We are so inured to bank regulation that we seldom stop to ask why we do it. Yet, when one looks deeply into the question, it is difficult to identify a sound policy reason for regulating banks. Most of the conventional explanations--inherent bank instability, deposit insurance, the Federal Reserve’s role as lender of last resort, or the Fed’s role in the large-dollar payment system--turn out on examination to be either unfounded or based on risks that the government need not take in order to foster the growth of the economy. Regulation, however, is not without significant costs. In recent decades, we have experienced the wholesale collapse of both the banking and savings-and-loan industries, with huge costs to the taxpayers and the economy. The fact that this happened to heavily regulated industries, and never happens in the absence of regulation, should tell us something about whether regulation does more harm than good. Although there is clearly a political consensus that regulation is efficacious or necessary, thoughtful students of the financial system should consider whether that consensus will survive in the absence of a compelling policy rationale.
In modern financial history there have only been two cases--banks and savings and loan associations (S&Ls) in the 1980s and early 1990s--where entire industries suffered wholesale collapse. That this was not a result of the deregulation of the S&Ls is shown by the fact that more commercial banks than S&Ls perished between 1980 and 1994, when the banking crisis essentially ended. During that period, 1,617 banks failed as compared with 1,295 S&Ls.
In other words, the two most heavily regulated industries turned out to be the least stable when economic conditions were unfavorable. There is no mystery about why this happened, and the problem is not that regulators are incompetent. The reason is the absence or reduction in market discipline--the loss of the wariness and skepticism that investors should feel when they commit funds to an enterprise--because of government regulatory policies. In the case of regulated depository institutions, these policies communicated a general sense to investors that government regulation was seeing to the safety and soundness of these enterprises, and thus that the usual risks of investment, and the usual need for monitoring, were much reduced. For ordinary depositors, market discipline was essentially eliminated by deposit insurance, and uninsured depositors and shareholders were lulled by previous actions of regulators into believing that even if the institution failed they would somehow be protected--and indeed many were. This condition--this lulling of investors--has a name: moral hazard.
The absence of market discipline explains the collapse of the banking and S&L industries. The resulting losses fell on the taxpayers--who were compelled to pay more than $150 billion to bail out insured depositors--and on the economy in general, which experienced a recession over several years because of the depressed real estate market that resulted. There was also a spectacular political effect, with a president turned out of office because of voter dissatisfaction with the weak economy and the tax increases that were thought necessary, in part, to pay for the depositor bailout.
With this record, one might suppose that the best course would have been to rethink the whole idea of regulating depository institutions, but this was not to be. In Washington, the preferred course is always to redouble the bet on government, and this is indeed what happened. The Federal Deposit Insurance Corporation Improvement Act of 1991 (FDICIA) added significant new regulations, including draconian penalties for violating the regulations or the orders of bank and S&L supervisors. Ironically, however, the most successful elements of FDICIA were rules governing the behavior of the regulators themselves. These required them to take “prompt corrective action” if the capital of depository institutions began to weaken--actions that became more stringent as the institutions’ capital levels declined. Leaving aside the difficult question of determining capital levels, this stratagem was successful because it became difficult for regulators to forbear in closing failing institutions--a natural human tendency to push off into the future unpleasant duties that do not have to be performed today.
Are Banks Inherently Unstable?
Despite the fact that regulation has produced some disastrous results, it might still be possible to justify it on various supervening grounds. In this paper, we will review these grounds and assess whether they continue to have force. Before addressing this question, however, it is necessary to consider the view that depository institutions such as banks and S&Ls are inherently less stable than other businesses. The reasoning behind this claim is that because these institutions take deposits that are withdrawable on demand, they are subject to panic “runs” that may bring down otherwise solvent institutions.
Work by my AEI colleague Charles W. Calomiris shows that historically the principal cause of bank instability was unit banking--laws and regulations that limited branch banking and thus the geographical diversification of banks. These laws made them unduly subject to local economic conditions. Banks with fewer restrictions on diversification failed less often than banks that were confined to small geographical areas or that were unit banks that did not have authority to branch at all. Indeed, in some states that permitted branch banking, failure rates for branching banks were negligible--less than .02 percent--suggesting that diversified banks are inherently stable enterprises that will survive as well as other businesses if left to their own devices. Moreover, it appears that depositors were generally able to distinguish solvent institutions from those that were failing, so runs generally occurred where the risk of failure--based on various ex ante indicators--was perceived to be high. The experience of the Canadian banking system, which has always been characterized by a small number of widely diversified banks, confirms the validity of the association between diversification and financial soundness. Canadian banks have never suffered the kinds of failures and panics that have recurred in the United States over the last two centuries, despite the fact that the Canadian
economy is closely tied to the U.S. economy and subject to the same economic and climatic conditions.
This suggests, ironically, that the principal cause of the instability of U.S. banks--the perceived need for more bank regulation--was the weakness caused by regulatory policies already in place. All businesses suffer losses and insolvencies--a healthy process that weeds out the weak managements and products--but only regulated depository institutions have failed simultaneously in wholesale numbers. Because policies that limited or prevented branching and diversification of banks were the rule in the United States for two hundred years--ending only near the end of the twentieth century--it is not possible to know whether fully diversified banks would have any greater rates of failure than any other business. But it seems likely that the instability banks have exhibited on occasions in the past is the result of regulation rather than a sensible rationale for regulation.
Why Regulate Stable, Diversified Banks?
If you ask most financially sophisticated people today why we regulate banks you will get three answers: (i) federal deposit insurance, which is necessary to protect the savings of small depositors, creates moral hazard and exposes the government and the taxpayers to risk; regulation is necessary to reduce or eliminate this risk; (ii) the Federal Reserve is a lender of last resort to the banking system and, again, regulation protects the Fed and the FDIC against losses that will occur when it lends to banks that later fail; and (iii) the Fed also operates the payment system, in which banks transfer funds among themselves through the Federal Reserve Banks; in the process, the Fed takes the risk of transferring good funds intra-day on behalf of a bank that may fail before final settlement occurs at day’s end; regulation of banks, again, protects against this risk. But when considered individually, none of these turns out to be a sound rationale for government regulation.
Deposit Insurance. Although the Federal Deposit Insurance Corporation administers what is nominally a government insurance system, it is not correct to say that either the government or the taxpayers are at risk for bailing out depositors at insured banks or S&Ls that fail. Since the adoption of FDICIA in 1991, the capital of the banking system as a whole has backed the deposits in insured banks. This is because the act authorized the FDIC to require an increase in bank and S&L deposit insurance premiums whenever the amount in the Bank Insurance Fund (BIF) or the Savings Association Insurance Fund (SAIF) falls below 1.25 percent of total insured deposits. Under prior law, the FDIC and its S&L counterpart were required to pay deposit insurance out of accumulated funds that were created by an annual fixed premium on insured institutions. If the total losses of the fund ever exceeded the amount accumulated through premiums, the U.S. government--and hence the taxpayers--had to make up the difference. This is exactly what happened in the early 1990s, when the losses of the S&L industry were so substantial that they outstripped the accumulated amount in the insurance fund.
However, when Congress authorized the FDIC to increase premiums on banks and S&Ls--at any time and from time to time--to replenish the funds that back the promise of deposit insurance, the deposit insurance system was fundamentally changed. The FDIC’s authority to tax all insured institutions, whenever it needs funds to pay off the depositors of a failing bank or S&L, means that the capital of insured depository institutions--and not the resources of the U.S. government--has become the primary guarantor of the deposit insurance system. This amount--over $866 billion for banks and $186 billion for S&Ls--is more than sufficient to protect insured depositors under any conceivable set of adverse economic circumstances in the future. In any realistic sense, then, the U.S. government and the taxpayers are no longer liable for insuring deposits. Although it might be argued that Congress will step in anyway in the event of substantial losses among insured depository institutions, this is unlikely. The losses would have to be truly stupendous before the taxpayers would tolerate the use of their money in the place of bank and S&L capital that exceeds $1 trillion.
Under these circumstances, it seems clear that regulation of insured depository institutions is not necessary to protect the government or the taxpayers against losses from failed banks or S&Ls. The real parties in interest are insured depository institutions themselves, even though the FDIC is administering the deposit insurance system and most depositors believe that their deposits are protected by the federal government.
This raises a whole series of issues beyond the scope of this essay. For example, if all depository institutions are now responsible for the losses suffered by depositors in the few institutions that fail, should not banks and S&Ls have more of a say in the regulatory policies of the FDIC and how the BIF and SAIF funds are administered? If the government’s interest is only in assuring that small depositors have a safe place to deposit their funds, would it not be sufficient to require only that small deposits have preference over all other creditors in the event of a bank or S&L failure? The losses in almost all failures seldom exceed 10 percent of assets, leaving quite a lot with which to make small depositors whole before large depositors receive anything. To the extent that some degree of moral hazard still results from a deposit insurance system that is government-administered (although backed by the capital of banks and S&Ls), could that not be overcome by a mandatory subordinated debt structure such as that recommended by the Shadow Financial Regulatory Committee in March 2000?
A structure in which the FDIC makes the policies but depository institutions bear the losses is probably inherently unstable over the long term. Over time, insured depository institutions, which have the incentives to protect themselves against the failure of weak insured institutions, should take over the policymaking function from the FDIC, even though that agency might continue to administer a compensation system for small depositors. However this ultimately comes out, since FDICIA there has been no sound policy basis for the federal government to protect itself by regulation against losses it might suffer through the deposit insurance system.
The Fed as Lender of Last Resort. From its inception in 1913, the Federal Reserve System was structured as a lender of last resort for member banks. Later, legislation authorized the Fed to lend to all insured depository institutions. In this capacity, the Fed was to provide liquidity assistance to banks that were encountering liquidity difficulties but were still solvent institutions. Of course, it is difficult to determine with any certainty whether a bank that is suffering from liquidity stress is solvent at that moment, so to reduce its risks the Fed has always required good collateral for this lending activity. Under these circumstances, the Fed’s risks were virtually nonexistent. If the borrowing bank ultimately failed, the Fed had collateral to cover its unpaid advances.
When a federal deposit insurance system was adopted in 1933, the Fed’s policies did not change, but the consequences of these policies were potentially very different. If, after receiving liquidity assistance from the Fed, a bank ultimately failed, it was no longer the bank’s depositors who suffered losses. Now, the FDIC, as deposit insurer, was the loser, and thus Fed liquidity assistance lending acquired a certain “beggar thy neighbor” character. Liquidity assistance provided by the Fed enabled uninsured depositors to withdraw their funds from failing banks, increasing the ratio of insured deposits in the failed bank and the FDIC’s ultimate costs of resolving the institution and paying off its depositors. Also, since the Fed took good assets as collateral to secure its liquidity advances--known as “discount window” lending because the Fed discounted short-term collateral offered by borrowers--that also left fewer good assets to reduce the FDIC’s losses when it had paid off the insured depositors.
When, in effect, FDICIA changed the ultimate guarantor of the deposit insurance system--substituting the capital of depository institutions for the U.S. government and the taxpayers--it changed substantially the government’s stake in how the Fed conducted its lender of last resort operations. Now, the liabilities and risks involved returned again to the private sector. Taking collateral for this lending, the Fed is fully protected against losses. If its lending prolongs the life of a failing bank so that the bank’s uninsured depositors could get out before its collapse, or if the Fed takes the good assets as collateral and leaves little for the FDIC, it is no longer particularly important from the government’s point of view.
Perhaps these policies should not be followed, but they do not provide a basis for government regulation because they do not implicate the government’s funds. Although the BIF or SAIF might be taxed by the Fed’s activities as a lender of last resort, the losses will ultimately fall on the class of depository institutions whose deposits are backed by either of those funds. These institutions may have to pay higher deposit insurance premiums, but neither the government’s resources nor the taxpayers will be adversely affected. Again, the government has no financial stake in the outcome, and thus has no policy basis for protecting itself by regulating banks and S&Ls. And again, this situation suggests that insured depository institutions, and not the Fed or the FDIC, are the ones that should be making the policies in this area, since they are the ones that will ultimately suffer the losses.
Indeed, the framers of FDICIA seem to have recognized this fact, requiring that if a large bank is bailed out because it is deemed “too big to fail,” the costs of the bailout will be borne primarily by other large banks, and not by the smaller institutions that also contribute to the BIF or SAIF. This provision recognizes that there is an inverse relationship between the enthusiasm of depository institutions for a government bailout and the likelihood that the costs will fall on them. The effects of this rule are also salutary. It reduces the tendency of otherwise healthy large banks to lobby the Fed and the other regulators for a bailout of a large bank that has become insolvent, and it makes uninsured depositors--recognizing the diminished likelihood of a bailout--more likely to pay attention to and exert market discipline with respect to the financial condition and risks of large depository institutions.
For good measure, FDICIA also requires that the secretary of the Treasury (in consultation with the president) and two-thirds of the members of both the Fed board of governors and the board of directors of the FDIC approve a bailout of a depository institution deemed too-big-to-fail. But this is largely icing on the cake; political officials, fearful of a systemic effect from the failure of a large bank on their watch, are much more likely to approve a bailout than the banks that will have to pay for it. In any event, once the losses fall on the private sector rather than the taxpayers, the rationale for government regulation--that it is intended to protect the Fed, the FDIC, or the taxpayers--wholly disappears.
The Fed and the Payment System. The final rationale for regulating banks is the U.S. payment system, which is operated by the Fed. Here, the Fed does take risks on the financial condition of participating depository institutions, although it has made efforts to reduce its exposure.
The Fed operates several different payment systems: (i) the clearing of checks through physical delivery to paying banks; (ii) a fully automated clearing house system for electronic payments among banks; (iii) credit and debit card payments; and (iv) a large-value funds transfer system known as Fedwire. The first three systems settle at the end of the day and the Fed incurs no liability. If a depository institution in the system fails and cannot meet its obligations to others when settlements occur, the transactions involved are unwound and those institutions that made payments to customers but did not receive expected reimbursement from the failed institution suffer losses. As a practical matter, these losses--although potentially large in gross terms--are spread widely throughout the banking system and are unlikely to threaten the financial condition of any depository institution that is not reimbursed.
Moreover, it is unlikely that those institutions not reimbursed have advanced funds that they cannot recover. Bank customers are familiar with how banks protect themselves in cases like this through the delay customers experience before they can obtain cash for a deposited check. What is happening during this period is that the check is being presented to the bank on which it was drawn. If the bank honors the check, it has “cleared.” If, however, that drawee bank fails and cannot honor the check, the customer who deposited the check will not receive credit from the bank in which the deposit was made and will not be able to withdraw an equivalent amount in cash. The customer, and probably thousands more who were depositors in the failed bank, might suffer losses if uninsured, but the bank in which the check was deposited does not. Neither, in this case, does the Fed.
However, the same is not true in the large-dollar Fedwire payment system. That system is a real-time gross settlement system in which the Fed makes a payment to a recipient bank at the request of a payor bank, without knowing whether the payor bank will be able to cover the advance when final settlement occurs at the end of the day. If the Fed’s advance occurs at a time when the payor bank has a net debit position with respect to the Fed (i.e., it has paid out more than it has received), the Fed’s advance is known as a “daylight overdraft.” Nevertheless, once the Fed makes the payment, it is considered final and non-recoverable from the payee, and represents a potential loss to the Fed if the payor bank cannot fully settle. The rationale for this system--and the reason the Fed takes this risk--is the need for finality. Recipient banks are assured that when they receive payment through Fedwire they can make payments to their customers immediately, without fear that the failure of the payor bank will mean that they will not receive funds to cover the disbursement. Clearly, the involvement of the Fed as the guarantor of these large-value payments promotes stability, efficiency, and speed in the settlement of transactions.
An example of the importance of finality is the closing of a home purchase. In that case, the buyer’s bank (which may be in Omaha) in effect directs the Fed to deliver funds to the seller or the seller’s bank (which may be in New York) in exchange for a deed delivered to the buyer; the deed is then recorded as a transfer of title. If the buyer’s bank fails at the end of the day and cannot make good on the funds it directed the Fed to deliver, the Fed suffers a loss which it may not be able to recover. However, the buyer and seller are not affected; their transaction is final.
To be sure, the Fed takes steps to limit its risks in Fedwire by imposing net debit caps (the amount any bank can be indebted to the Fed for daylight overdrafts) and fees for overdrafts. Net debit caps are established for each participant in the system, and as settlement proceeds in real time all participants must remain within these caps, which in effect place a limit on the Fed’s exposure to a participant’s failure. The overdraft fees that the Fed imposes also tend to reduce the incentive for overdrafting. Neither of these measures, however, completely eliminates the risk that the Fed faces in operating Fedwire. Arguably, then, in order to obtain the finality benefits of Fedwire, regulation of banks and other depository institutions may be necessary to reduce the risk the Fed assumes under Fedwire, which falls on the government and ultimately on the taxpayers.
But the question here is whether the same payment finality can be achieved without the participation of the Fed, or--even if the Fed continues to operate the payment system--without requiring the Fed to take any substantial risk. If so, the Fed and the U.S. government are volunteering to assume risks that are not necessary for the safe and efficient operation of the payment system. Indeed, many private payment systems and at least one governmental system have developed mechanisms that permit the achievement of payment finality without government involvement and without the risk of loss to intermediaries that have roles similar to that of the Fed in the payment system.
Among the simplest of these systems is the clearinghouse, which acts as an intermediary-counterparty for its members. In this position, it will buy from one and sell to another, with the seller never having to weigh the buyer’s creditworthiness. The clearinghouse in turn protects itself against the associated credit risk by requiring the buyer to post collateral sufficient to cover the net exposure of the clearinghouse--i.e., the difference between what the buyer has promised to pay to the clearinghouse and what the clearinghouse will receive from other members for the account of the buyer when the latter has acted as a seller. A somewhat more sophisticated but similar system is represented by the Derivatives Products Company or DPC. These entities are frequently owned by a single sponsor and are paid for providing intermediary services in markets, such as derivatives markets, where there is no formal exchange and no formal membership. The DPC limits its risks and maintains a triple-A rating through a combination of hedging, capitalization, marking-to-market, and collateralization by its sponsoring organization. Both structures offer finality of payment.
A more sophisticated mechanism is the Clearinghouse Interbank Payments System, known as CHIPS, owned by a large number of financial institutions and countries and operated by the New York Clearinghouse Association, an association of New York banks. CHIPS is a rival of the Fed for sheer size, handling in the first half of 2005 about 275,000 payments per day with a daily dollar aggregate average value of more than $1.3 trillion. In contrast, Fedwire during the same period handled approximately 511,000 transactions per day with an average value of more than $2 trillion.
Since January 2001, CHIPS has operated an intra-day multilateral payment system that creates settlement finality without credit risk to CHIPS itself. Under the CHIPS system, a participant pre-funds its account at the beginning of the business day. Its intra-day payments are then added to a queue when made and are released when funds are or become available from two sources: (i) the pre-funding amount, and (ii) payments from other participants that come in to the CHIPS system for the paying participant’s account. Any funds released from the queue are final and irrevocable, but can be revoked any time before release. At the end of the day, if there are still payments left in the queue, the paying participants must either fund these payments, allowing them to be released, or withdraw them. If they are withdrawn, the intended recipients are immediately notified. Although this system seems to sacrifice speed to achieve finality--because payments are held in a queue unless pre-funded or funded by incoming payments--most CHIPS payments clear in 15 seconds.
The Bank of Canada also participates in a large dollar real time payment system known as the Large Value Transfer System (LTVS) that produces payment finality with virtually no risk to the government. Developed and administered by the Canadian Payments Association (CPA), the system relies on a special collateralization structure to eliminate risk. At the beginning of every business day, each participant in the system states the amount of credit exposure it is willing to accept, as a recipient, from each of the other participants. This aggregate amount, multiplied by 24 percent, then establishes every participant’s net debit cap as a sender. All participants must then post collateral with the Bank of Canada equal to their net debit cap and remain under this cap throughout the day. The net debit cap is increased by payments received during the day and reduced by payments sent. Before any payment may be made, it must be cleared against the payer’s net debit cap, and is refused if the cap would be breached. Payments that are cleared intra-day are final and irrevocable.
At the end of the day, settlement occurs, but since no participant is able to exceed its net debit cap--which in turn is collateralized--no losses can occur to other participants if one participant fails at day’s end. Although the CPA system is designed to prevent any losses that may arise out of the failure of one bank, the CPA and the Bank of Canada believe that the chance of a failure by more than one bank is remote: “The Bank of Canada has agreed to guarantee settlement in the extremely unlikely circumstance that more than one LVTS participant were to fail on the same day, during LTVS operating hours, and the sum of exposures of failed institutions were to exceed the collateral pledge to support their positions.”
The CHIPS and CPA systems both demonstrate that it is possible to achieve payment finality without government exposure to loss. If the United States were to adopt this system, the Fed would no longer be exposed to potential losses on daylight overdrafts. The last remaining policy reason for regulating banks would be removed.
* * * *
The answer, then, to the question “Why do we regulate banks?” is that we do so because we want to, not because we must.
Peter J. Wallison is a resident fellow at AEI.
1. Federal Deposit Insurance Corporation, Managing the Crisis: The FDIC and RTC Experience 1980–1994 (Washington, D.C.: FDIC, 1998), 4–5, available at www.fdic.gov/bank/historical/managing/contents.pdf.
2. Charles W. Calomiris, U.S. Bank Deregulation in Historical Perspective (Cambridge: Cambridge University Press, 2000), 22–28.
3. Ibid., 25–27.
4. Ibid., xx–xxi.
5. Ibid., 14–19, 24.
6. In a comprehensive survey published by AEI in 1999, George J. Benston explored the basis for regulation of banks and other financial services providers, and identified several other reasons for regulating banks, including benefits to the government, to legislators and to the institutions themselves. Regulating Financial Markets: A Critique and Some Proposals (Washington, D.C.: AEI Press, 1999), available at www.aei.org/publication218. Benston concludes that the only regulation required for banks is minimum capital. There are indeed additional political economy reasons for regulating banks, but those discussed in this Financial Services Outlook are the reasons cited in the ongoing debate over this policy.
7. Federal Deposit Insurance Corporation, Quarterly Banking Profile: First Quarter 2005, 5.
8. Reforming Bank Capital Regulation: A Proposal by the U. S. Shadow Financial Regulatory Committee (Washington, D.C.: AEI Press, 2000); statement no. 160 of the Shadow Financial Regulatory Committee, March 2, 2000, available at www.aei.org/
9. George G. Kaufman, “Do Lender of Last Resort Operations Require Bank Regulation?” (paper, “Is Bank Regulation Necessary?” AEI conference, October 27, 1999), 5, available at www.aei.org/event1062.
10. George J.Benston and George G. Kaufman, “Deposit Insurance Reform in the FDIC Improvement Act: The Experience to Date,” Economic Perspectives (Federal Reserve Bank of Chicago) 22, no. 2:2–20 (1998): 10. See 12 U.S. Code 1823 (1998) (G)(ii): “Repayment of loss.--The [FDIC] shall recover the loss to the appropriate insurance fund arising from any action taken or assistance provided [because an institution was considered too-big-to-fail] expeditiously from 1 or more emergency special assessments [on other members of the same insurance fund as the institution that was bailed out] . . . equal to the product of (I) an assessment rate established by the [FDIC]; and (II) the amount of each member’s average [net assets].” This language, which bases the recovery amount on net assets rather than deposits, falls disproportionately on large institutions.
11. A more complete discussion of these mechanisms is contained in Randall S. Kroszner, “Can Financial Markets Privately Regulate Risk in Payments and Clearing Systems?” (paper, “Is Bank Regulation Necessary?” AEI conference, October 27, 1999), available at www.aei.org/event1062.
12. Clearinghouse Interbank Payment System, Annual Stats, June 2005, available at www.chips.org/annual_stats.php.
13. The Federal Reserve Board, Fedwire Funds Service, last updated June 1, 2005, available at www.federalreserve.gov/PaymentSystems/Fedwire/fedwirefundstrfqtr.htm. The size of CHIPS in relation to Fedwire is particularly remarkable, given the fact that CHIPS only transacts with its fifty-four member banks and twenty-two member countries, while the Fedwire data reflects transactions by the Fed involving thousands of banks and other depository institutions.
14. Breffni McGuire, “Delivering Payment Value Online: CHIPS Ventures Into Web-Based Management Services,” Tower Group Viewpoint, January 2004, 3.
15. Canadian Payments Association, “LVTS Risk Controls,” www.cdnpay.ca/systems/lvts_controls.asp.