Mr. Chairman, Ranking Member Bachus, and members of the Committee:

The Committee has asked the witness today to address a single question: Are some institutions too big to fail and if so what should we do about it? In this testimony I will address those two questions.

Are some institutions too big to fail?

The answer to that question is yes, but as discussed below those institutions are only large commercial banks.

When we say that a company is too big to fail (TBTF), we mean that if it fails it will cause damage to the financial system as a whole--in other words, that its failure will cause many other companies to be seriously weakened or forced into bankruptcy. One of the problems associated with attempting to prevent a systemic breakdown is that it is difficult to determine, in advance, whether the failure of a particular company will cause a systemic breakdown or merely a temporary disruption. For the same reason, it is very difficult to determine, in advance, when a company is TBTF.

Nevertheless, it is important to distinguish between failures that will cause disruption and those that will cause a systemic breakdown of some kind. We should not want to rescue firms from failure if their bankruptcy would only create a temporary disruption in the economy. Those companies should not be regarded as TBTF. If we embarked on a path that would rescue all large firms, because their failure would cause economic disruption, we would create moral hazard. Market discipline would be impaired as creditors thought that they would be rescued by the government. Bad managements and bad business models would be preserved--when, in reality, they should disappear to make room for new and better managements and business models. That's how innovation, efficiency and change occur in our economy.

In theory, however, it is possible to visualize how the failure of a large bank might have more than simply a disruptive effect. Companies deposit their payrolls in banks pending use; individuals deposit funds in banks in order to pay their bills and their mortgages; and small banks deposit funds in large banks as part of the payment system. So when a large bank fails there could be a cascade of immediate losses through the economy. If this happens, companies will not be able to meet their payrolls, individuals will not be able to pay their mortgages, and smaller banks will not be able to meet their obligations to pay out the funds that are withdrawable on demand. In other words, the failure of a large bank can cause a cascade of losses and failures through the economy that might qualify as a systemic breakdown--that is, something more than a mere temporary disruption.

Can a nonbank financial institution create systemic risk and thus be TBTF? On the other hand, it is very difficult to see how a nonbank financial institution like a bank holding company, insurance company, securities firm, finance company or hedge fund--no matter what its size--can cause a systemic breakdown or become TBTF. Using a bank holding company (BHC) as an example, it's useful to consider what would happen if a large nonbank financial institution like that were to fail. Importantly, its liabilities are not deposits, and are not withdrawable on demand. As a result, if it fails, very few of its creditors suffer any immediate cash losses. No business, for example, deposits its payroll with a BHC. The BHC may have short term creditors, but unless it's a very financially strong company these short term obligations are likely to be collateralized, and thus short term creditors can make themselves whole by selling their collateral. Most of a BHC's creditors are long term, however, and they will suffer a loss over time. Of course, if the company defaults on any of its debts, all its obligations will usually come due because of cross-default provisions in its outstanding borrowings, and it will thus be forced into bankruptcy.

This does not change the condition of long-term creditors; they simply now get in line to receive their share of the bankrupt company's assets, or they approve a plan of reorganization that returns the company to viability if they believe the company will eventually be able to pay them back.

But, still, how does a BHC's bankruptcy create a systemic breakdown? If most of its short-term creditors are made whole through the collateral they hold, and its long-term creditors will eventually take losses as the company goes through bankruptcy, it's not obvious that a systemic breakdown can occur. Even if the long-term creditors take losses, these losses occur over time; they will not be the immediate cash losses that occur when a bank fails. Moreover, the BHC's creditors are very likely to be institutions whose lending is widely diversified. They will lick their wounds, but will not be forced into bankruptcy because of the failure of the bank holding company. They will continue functioning. There would be no systemic breakdown.

Moreover, there is no evidence--none--that credit default swaps (CDS) or other derivatives had anything to do with what happened after Lehman, or that if AIG had been allowed to fail there would have been a catastrophic effect on the financial markets. Lehman's CDS were all cleaned up after its bankruptcy for a total of $5.2 billion exchanged among the various counterparties. Goldman Sachs was the largest CDS counterparty of AIG, with contracts valued at $12.9 billion. But when a spokesman for Goldman was asked what would have been the effect on Goldman if AIG had failed, the answer was that the effect would have been "negligible." As required in most CDS contracts, Goldman had received collateral from AIG before its rescue and had also hedged its AIG exposure. If Goldman, AIG's largest counterparty, would not have suffered significant losses, there is no reason to believe that anyone else would have suffered systemically significant losses either. After all, AIG's CDS--like all CDS--were simply like insurance or reimbursement contracts, with AIG in the position of the insurer. If AIG had failed, its counterparties--like the homeowner whose insurance company fails before he has a loss on his home--would have been required to find another insurer, but they would not have suffered any major loss.

Finally, we read all the time that financial companies are "interconnected," and that's the reason they must be treated specially. It's certainly true that financial firms are interconnected in some sense--that's the nature of financial firms, which are in the business of moving money from a place where it's not well-used to a place where it is better employed. To accomplish that, interconnections are necessary. But the question is not whether these firms are interconnected; it is whether these interconnections create cross-obligations that are so large as to make it probable that if one nonbank financial firm fails it will bring others down with it. There is no evidence for this, and it is highly unlikely for the reasons stated above. After Lehman failed, for example, there was only one case of another company encountering trouble. In that case, a money market mutual fund (the Reserve Fund) was unable to maintain the value of its shares at one dollar, and suffered a run. But beyond that, there is no indication that any other firm suffered serious losses as a result of Lehman's failure. Even the CDSs on Lehman, as noted above, were quickly settled with no known adverse effects.

I should add here, as an aside, that our banking laws have been structured so that the failure of a bank holding company should have no effect on the underlying bank or banks. It's simply the failure of a bank's shareholder. Banks are restricted by banking law and regulations from making loans of significant size to their parent company or affiliates, so that the bank is insulated from the failure of the holding company. The reason is that the holding company is or could be engaged in activities that are riskier than the activities of a bank, and is more likely to fail for that reason. Many on the committee will remember that this restriction was put in place to prevent the extension of the so-called "federal safety net" beyond banks themselves. It is ironic that the administration is now proposing to extend a safety net to the same companies that were not supposed to cost the government anything.

The important point, however, is that if a BHC fails there are very few immediate cash losses that render its creditors unable to meet their own obligations, and thus no cascade of losses through the economy. So if we define a systemic event as a kind of contagion in which the losses of one company spread to others and affect the whole economy, it seems that only the failure of a large bank can have this effect. In other words, in my view, only a large bank can be too big to fail.

The Lehman case. Having said this, there is one category of events that is frequently called a systemic breakdown but is not. Here I am referring to the kind of turmoil that occurred after the failure of Lehman Brothers in September 2008. In that case, there was an immediate freeze-up of lending by banks and other financial intermediaries around the world. Because no direct losses are known to have occurred as a result of Lehman's failure (except the Reserve Fund as described above), this was not a classic case of a systemic breakdown in which losses were transmitted through an economy or financial system. What happened after Lehman Brothers' failure is what is known as a "common shock"--an event that causes a market to stop functioning because the participants have encountered new information that nullifies their previous expectations about the future. In this case, in a classic example of moral hazard, market participants were shocked to learn that--despite the rescue of Bear Stearns the preceding March--the government did not intend to rescue every firm that was larger than Bear. This new and highly adverse information required all market participants to reassess whether their counterparties and borrowers were solvent and safe, since a government rescue could no longer be considered a near certainty. The result was a freeze-up in lending as every major institution hoarded cash while it reassessed the financial condition of its counterparties.

A market freeze-up that results from a common shock is not the same thing as a systemic event and can't be prevented by the regulation of individual institutions. It is the result of a loss of confidence in the future by market participants as a group, not the failure of a particular institution. In reality, there were two common shocks that led to the current crisis. The first was the recognition by market participants in the summer of 2007 that defaults on U.S. mortgages were much higher than expected and mortgage-backed securities backed by these mortgages and rated AAA were not nearly as safe as previously thought. This led to the downgrading of mortgage securities portfolios, the shutdown of the asset-backed securities market, and large financial losses at banks because of the influence of mark-to-market accounting. The second shock was the failure of Lehman. It is highly unlikely that the second shock would not have had the adverse effect that it did without (i) the prior rescue of Bear Stearns, which made Lehman's failure a shock, and (ii) the weakening of bank capital positions because of the shutdown of the asset backed market in mid-2007 and the resulting sharp loss in the value of asset-backed securities.

Bank regulation failed to prevent the losses at individual banks because neither the banks nor their supervisors recognized that the assets they were acquiring in the form of mortgage-backed securities (MBS) and other asset-backed securities (ABS) were not of AAA quality, and that the market for these securities would completely dry up when the poor quality of these securities became known. Even more important, there was no general recognition anywhere in the system that virtually all the world's major banks were buying and holding the same weak assets. This made them all subject to the same effect when the first shock--the loss of value for MBS and ABS--occurred in mid-2007. Once all these institutions were weakened at the same time, they became vulnerable to any shock that caused a sharp loss of confidence about the future. Lehman was that shock.

The crises of the past did not result in similar global financial collapses because most financial institutions were considered adequately capitalized and financially strong enough to survive a substantial change in circumstances. Accordingly, the failure of the large securities firm Drexel Burnham Lambert in 1990, the collapse of the Thai Baht and the Russian default later in that decade, and the failure of Penn Central and the relatively small Herstatt bank in the 1970s, all caused major disruptions in the financial markets when they occurred, but none caused a global financial meltdown. However, once all or almost all major banks are perceived as weak and unstable--as they were in 2008--anything that shook market confidence and disrupted expectations would have had the same effect as Lehman's failure.

This would include a major earthquake in the United States, the collapse of the government of a major oil exporting nation, or some other natural or unexpected catastrophe that causes market participants to recalibrate who is safe to deal with and who is not.

This leads to the conclusion that if we are to prevent a financial crisis in the future we should take steps to prevent virtually all major banks from taking on the same risks and becoming weak at the same time. To carry out this policy, it will be necessary to recognize in advance that the elements for a severe common shock are coming together. Thus, in order to prevent a recurrence of the financial crisis, the regulation of commercial banks should focus not only the safety and soundness of the individual bank, but also on safety and soundness of the banking system as a whole. In this way, we can minimize the chances that the failure of a large bank will create a systemic event, and the chances that the banking system as a whole will become so weak that any Lehman-like common shock will cause a financial meltdown. As outlined below, then, we should adopt a form of what might be called macro-prudential regulation.

This strategy also avoids the negative effects on economic growth that would flow from regulating nonbank financial institutions the way we regulate banks. These nonbank institutions are not backed by the federal government, and are still controlled by market discipline. Placing them under government regulation, as the administration proposes, would create moral hazard and give them substantial funding advantages over their smaller competitors. It would be like creating Fannie Maes and Freddie Macs in every sector of the financial economy where these institutions are designated for special regulatory treatment. Even more important, as distinguished from banks, these institutions are supposed to be risk-takers; they are supposed to fail at higher rates than commercial banks. There is no reason to keep them from failing. If we were to regulate all these institutions the way we regulate commercial banks we would suppress the risk-taking that drives growth and innovation in our economy.

If large commercial banks are too big to fail, what should we do about it?

Once we focus on large banks as the most likely sources of systemic risk--and as a bulwark against devastating common shocks--there are a number of steps we can take. These are generally of two kinds: first, to create a means for discovering conditions in the financial markets that might make the financial system vulnerable to a common shock; and, second, to place supervisory limits on banks that will (i) restrict their risk-taking, (ii) limit the their procyclical tendency to lend freely when asset prices are rising, and (ii) ensure that they have the capital to remain strong when the inevitable asset bubbles deflate.

1. A systemic risk council. As outlined above, one of the reasons for the current crisis is that virtually all large banks held the same weak assets--weak because they were not of high quality themselves and were subject to rapid devaluation if the market for them disappeared.. One way to address this problem would be to authorize some regulatory body to monitor the worldwide financial system and report to Congress and the public on the possible growth of systemic risk or the factors that might produce a serious common shock. A suitable body for this purpose would be the President's Working Group, reconstituted as a Systemic Risk Council. The Council, which would have a small staff of its own, would be able to use the combined knowledge of the bank regulators, as well as the SEC and the CFTC, to broaden its perspective on the markets.

2. Metrics of risk-taking. The bank supervisors, working with banks and bank analysts, should develop metrics and indicators of risk-taking that all banks would be required to publish regularly. One of the continuing functions of supervisors would be to assure that these metrics were kept up to date and consistently calculated and reported by the banks under their supervision. If properly designed, metrics of risk-taking would signal when a bank is holding assets that are subject to sharp declines in values, assets that are highly correlated with assets held by other banks, or that a bank is relying excessively on short term liabilities to fund long term assets. Regular publication of these metrics would enhance market discipline by alerting creditors more effectively to bank risk-taking.

3. Subordinated debt. The largest banks should be required to issue subordinated debt that by law could not be bailed out by the government. If the interest rate on these instruments were to rise substantially above the rate on Treasury securities, it would signal to regulators that the market perceives excessive risk-taking in the issuer bank.

4. Higher capital requirements for banks. We could require very large banks to reconsider the benefits of size by imposing higher capital requirements as banks grow above a certain level. In this way, the largest banks would be protecting themselves and the financial system against the possibility of their own failure, and would also have a strong financial incentive not to grow larger.

5. Countercyclical capital increases and other measures. We could put in place regulatory requirements that would operate countercyclically, tending to restrain bank growth when asset prices are rising and cushion bank losses when asset prices are falling. For example, requiring higher reserves or capital levels as asset values rise would accomplish this. Eventually, those values will deflate, and at that time we want banks to have enough capital cushions so that market confidence in their health is not eroded. Capital requirements could also be increased if a bank's ratio of short term liabilities to long term assets rises above a predetermined level. This would tend to discourage banks from borrowing short term in the money markets in order to profit from the spread between short term money costs and the returns on long term assets. This would reduce the tendency of banks to act procyclically in fostering asset bubbles.

6. Countercyclical macro-prudential measures. The Systemic Risk Council could be authorized to establish an acceptable level of bank growth and impose appropriate limits on growth that are not consistent with these limits. For example, the council could impose a higher leverage ratio on banks when it appears that asset prices have risen too quickly. The leverage ratio for U.S. banks is defined as total common equity divided by total assets. Well capitalized banks must maintain a leverage ratio of 5%; the minimum is 3%. Raising the bank leverage ratio would require banks to sell assets or restrict lending, which would tend to mitigate the growth of asset bubbles. This would be a more direct way of limiting bank contributions to asset bubbles than expecting the Fed to raise interest rates.

If these measures were put in place, and coupled solely with a focus on large commercial banks, we would minimize the likelihood of another financial crisis while maintaining the dynamism and risk-taking that economic growth requires.

Peter J. Wallison is the Arthur F. Burns Fellow in Financial Policy Studies at AEI.