sold in order to defray the cost of the new protection. CDS contracts normally require a party like AIG that has sold protection to post collateral as assurance to its counterparties that it can meet its obligations when they come due.
This analysis is consistent with the publicly known facts about AIG. In mid-March, the names of some of the counterparties that AIG had protected with CDSs became public. The largest of these counter- parties was Goldman Sachs. The obligation to Goldman was reported as $12.9 billion; the others named were Merrill Lynch ($6.8 billion), Bank of America ($5.2 billion), Citigroup ($2.3 billion), and Wachovia ($1.5 billion). Recall that the loss of CDS coverage--the obligation in this case--is not an actual cash loss or anything like it; it is only the loss of coverage for a debt that is held by a protected party. For institutions of this size, with the exception of Goldman, the loss of AIG's CDS protection would not have been problematic, even if they had in fact already suffered losses on the underlying obligations that AIG was protecting. Moreover, when questioned about what it would have lost if AIG had defaulted, Goldman said its losses would have been "negligible." This is entirely plausible. Its spokesman cited both the collateral it had received from AIG under the CDS contracts and the fact that it had hedged its AIG risk by buying protection against AIG's default from third parties.
Also, as noted above, Goldman only suffered the loss of its CDS coverage, not a loss on the underlying debt the CDS was supposed to cover. If Goldman, the largest counterparty in AIG's list, would not have suffered substantial losses, then AIG's default on its CDS contracts would have had no serious consequences in the market. This strongly suggests that AIG could have been put into bankruptcy with no costs to the taxpayers, and if it had not been rescued its failure would not have caused any kind of systemic risk. On the other hand, it is highly likely that a systemic regulator would have rescued AIG--just as the Fed did--creating an unnecessary cost for U.S. taxpayers and an unnecessary windfall for AIG's counterparties.
Lehman's failure did not cause a systemic event. Despite the contrary analyses by Taylor, Skeel, and Ayotte, it is widely believed that Lehman's failure proves that a large company's default, especially when it is "interconnected" through CDSs, can cause a systemic breakdown. If that were true, then it might make sense to set up a regulatory structure to prevent a failure by a systemically important company. But it is not true. Even if we accept that Lehman's failure somehow precipitated the market freeze that followed, that says nothing about whether, in normal market conditions, Lehman's failure would have caused the same market reaction. In fact, analyzed in light of later events, it is likely that Lehman's bankruptcy would have had no substantial adverse effect on the financial condition of its counterparties. In other words, the failure would not--in a normal market--have caused the kind of cascade of losses that defines a systemic breakdown.
After Lehman's collapse, there is only one example of any other organization encountering financial difficulty because of Lehman's default. That example is the Reserve Fund, a money market mutual fund that held a large amount of Lehman's commercial paper at the time Lehman defaulted. This caused the Reserve Fund to "break the buck"--to fail to maintain its share price at exactly one dollar--and it was rescued by the Treasury and Fed. The need to rescue the Reserve Fund was itself another artifact of the panicky conditions in the market at the time. That particular fund was an outlier among all funds in terms of its risks and returns. The fact that there were no other such cases, among money market funds or elsewhere, demonstrates that the failure of Lehman in a calmer and more normal market would not have produced any of the significant knock-on effects that are the hallmark of a systemic event. It is noteworthy, in this connection, that a large securities firm, Drexel Burnham Lambert, failed in 1990 and went into bankruptcy without any serious systemic effects. In addition, when Lehman's CDS obligations were resolved a month after its bankruptcy, they were all resolved by the exchange of only $5.2 billion among all the counterparties, a minor sum in the financial markets and certainly nothing that in and of itself would have caused a market meltdown.
So, what relationship did Lehman's failure actually have to the market crisis that followed? The problems that were responsible for the crisis had actually begun more than a year earlier, when investors lost confidence in the quality of securities--particularly mortgage-backed securities (MBS)--that had been rated AAA by rating agencies. As a result, the entire market for asset-backed securities of all kinds became nonfunctional, and these assets simply could not be sold at anything but a distress price. With large portfolios of these securities on the balance sheets of most of the world's largest financial institutions, the stability and even the solvency of these institutions--banks and others--were in question.
In this market environment, Bear Stearns was rescued through a Fed-assisted sale to JPMorgan Chase in March 2008. The rescue was not necessitated because failure would have caused substantial losses to firms "interconnected" with Bear, but because the failure of a large financial institution in this fragile market environment would have caused a further loss of confidence--by investors, creditors, and counterparties--in the stability of other financial institutions. This phenomenon is described in a 2003 article by professors George Kaufman and Kenneth Scott, who write frequently on the subject of systemic risk. They point out that when one company fails, investors and counterparties look to see whether the risk exposure of their own investments or counterparties is similar: "The more similar the risk-exposure profile to that of the initial [failed company] economically, politically, or otherwise, the greater is the probability of loss and the more likely are the participants to withdraw funds as soon as possible. The response may induce liquidity and even more fundamental solvency problems. This pattern may be referred to as a ‘common shock' or ‘reassessment shock' effect and represents correlation without direct causation." In March 2008, such an inquiry would have been very worrisome; virtually all large financial institutions around the world held, to a greater or lesser extent, the same assets that drove Bear toward default.
Although the rescue of Bear temporarily calmed the markets, it led to a form of moral hazard—the belief that in the future governments would rescue all financial institutions larger than Bear. Market participants simply did not believe that Lehman, just such a firm, would not be rescued. This expectation was shattered on September 15, 2008, when Lehman was allowed to fail, leading to exactly the kind of reappraisal of the financial health and safety of other institutions described by Kaufman and Scott. That is why the market froze at that point; market participants were no longer sure that the financial institutions they were dealing with would be rescued, and thus it was necessary to examine the financial condition of their counterparties much more carefully. For a period of time, the world's major banks would not even lend to one another. So what happened after Lehman was not the classic case of a large institution's failure creating losses at others--the kind of systemic event that has stimulated the administration's effort to regulate systemically important firms. It was caused by the weakness and fragility of the financial system as a whole that began almost a year earlier, when the quality of MBS and other asset-backed securities was called into question and became unmarketable. If Lehman should have been bailed out, it was not because its failure would have caused losses to others--the reason for the designation of systemically important or TBTF firms--but because the market was in an unprecedented condition of weakness and fragility. The correct policy conclusion arising out of the Lehman experience is not to impose new regulation on the financial markets, but to adopt policies that will prevent the correlation of risks that created a weak and fragile world-wide financial market well before Lehman failed.
Thus, Lehman didn't cause, and AIG (if it had been allowed to fail) wouldn't have caused, a systemic breakdown. They are not, then, examples of why it is necessary to set up a special resolution system, outside the bankruptcy process, to resolve them or other large nonbank financial firms. Moreover, and equally important, a focus on Lehman and AIG as the supposed sources of systemic risk is leading policymakers away from the real problem, which is the herd and other behavior that causes all financial institutions to become weak at the same time.
The funding question. There is also the question of how a resolution system of the kind the administration has proposed would be financed. Funds from some source are always required if a financial institution is either resolved or rescued. The resolution of banks is paid for by a fund created from the premiums that banks pay for deposit insurance; only depositors are protected, and then only up to $250,000. Unless the idea is to create an industry-supported fund of some kind for liquidations or bailouts, the administration's proposal will require the availability of taxpayer funds for winding up or bailing out firms considered to be systemically important. If the funding source is intended to be the financial industry itself, it would have to entail a very large levy on the industry. The funds used to bail out AIG alone are four times the size of the FDIC fund for banks and S&Ls when that fund was at its highest point--about $52 billion in early 2007. If the financial industry were to be taxed in some way to create such a fund, it would put all of these firms--including the largest--at a competitive disadvantage vis-à-vis foreign competitors and would, of course, substantially raise consumer prices and interest rates for financial services.
The 28 percent loss rate that the FDIC has suffered on failed banks during the past year should provide some idea of what it will cost the taxpayers to wind up or (more likely) bail out failed or failing financial institutions that the regulators flag as systemically important. The taxpayers would have to be called upon for most, if not all, of the funds necessary for this purpose. So, while it might be attractive to imagine the FDIC will resolve financial institutions of all kinds more effectively than the way it resolves failed or failing banks, a government-run resolution system opens the door for the use of taxpayer funds for unnecessary bailouts of companies that would not cause systemic breakdowns if they were actually allowed to fail.
Sometimes it is argued that bank holding companies (BHCs) must be made subject to the same resolution system as the banks themselves, but there is no apparent reason why this should be true. The whole theory of separating banks and BHCs is to be sure that BHCs could fail without implicating or damaging the bank, and this has happened frequently. If a holding company of any kind fails, its subsidiaries can remain healthy, just as the subsidiaries of a holding company can go into bankruptcy without the parent becoming insolvent. If a holding company with many subsidiaries regulated by different regulators should go into bankruptcy, there is no apparent reason why the subsidiaries cannot be sold off if they are healthy and functioning, just as Lehman's broker-dealer and other subsidiaries were promptly sold off after Lehman declared bankruptcy. If there is some conflict between regulators, these--like conflicts between creditors--would be resolved by the bankruptcy court.
Moreover, if the creditors, regulators, and stakeholders of a company believe that it is still a viable entity, Chapter 11 of the Bankruptcy Code provides that the enterprise can continue functioning as a "debtor in possession" and come out of the proceeding as a slimmed-down and healthy business. Several airlines that are functioning today went through this process, and--ironically--some form of prepackaged bankruptcy that will relieve the auto companies of their burdensome obligations is one of the options the administration is considering for that industry. (Why bankruptcy is considered workable for the auto companies but not financial companies is something of a mystery.) In other words, even if it were likely to be effective and efficient--which is doubtful--a special resolution procedure for financial firms is unlikely to achieve more than the bankruptcy laws now permit.
In addition to increasing the likelihood that systemically important firms will be bailed out by the government, the resolution plan offered by the administration will also raise doubts about priorities among lenders, counterparties, shareholders, and other stakeholders when a financial firm is resolved or rescued under the government's control. In bankruptcy, the various classes of creditors decide, under the supervision of a court, how to divide the remaining resources of the bankrupt firm, and whether the firm's business and management are sufficiently strong to return it to health. In an FDIC resolution, insured depositors have a preference over other creditors, but it is not clear who would get bailed out and who would take losses under the administration's plan. One of the dangers is that politically favored groups will be given preferences, depending on which party is in power at the time a systemically important firm is bailed out.
Perhaps even more important, the FDIC's loss rate even under PCA demonstrates that the closing down of losing operations is slow and inefficient when managed by the government. Under the bankruptcy laws, the creditors have strong incentives to close a failing company and stop its losses from growing. As the FDIC experience shows, government agencies have a tendency to forbear, allowing time for the losses in a failing firm to grow even greater.
Given that bailouts are going to be much more likely than liquidations, especially for systemically important firms, a special government resolution or rescue process will also undermine market discipline and promote more risk-taking in the financial sector. In bailouts, the creditors will be saved in order to prevent a purported systemic breakdown, reducing the risks that creditors believe they will be taking in lending to systemically important firms. Over time, the process of saving some firms from failure will weaken all firms in the financial sector. Weak managements and bad business models should be allowed to fail. That makes room for better managements and better business models to grow. Introducing a formal rescue mechanism will only end up preserving bad managements and bad business models that should have been allowed to disappear while stunting or preventing the growth of their better-managed rivals. Finally, as academic work has shown again and again, regulation suppresses innovation and competition and adds to consumer costs.
Accordingly, there is no need to establish a special government system for resolving nonbank financial institutions, just as there is no need to do so for large operating companies like GM. If such a system were to be created for financial institutions other than banks--for which a special system is necessary--the unintended consequences and adverse results for the economy and the financial system would far outweigh any benefits.
Peter J. Wallison is the Arthur F. Burns Fellow in Financial Policy Studies at AEI.