The management of liquidity risk is central to banking. Banks fund themselves in large part with deposits redeemable on demand, and they invest in illiquid loans. An important component of bank examination therefore has always been assessing how well a bank is managing that risk, including ensuring that the bank has robust plans for dealing with adverse liquidity contingencies.
The nature of those assessments underwent a sea change in the years after the Global Financial Crisis (GFC). In particular, instead of viewing the supply of liquidity broadly—including drawing on committed lines of credit from financial institutions and the central bank, for example—liquidity regulations were amended to define the supply of liquidity exclusively as the bank’s stock of high-quality liquid assets (HQLA).[1] These consist almost entirely of a bank’s holdings of reserve balances (deposits at a Federal Reserve Bank), Treasury securities and, to a lesser extent, agency (that is, mostly Fannie Mae- or Freddie Mac-guaranteed) mortgage-backed securities. The change had several motivations: primarily preventing a freeze in short-term funding markets from creating financial instability, but also a desire to reduce interconnectedness.
Overall, however, this change has not been for the better. To be clear, banks’ liquidity profiles have strengthened, but the focus on HQLA as the sole supply of contingent liquidity has produced three bad consequences. First, as discussed in the “Consequences” section, banks have had to commit more of their balance sheets to HQLA rather than loans to businesses and households, reducing economic activity and employment, and pushing credit intermediation into the nonbank sector. (Most people would be surprised to know that the eight largest bank holding companies now hold 25 percent of their assets as HQLA.) Second, the narrow focus on HQLA has left banks unwilling to use their HQLA in times of economic stress when it is needed to support the economy lest the banks fall below a minimum regulatory standard or expectation. Third, the increased demand for reserve balances has contributed to an upward spiral in the size of the Federal Reserve.
Liquidity regulation doesn’t have to be this way – the gains from having more liquid banks can be preserved while reducing social costs considerably. Returning to a broader conception of the supply of contingent liquidity, with an emphasis on maintaining diversified sources of funding, can achieve the same benefits with fewer costs. In particular—subject to stringent requirements, appropriate haircuts and specific circumstances—liquidity regulation should permit banks to assume that, during periods of stress, they can source funds from other financial institutions and the central bank. Such changes could create a virtuous spiral. The current reliance on central bank reserves for liquid assets would decline, and interbank markets would recover. This would increase financial market liquidity broadly and restore the market discipline imposed on banks by having to maintain access to external sources of funding. At the same time, without sacrificing liquidity, banks could reduce the share of their balance sheets devoted to unusable and unproductive stockpiles of HQLA and increase the share committed to providing credit to businesses and households.
That said, there are no easy answers. As a key first step, the prudential U.S. banking agencies, and the Basel Committee, should conduct a “holistic review” of liquidity requirements – similar to how the Fed is currently conducting a holistic review of capital requirements.[2] The review should consider whether broadening the definition of bank liquidity and shifting the focus away from simple metrics towards banks’ existing internal liquidity stress tests would be prudent and beneficial.
This note is in two parts. First, we review how bank liquidity was conceived before the Global Financial Crisis, how it is conceived now and what caused the change. Then we propose a holistic review of liquidity assessments, as well as some specific adjustments that would reduce the social cost of ensuring appropriate levels of liquidity without reducing safety and soundness or financial stability.
Pre-GFC Views of Liquidity
Before the GFC, the keystone to a bank’s liquidity was maintaining well-diversified and reliable sources of funding. The Federal Reserve’s Commercial Bank Examination Manual from 2003 stated:
A fundamental principle in designing contingency plans for each of these liquidity tenors is to ensure adequate diversification in the potential sources of funds that could be used to provide liquidity. Such diversification should not only focus on the number of potential funds providers, but on the underlying stability, availability, and flexibility of funds sources in the context of the type of liquidity event they are expected to address.[3] [Emphasis added]
In contrast, relying on assets as a source of liquidity was viewed somewhat negatively and primarily the approach of smaller banks.
Because many banks (primarily the smaller ones) tend to have little influence on the size of their total liabilities, liquid assets enable a bank to provide funds to satisfy increased loan demand. . . . However assets that are often assumed to be liquid are sometimes difficult to liquidate. . . . Furthermore, the holding of liquid assets for liquidity purposes is less attractive because of their thin profit spreads. . . . The number of banks relying solely on manipulation of the asset structure to meet liquidity needs is declining rapidly.[4]
Even 50 years ago, bank examiners judged an overreliance on assets as a source of liquidity, rather than lines of credit with correspondent banks, as problematic and lazy. This is clearly indicated in this selection from an examination report from the 1960s (note the acerbic “pressing demands of daily activities”).
The bank, not having borrowed [from the discount window] for several years, maintains at all times an apparently excessive amount of liquid assets. This excessively liquid position is not the result of board policy, but stems rather from an apparent total absence of effort on the part of management to employ profitably all available funds. In 1965, the reserve account balance was in excess of that required by a daily average of approximately $125,000. Because of the pressing demands of daily activities, the management admittedly maintains a “safe cushion” in the reserve account so that it will not be forced to make a daily calculation of the requirement. The regular offers of a correspondent bank to purchase excess Federal funds are always refused because the reserve position is unknown. In addition, the bank sold to a correspondent bank mortgage participations aggregating about $300,000 on June 1, 1965. As of examination date, about $100,000 of the proceeds of this transaction had not been reinvested and remained with the correspondent bank. The elimination of these excess balances would leave the bank still in a highly liquid position, with about 20 percent of the investment account in Treasury bills. As a standby liquidity protection, management has an open commitment from the correspondent bank to purchase an unlimited amount of the bank’s mortgage portfolio. The loss of potential earnings inherent in a situation such as this was discussed with management.[5]
The 2003 examination handbook also included a section extolling the virtues of the Fed’s discount window as a potential source of contingency liquidity.
The Federal Reserve’s primary credit program (discount window) offers depository institutions an additional source of available funds (at a rate above the target federal funds rate) for managing short-term liquidity risk. . . . Management may find it appropriate to incorporate the availability of the primary credit program into their institution’s policies, procedures, and contingency plans.[6]
The ability to borrow in the interbank market was considered a strength rather than a weakness of liquidity management. Each day, the Federal Reserve offered a finite supply of reserve balances calibrated to just meet the industries’ aggregate demand for reserves. Because the Fed did not pay interest on reserves, all banks sought to keep excess reserves as low as possible.[7] If one bank had insufficient reserves, then another almost always had an excess, and the reserves were redistributed at the end of the day in the federal funds market. As a result, each bank did not need to maintain a supply of reserve balances necessary to meet all of its idiosyncratic liquidity shortfalls with high certainty; those needs were diversified across the banking system. If the system as a whole was short of liquidity on any particular day, one or several banks would borrow from the discount window, creating the necessary reserves.[8]
The value of diversifying liquidity risks in the interbank market is recognized in the economic literature as well. Bhattacharya and Gale (1987) showed that the interbank market could fix banks’ susceptibility to failures as a result of self-fulfilling depositor runs, described by Diamond and Dybvig (1983). If banks not experiencing a run can lend to banks going through one in the interbank market, the risk from runs can be eliminated. Wilson Ervin (2017) presents a numerical example to analyze the benefit of diversification within a bank. He considers an international bank with four subsidiaries in multiple jurisdictions and then compared different resource sharing scenarios. Ervin finds under a plausible calibration that the probability of failure is five times higher if the institution is ring-fenced (i.e., cannot distribute excess resources to subsidiaries in need), compared to a scenario where excess resources can be shared. While his analysis focused on capital sharing, the logic and implications are similar for other critical resources, such as liquidity.
Bank regulators recognized the value of maintaining good access to the interbank market. For example, the Basel Committee’s Sound Practices for Managing Liquidity in Banking Organisations (2000) stated:
Senior management needs to ensure that market access is being actively managed by the appropriate staff within the bank. Relationships might exist with trading counterparties, correspondent banks, corporate customers and payments systems. Building strong relationships with key providers of funding can provide a line of defence in a liquidity problem and form an integral part of a bank’s liquidity management. The frequency of contact and the frequency of use of a funding source are two possible indicators of the strength of a funding relationship.[9]
In addition, being able to access funding in the interbank market acted as a source of discipline on banks. Reduced or lost access was an early warning that something was going wrong.
Federal Reserve Views on Oversupplying Reserve Balances on the Cusp of the GFC
When the Federal Reserve was granted the ability to pay interest on reserve balances, the Federal Open Market Committee asked staff to prepare a study on how to use the new authority to implement monetary policy. The study, Interest on Reserves: A Preliminary Analysis of Basic Options, was published in April 2008. When the study was prepared, the Fed’s authority to pay interest on reserves was scheduled to begin in 2011, but the onset of the GFC led Congress to accelerate the start of the authority to October 2008.
This study considered five different ways the Fed could implement monetary policy using its new authority. One of the options, “Floor with High Balances,” involved the Fed providing more reserve balances than banks would demand to satisfy reserve requirements, meet clearing needs and avoid overnight overdrafts. This would then push the fed funds rate down near the interest rate that the Fed would pay on reserve balances.
The Floor with High Balances approach is essentially how the Fed ended up conducting monetary policy starting in October 2008, when it began to borrow heavily from banks in the form of reserve balances to fund its expanding emergency lending facilities.[10] The Fed continued to conduct policy using a floor system as successive rounds of QE increased the quantity of reserve balances further. It then officially adopted the floor system implementation approach in January 2019 (see “Statement Regarding Monetary Policy Implementation and Balance Sheet Normalization”).
When the staff evaluated a floor system in 2008, they envisioned a much more modest floor system than what transpired. The approach was expected to require about $35 billion in reserve balances, and the interest rate on reserve balances was expected to be 10 to 15 basis points below the fed funds rate. The Fed currently estimates that $2.3 trillion in reserve balances are necessary to implement a floor system. Except for October 2018 to September 2019, the interest rate on reserve balances has generally been above the fed funds rate.[11]
Nevertheless, staff raised several serious and prescient concerns about even this modest floor system. One concern was that reserve balances would become attractive for banks, resulting in extremely high levels of reserve balances and hoarding during periods of stress (as in September 2019).
One question that arises is the extent to which this option creates a highly desirable risk-free asset for individual banks or the system as a whole. For example, there may be perverse outcomes in which an individual institution might find Fed account balances to be a very attractive asset and choose to hold a very large quantity of balances rather than lend in the market. Banks might see Fed account balances as largely superior to holding Treasury bills and other very short-term assets, potentially creating upward pressure on Treasury yields. A particular risk arises when there is a sudden change in the financial climate that makes banks want to increase their hoarding of reserves, leaving other banks deficient despite a very large supply of reserves in the aggregate. [p. 35]
Another concern was that oversupplying reserve balances would substantially reduce the size of the federal funds market and lead to a degradation of the efficiency of interbank markets.
Depository institutions would have little incentive or need to manage their balances on a day-to-day basis, so federal funds trading could diminish substantially. [p. 9]
A significant question mark in evaluating potential implications of option 4 for payment is related to the possible impact of these options on the efficiency of interbank markets. The usual incentives for banks to trade reserves—the desire to avoid penalties for reserve deficiencies and overnight overdrafts and the opportunity costs associated with holding non-interest bearing excess reserves—would be much attenuated. . . . With little trading activity and the funds rate mostly pegged very close to the target, the role for federal funds brokers could be considerably diminished. This possible degradation in the efficiency of the funds market could also have some corresponding impact on the efficiency of the payment system. [p. 38]
Perhaps surprisingly, the fact that banks would not need to manage their liquidity situation carefully was seen as a benefit rather than a cost (in striking contrast to the 1960s examiner report quoted above).
With ample balances earning a rate of return equal to the risk-adjusted federal funds rate, depository institutions would have little need or incentive to manage their balances carefully and could reduce the resources they devote to that activity. [p. 10]
GFC and Beyond
In reaction to the Global Financial Crisis, supervisory views of appropriate liquidity risk management shifted toward maintaining a large on-balance-sheet stock of HQLA. For example, the Basel Committee issued new Principles for Sound Liquidity Risk Management and Supervision in September 2008. Principle 1, identified as the (new) “Fundamental principle” states:
A bank should establish a robust liquidity risk management framework that ensures it maintains sufficient liquidity, including a cushion of unencumbered, high quality liquid assets, to withstand a range of stress events, including those involving the loss or impairment of both unsecured and secured funding sources. [p. 3]
Consistent with this principle, the liquidity coverage ratio—the centerpiece of the Basel III standard for liquidity—defines the supply of liquidity, the numerator of the ratio, as deposits at the central bank, government securities, agency securities, and, with a steep haircut, a few other types of securities. In practice, banks hold mostly reserve balances, Treasuries, and agency MBS as their HQLA. For the 20 U.S. banks that filed LCR disclosures in 2022Q3, Level 1 HQLA (the sum of Treasury securities and reserve balances) made up 86.7 percent of total HQLA, Level 2A HQLA (agency debt and agency MBS) made up 13.1 percent and Level 2B HQLA (everything else) made up only 0.2 percent.
The LCR requires banks to have HQLA sufficient to withstand 30 days of idiosyncratic and systemic stress. Banks are not allowed to assume that they draw on any lines of credit, borrow additional funds in the money market, borrow from central bank, or borrow from FHLBs. The Basel LCR standard justifies the exclusion of draws on credit lines as primarily to avoid contagion from interconnectedness:
No credit facilities, liquidity facilities or other contingent funding facilities that the bank holds at other institutions for its own purposes are assumed to be able to be drawn. Such facilities receive a 0% inflow rate, meaning that this scenario does not consider inflows from committed credit or liquidity facilities. This is to reduce the contagion risk of liquidity shortages at one bank causing shortages at other banks and to reflect the risk that other banks may not be in a position to honor credit facilities, or may decide to incur the legal and reputational risk involved in not honoring the commitment, in order to conserve their own liquidity or reduce their exposure to that bank.[12]
As we will discuss, the LCR does allow central banks to offer committed lines of credit that count as part of HQLA, but only on terms that make the committed liquidity facilities (CLFs) unworkable, except in jurisdictions that have an insufficient supply of government debt to meet banks’ HQLA needs such as Australia. With the exception of the unworkable CLFs, borrowing capacity from collateral pledged to the central bank does not count as HQLA. Although the LCR compares HQLA to net cash outflows that include an estimate of cash inflows, the inflows only consist of payments from maturing loans. And even those are limited by the standardized LCR assumptions.
The less important net stable funding ratio requirement (roughly designed to ensure the liquidity of a bank over a year of stress) assumes that a bank is unable to draw on any lines of credit, borrow from the central bank or FHLBs, or borrow in the interbank market.
In the United States, larger banks are required under Regulation YY to perform internal liquidity stress tests at least monthly across several time horizons. These liquidity stress tests must include (1) a scenario with adverse market conditions; (2) a scenario with an idiosyncratic stress event; (3) a scenario with combined market and idiosyncratic stress; and (4) any other appropriate scenario based on a BHC’s financial condition, size, complexity, risk profile, and scope of operations or activities. The precise details of internal liquidity stress test requirements are secret. Banks are permitted to make their own cash flow projection assumptions and their own determinations of what are considered “highly liquid assets.”
However, some banks have been subject to supervisory pressure to change those assumptions if they are not consistent with the LCR construct.[13] As a result, those banks have a reduced incentive to devote time and energy to modeling liquidity risks in so far as their professional judgement will be overridden by the calibration of the LCR. In addition, Reg YY states that banks cannot count on lines of credit for liquidity over the first 30 days of projected stress. It is a reasonable guess that the only permissible source of liquidity within the first 30 days is on-balance-sheet HQLA.
Resolution liquidity requirements include Resolution Liquidity Execution Need (RLEN) and Resolution Liquidity Adequacy and Positioning (RLAP). RLAP requires firms to estimate standalone liquidity needs for each material subsidiary over a minimum of 30 days of stress, and also ensure sufficient HQLA is either pre-positioned in the subsidiary or otherwise available at the parent to meet deficits. The parent must hold sufficient HQLA to cover the sum of all net liquidity deficits at every material subsidiary. RLEN requires firms to further account for the estimated liquidity needed post-bankruptcy filing to support the surviving or wind-down subsidiaries. This could lead to a requirement for even more liquid assets at the subsidiary level. The sizes of the liquidity requirements imposed by RLAP and RLEN are treated as confidential supervisory information. However, many large banks have reported that resolution liquidity requirements are the most binding constraint.
What Accounts for the Change in Perspective?
Several developments during and after the GFC contributed to the shift in perspective on how to measure a bank’s liquidity. The shift was also driven by LCR design considerations.
The GFC was characterized by widespread deterioration in the liquidity of financial markets. The markets for term money market funding were especially hard hit as financial markets pulled back from offering term funding, because of counterparty credit concerns and a desire to husband liquidity. For example, the spread between the 30-day Eurodollar rate and the 30-day Treasury bill rate (the TED spread), which is normally well below 50 basis points, rose sharply during the crisis, peaking at over 3 percentage points. Firms that wanted term-unsecured funding either paid exorbitant rates or couldn’t get it at all. Although markets for overnight funding continued to function, the liquidity risk of relying on such funding is extreme.
Securitization markets were equally hard hit. Pre-GFC supervisory manuals discuss securitizing loans as a way to raise liquidity under stress. During the crisis, the markets for asset-backed commercial paper, asset-backed securities of all types, collateralized loan obligations and private mortgage-backed securities all froze up, with new-issuance essentially falling to zero. Even agency MBS markets experienced significant strains. Rather than serving as a reliable source of contingent liquidity, securitization markets became a source of liquidity strains, since pipelines of loans intended for securitization needed to be funded.
During and after the crisis, central bank loans to banks were widely seen as bailouts, and therefore to be avoided. Indeed, for some, the entire purpose of a liquidity regulation was to prevent banks from borrowing from the central bank. For example, in January 2013, when the LCR was approved by the oversight body of the Basel Committee, Stefan Ingves, the head of the Committee, stated, “Today’s agreement is a clear commitment to ensure that banks hold sufficient liquid assets to prevent central banks becoming the ‘lender of first resort’.”
That view is odd for several reasons. Loans to banks were then the primary asset of the ECB, continuing the tradition of the Bundesbank. In addition, lending during periods of liquidity shortfalls is a normal and traditional function of the central bank. The Federal Reserve Act, after all, states that the purpose of the Federal Reserve is to create a “flexible currency.” And all the loans that the Federal Reserve made during the GFC were repaid, on time, with interest. Also note that the FHLB, which is a GSE just like Fannie and Freddie, is now the lender of first resort.
Largely as a consequence of this negative view of the discount window, the LCR, with one exception, does not permit borrowing capacity at the central bank to count as a source of liquidity. In the United States, for example, banks keep pools of collateral pledged to the Fed to maintain discount window borrowing capacity and cover potential daylight overdrafts. The Fed has emphasized that the discount window should be incorporated into banks’ liquidity plans and that discount window loans are available, no questions asked. Nevertheless, if a bank has pledged collateral to the Fed, thereby creating, say, $100 million in borrowing capacity, it may not count that $100 million as a source of liquidity (see “Against What Liquidity Risks should a Bank Self-insure?”). Borrowing from the central bank is also not included as a source of funding in the NSFR.
The exception in the international standard is “committed liquidity facilities”—guaranteed lines of credit from the central bank. From the outset, the LCR included borrowing capacity at CLFs as allowable HQLA in jurisdictions with insufficient government debt to satisfy banks’ HQLA needs. One reason why central bank borrowing capacity was not included in the LCR is that each central bank offers different terms on their loans that change over time, making it hard to include the capacity in a single international standard. CLFs fix that problem because the terms are standardized by the regulation. Indeed, the final official change to the LCR was including CLFs as an option for all jurisdictions. However, the standard requires the CLFs be provided at a fee so high that the facilities are not viable. No central banks in jurisdictions required to use the more onerous terms have offered CLFs to their banks.
What happened during the GFC also highlighted the potential risk of interconnectedness between financial institutions. For example, losses on Lehman paper led the Reserve Primary Fund to break the buck, and AIG’s extensive network of CDS swaps was a key reason the firm could not be allowed to fail. The designers of the LCR did not want the new requirement to lead to increased interconnectedness. Concern about interconnectedness is largely why what is universally called “cash”—deposits at a bank—is not included in the list of allowed HQLA.[14] Not wanting to promote interconnectedness is also one reason why lines of credit from financial institutions do not count as HQLA.
Another reason to exclude deposits at or lines of credit from other financial institutions was to avoid creating a situation where two or more institutions could boost each other’s liquidity metrics by offering each other equal and opposite financial products. For example, if bank A deposits $100 at bank B and bank B deposits $100 at bank A, and if the deposits counted as HQLA, each bank’s HQLA would go up. The change in each bank’s LCR depends on the assumption about how the deposits would behave under stress. In particular, if 100 percent of the deposits were assumed to be withdrawn, then there would be a cash outflow that exactly matched the added HQLA, so the LCR would be essentially unchanged. Any lower outflow assumption and the LCR would rise. Although preventing such outcomes may seem obviously desirable, as we will discuss, in reality such an arrangement would help each bank handle an idiosyncratic stress event.
The difficulty of creating a single international liquidity standard for measuring a bank’s liquidity situation was also a reason why the supply of liquidity was limited to a short list of assets held on the books of the bank. Objective, internationally consistent, criteria for external sources of liquidity are difficult to define. For example, each central bank has its own lending terms. At that point, the Fed only lent to offer backup funding for short periods, while the ECB offered ongoing funding. As a consequence, just because an asset was eligible as collateral for a central bank loan did not mean it could reliably secure 30 days of funding in every jurisdiction. Moreover, having reliable access to financial markets may not be something that can be measured with enough precision for a standard. In addition, many wanted a simple ratio similar to capital ratios. The supply of liquidity was therefore limited to HQLA, while cash inflows were included in the denominator as part of the calculation of net cash outflows and limited to a fraction of scheduled loan repayments, rather than any additional borrowing.
An external development that contributed to a focus on HQLA was the Fed’s payment of interest on reserves. Before 2008, the Fed paid no interest on reserves, because it did not have the legal authority to do so. Banks sought to minimize their holdings of reserve balances, and no examiner would expect a bank to maintain a stock of reserves to meet contingencies.
But once the Fed started paying a market rate on reserve balances, reserves became a viable asset to hold for liquidity needs. Indeed, as the Fed’s balance sheet swelled through QE 1, 2, 3 and 4, the quantity of reserve balances the banking system had to hold rose nearly dollar-for-dollar. Whereas banks maintained about $10 billion in reserve balances before the Global Financial Crisis, the banking system held $3¼ trillion in reserve balances in May 2020. As we noted, the quantity of reserve balances that the banking system had to absorb became so vast that the interest rate on reserve balances has usually been above the federal funds rate. With reserve balances so abundant and cheap, banks chose to use them to solve nearly all their liquidity problems, and examiners have come to expect nearly all liquidity problems to be solved with reserve balances.
It is worth noting that a lot of the lessons from the Global Financial Crisis and the risks of short-term wholesale funding were derived from the experience of non-banks like Bear Stearns, Lehman Brothers, and AIG. Liquidity regulations designed for non-banks must recognize that the institutions do not have access to discount window funding. Even if part of a bank holding company, there are significant restrictions on affiliates borrowing from the bank subsidiary. Any redesign of regulatory assessments of liquidity would need to recognize the difference funding source available to banks, non-banks and bank holding companies.
What are the Consequences?
The change in perspective, from strong liquidity meaning well-diversified sources of funds to meaning a stockpile of reserve balances and similar assets, has had several negative consequences. First and foremost, as discussed below, it has reduced economic growth and employment by curtailing the supply of credit to businesses and households. If each bank needs to hold enough liquid assets to meet all its potential liquidity needs with high certainty, the fraction of each bank’s balance sheet that is HQLA must be excessive. Indeed, at a certain point, banks are no longer engaging in liquidity transformation at all.
In 2005, liquid assets (defined as reserve balances, Treasury securities, agency debt, and agency MBS) were 10.6 percent of bank holding company assets. At the end of 2022, they were 23.6 percent. Because of the balance sheet constraints imposed by capital requirements, that increased share of liquid assets implies a reduced supply of loans to businesses and households.[15] Although some increase in holdings of liquid assets relative to pre-GFC levels was surely warranted, it is far from clear that a more-than-doubling to a quarter of banks’ balance sheets is socially optimal. The BIS estimates that compliance with the NSFR alone has reduced GDP permanently by 8 basis points (LEI 2010) and that at current levels of capital, the social benefit is less than the social cost (see Nelson and Covas 2017). A BIS review of the literature reported findings that compliance with the LCR reduces the supply of bank credit by 3 to 6 percent.
Moreover, it is unclear that the stockpiles of HQLA are serving much purpose. A debilitating problem with the LCR is that banks are unwilling to use their HQLA to meet liquidity needs if doing so reduces their LCR to near or below the regulatory standard of 100 percent. The usability problem has been a concern since the creation of the standard, which is why there is language in both the international standard and the U.S. implementing regulation emphasizing that it is OK to make use of HQLA.[16]
Nevertheless, banks do not appear willing to use their HQLA. The Bank of England released a discussion paper in March 2022 in which they sought feedback on how to fix the usability problem. They noted that during the Covid-19 stress, “. . . a range of banks in the UK and internationally took defensive actions to protect and bolster their liquidity positions.” In addition, during a liquidity exercise conducted by the BoE in 2019, “Banks’ submissions suggested that, on the whole, they were unwilling to allow their LCRs to fall below 100%, even in extremely severe stress, if they could prevent them from doing so.” Similarly, a BPI survey of bank treasurers found that none of the treasurers would be willing to use HQLA to meet a funding need if doing so put the bank below a regulatory standard. Moreover, several noted that they raised liquidity in March and April 2020 by selling illiquid assets or borrowing at term specifically to remain in compliance, actions that run contrary to the objective of the requirements. Thus, the LCR appears to be having exactly the opposite effect than intended.
Relatedly, the diminished use of the discount window, along with the examiner and bank management focus on discouraging use, has contributed to the severe stigma associated with borrowing from the central bank. The stigma has reduced the effectiveness of the window as a monetary policy tool and a crisis response tool (see here and here).
In addition, the demand by banks for reserve balances to meet liquidity needs has fostered a vastly larger Federal Reserve. As the Fed has expanded the supply of reserve balances through successive rounds of QE, banks and bank examiners have come to expect banks to use the reserves supplied to meet liquidity needs. Each new provision of supply sets a new level of expectations, making it difficult to reverse. Because the Fed needs to keep reserve supply abundant to conduct monetary policy under its current operating regime, it is forced to provide an ever-growing quantity to meet demand (see “Bank Examiner Preferences and Obstructing Monetary Policy”).
The BoE seems to have concerns that a focus on HQLA as the sole source of liquidity may be contributing to the usability problem. In a section titled “The Importance of Usable HQLA,” the discussion paper states:
When and how to use market sources of funding, existing HQLA stocks, or central bank liquidity insurance is a decision for banks to make, based on their expert knowledge of their own liquidity needs and the availability and cost of options. In particular, there is no fixed order in which the Bank and PRA expect banks to use one form of liquidity over another. The existence of central bank liquidity facilities should provide banks with the confidence needed to operate for a reasonable period below the level of liquidity that the Bank and PRA expect them to maintain in normal times.
Furthermore, as the Fed provides more and more reserves at an interest rate that is above the federal funds rate, the interbank market has withered (as expected by Fed staff in the 2008 study discussed earlier). In a 2013 New York Fed blog post “Who is Lending in the Fed Funds Market?”, Afonso and colleagues report that fed funds lending (and therefore borrowing) fell after the GFC to one-fourth of its pre-GFC level. They attribute the decline to the expansion of the Federal Reserve’s balance sheet and to the payment of interest by the Fed on reserve balances. Banks now hold such large amounts of extra reserves that there is no need to redistribute liquidity at the end of the day. Currently, the federal funds market consists almost entirely of loans from FHLBs (which have extra funds to invest and have accounts at the Fed but do not earn interest) to the U.S. branches of foreign banking organizations, who borrow the funds and deposit them at the Fed, earning the spread. FBOs are the main participants in these transactions, because it is slightly cheaper for them to do so than domestic commercial banks, owing to how deposit insurance premiums are calculated.
Even though the federal funds market is currently essentially gone, it can be restored. As described by Sriya Anbil and Mark Carlson (2019), the federal funds rate reemerged from dormancy in the 1950s as reserves balances fell from abundant to scarce.
As banks increasingly needed to manage their liquidity amid monetary tightening, trading in the federal funds market gradually re-emerged. Through this experience, the size and liquidity conditions in the federal funds market became self-reinforcing; as the market developed, it attracted additional participants that further enhanced its usefulness in managing liquidity. This historical experience may have implications for current monetary policy because it suggests that as the Federal Reserve reduces the supply of reserves through tightening monetary policy, and as the incentives to participate in the market return, interbank trading in the federal funds market is likely to eventually revive.
If idiosyncratic liquidity risk can be diversified across the banking system in the interbank market or through private-sector committed lines of credit, the on-balance-sheet supply of liquidity that each bank needs to hold can be reduced. Similarly, if banks can count on the central bank to offer liquidity in a systemic crisis, the necessary quantity of HQLA can be lowered further (see Carlson et al. 2015).
Ironically, although the liquidity requirements were designed in part to reduce the likelihood that a bank would need to borrow from its central bank in the rare financial crisis, the result has been that banks rely on the Fed and the FHLBs (which have the implicit backing of the taxpayer) for liquidity all the time. Furthermore, the ability to cheaply stockpile reserves has meant that banks are not subject to the market discipline associated with maintaining access to money market funding. In 2010, concern about the negative consequences of holding excess reserves led Norges Bank (the central bank of Norway) to switch from a system with abundant reserves to one with more scarce ones. When seeking comment on their decision, they noted:
When Norges Bank keeps reserves relatively high for a period, it appears that banks gradually adjust to this level. . . . With ever increasing reserves in the banking system, there is a risk that Norges Bank assumes functions that should be left to the market. It is not Norges Bank’s role to provide funding for banks. . . . If a bank has a deficit of reserves towards the end of the day, banks must be able to deal with this by trading in the interbank market.
The Norges Bank’s concern that an overabundance of inexpensive reserve balances results in banks reducing their liquidity management echoes the pre-GFC supervisory concerns about banks relying on assets as their sole source of liquidity in contingencies. The Norges Bank’s concern stands in sharp contrast to the Federal Reserve monetary policy staff’s view discussed earlier that a benefit of oversupplying reserves is reducing the resources that banks need to devote to liquidity management.
The exclusive focus on internal stores of liquidity extends beyond the LCR to the internal liquidity stress tests that banks are required to perform. As we noted, banks are not able to anticipate drawing on any external sources of liquidity for the first 30 days. Such an assumption is at odds with what a bank would actually do.
In addition, the insistence that ILSTs be calibrated to be at least as stringent as the inflow and outflow assumptions in the LCR reduces the incentive for banks to devote resources into preparing contingency plans. For example, an employee at a regional bank (a bank required to perform internal liquidity stress tests but not comply with the LCR) showed the initiative to research how long it would take the bank’s municipal depositors to do the necessary paperwork to shift their funds elsewhere, learning that it would take several months. But their examiner required the bank to assume the deposits would be withdrawn on the timeline assumed in the LCR for purposes of the bank’s ILST.
Need for a Holistic Review
The Federal Reserve is currently engaged in a holistic review of capital requirements. It is time for a similar review of liquidity requirements as well. Current liquidity requirements were developed to achieve several different objectives, in addition to keeping the probability of a liquidity default or liquidity crisis low. Many have questioned whether the requirements as currently designed are worth the cost. It would be beneficial to revisit the results to consider scope for improvement.
There are inherent challenges to designing liquidity regulations. Along with the issues associated with defining the supply of liquidity discussed in this note, liquidity requirements suffer from a catch-22: Liquidity reserves can’t inspire counterparty confidence unless the counterparty has the reserves, but the reserves aren’t useful unless the bank can use them. Similarly, the usability problem is inherent in any contingency plan. Using a contingency source of funding inevitably indicates something has gone wrong. Using the discount window is stigmatized for the same reason.[17]
Recognizing the magnitude of the challenge, here are some suggestions for consideration. First, it may be extremely helpful to focus on banks’ current internal liquidity stress tests rather than the LCR or NSFR as the measure of a bank’s liquidity situation. In the U.S., these tests appropriately cover a range of horizons and a range of circumstances. Within the tests, it would be useful to recognize multiple sources of liquidity that vary depending on the circumstances and require banks to have well-diversified sources of contingent liquidity.
Just as capital stress tests evaluate whether banks after a stress remain above a minimum level of capital far below the acceptable normal level, the monthly ILSTs could be used to ensure banks remain above a minimum level of cash under stress. That minimum level should be well below the level currently required by the LCR. The test would therefore implicitly define a buffer requirement and a minimum. However, the buffer could be satisfied by different sources of contingency funding, depending on the situation.
The scenarios could examine outcomes designed to reward well-diversified sources of liquidity. In particular, the supply could be adjusted to fit the situation considered in each scenario. For example, in a scenario involving only idiosyncratic risk, the supply might not include borrowing in the interbank market or the central bank but would include HQLA and committed lines of credit from other counterparties. In a scenario involving systemic risk, the supply might include borrowing from the central bank, a fraction of committed lines from other counterparties, and no HQLA except cash. Perhaps borrowing in the interbank market would never qualify.
A benefit of ILSTs compared with the LCR or NSFR is that the current financial condition of the bank is integrated into the test. Just as GSIB surcharges depend in part on reliance on wholesale funding, liquidity stress tests are inherently more stringent for less capitalized banks than for better capitalized ones. That characteristic of the test could be extended to the determination of whether external sources of funding would be available. For example, as discussed in “Against What Liquidity Risks Should a Bank Self-insure?”, only financially sound banks have access to primary credit (a/k/a “the discount window”), so a bank with a marginal condition now would be projected to lose access to the window in a scenario where its condition deteriorated further.
Another option would be to develop a set of characteristics for committed lines of credit from other financial institutions that could count as part of the supply of credit, so that such lines could be included in the international standard. Haircuts on lines, and assumed drawdown rates on lines offered to others, should vary depending on the scenario. That is, the liquidity regulation framework need not treat committed lines as a binary option, instead allowing the lines to count to the extent that they can be expected to perform. But standards should generally recognize that diversification of liquidity risk across the banking system is beneficial.
HQLA could also be made more usable. If focus is shifted toward ILSTs, then “usability” would correspond to a willingness to fail the test for a while, before restoring the supply of liquidity or diminishing the projected need. This would no doubt also be a tough sell to banks and to examiners. The Bank Policy Institute offered several possible changes to increase HQLA usability in its response to the BoE request for comment (here). These include making adjustments to increase LCRs in times of stress, granting banks credit for unused capacity at the discount window, and recognizing that the liabilities that remain when other liabilities run, are more likely to continue to remain at the bank.
Another potentially promising way to convince banks that it would be OK to draw on their sources of liquidity is to conduct exercises designed to get bank management and bank examiners used to such draws. In 2019, the Bank of England conducted such an exercise. The BoE stated that a primary purpose of the exercise was to increase usability:
Banks hold regulatory liquidity buffers that the FPC expects to be used in a stress bringing liquidity coverage ratios below 100%. The exercise will explore how the reactions of banks and authorities to the stress would shape its impact on the broader financial system and the U.K. economy.[18]
Although the results have not been published, the BoE has indicated that the exercise revealed insights that helped shape the Bank’s response to the impact of the COVID-19 pandemic and would inform the Bank’s future work.[19]
Conclusion
It is time for a holistic review of liquidity regulations. Compared with banking before the GFC, regulations and examiner guidance now see maintaining a large stock of on-balance-sheet HQLA as the fundamental principle for sound liquidity, rather than maintaining well-diversified and reliable sources of contingency funding. The economic costs of that arrangement are substantial, and there is evidence that the benefits are smaller. The change reflects lessons learned during the GFC, but also some ancillary developments, such as the Fed paying interest on reserve balances.
Although the improvement in bank liquidity over the past 15 years has been beneficial, the change in the design of the liquidity regulation framework has had several significant negative consequences. Most importantly, one-fourth of bank balance sheets is now HQLA rather than loans to businesses and households, and the reduced supply of bank credit has reduced economic activity and employment and pushed credit intermediation into the nonbank sector. In addition, the focus on HQLA has contributed to banks’ unwillingness to use their liquid assets in times of stress when their use is needed most. Instead, banks are selling assets into illiquid markets rather than falling below a regulatory standard—exactly the outcome the standards are intended to prevent.
Moreover, the reliance on reserve balances as HQLA has fed an upward spiral in the Fed’s balance sheet. Indeed, given the central importance of reserve balances and Treasury securities as HQLA, liquidity regulations as currently implemented in the United States depend on there being a massive and growing central bank and ongoing deficit spending by the federal government.
It remains unclear that the new liquidity regulation framework has produced benefits that exceed its costs. However, many potential reforms could preserve the gains while reducing those costs. In particular, under the right circumstances and with appropriate controls, the concept of liquidity supply could be expanded. It could include, subject to appropriate haircuts and stringent requirements, external sources of funding, such as committed lines of credit from other financial institutions or the central bank. Moreover, a review could continue to promote efforts to make banks more willing to use their HQLA, even if doing so put them below a regulatory standard for a period of time.
References
Afonso, Gara, Entz, Alex, and LeSueur, Eric. (2013, December 2). Who’s Lending in the Fed Funds Market? New York: Liberty Street Economics, Federal Reserve Bank of New York.
Who’s Lending in the Fed Funds Market? – Liberty Street Economics (newyorkfed.org)
Anbil, Sriya, and Carlson, Mark. (2019, March 22). The Re-emergence of the Federal Reserve Funds Market in the 1950s. Finance and Economics Discussion Series (FEDS) Notes. Washington: Board of Governors of the Federal Reserve System.
The Fed – The Re-emergence of the Federal Reserve Funds Market in the 1950s
Bank of England. (2019, July). Financial Stability Report: Issue No. 45. London: Author.Financial Stability Report. Issue No. 45 (bankofengland.co.uk)
Bank of England. (2021, February 3). The Bank of England will not be restarting the 2019 liquidity Biennial Exploratory Scenario (BES). Press Release.
The Bank of England will not be restarting the 2019 liquidity Biennial Exploratory Scenario (BES).
Bank of England and Prudential Regulatory Authority. (2022, March 31). Discussion Paper 1/22—The Prudential Liquidity Framework: Supporting Liquid Asset Usability. London: Author.
Discussion Paper 1/22—The Prudential Liquidity Framework: Supporting Liquid Asset Usability.
Bank Policy Institute. (2018, November 26). Understanding the Fed’s Implementation Framework Debate. Washington, DC: Author.Understanding_the_Fed’s_implementation_framework_debate_Review05.pdf (bpi.com)
Barr, Michael S. (2022, December 1). Why Bank Capital Matters. Speech at the American Enterprise Institute, Washington, DC. [virtual]. Transcript and Videolink.
Speech by Vice Chair for Supervision Barr on why bank capital matters – Federal Reserve Board
Basel Committee on Banking Supervision. (2000, February). Sound Practices for Managing Liquidity in Banking Organisations. Basel: Bank for International Settlements.
Basel Committee Publications – Sound Practices for Managing Liquidity in Banking Organisations – Feb 2000 (bis.org)
Basel Committee on Banking Supervision. (2010, August). An Assessment of the Long-Term Economic Impact of Stronger Capital and Liquidity Requirements. Basel: Bank for International Settlements.An Assessment of the Long-Term Economic Impact of Stronger Capital and Liquidity Requirements (bis.org)
Basel Committee on Banking Supervision, Bank for International Settlements. (2013, January 6). Group of Governors and Heads of Supervision endorses revised liquidity standard. Press Release.
Group of Governors and Heads of Supervision endorses revised liquidity standard for banks (bis.org)
Basel Committee on Banking Supervision (2013). Basel III: The Liquidity Coverage Ration and Liquidity Risk Monitoring Tools. Basel: Bank for International Settlements.
Basel III: The Liquidity Coverage Ratio and Liquidity Risk Monitoring Tools (bis.org)
Basel Committee on Banking Supervision. (2016, March). Working Paper No. 30: Literature Review on integration of Regulatory Capital and Liquidity Instruments. Basel: Bank for International Settlements.
Literature review on integration of regulatory capital and liquidity instruments (bis.org)
Board of Governors of the Federal Reserve. (1971). Reappraisal of the Federal Reserve Discount Mechanisms. Washington, DC: Author. https://fraser.stlouisfed.org/files/docs/historical/federal%20reserve%20history/discountmech/bog_reappraisal_discount_197108_vol1.pdf
Board of Governors of the Federal Reserve. (2003, November). Commercial Bank Examination Manual. Washington, DC: Author.
0312cbem.pdf (famguardian.org)
Board of Governors of the Federal Reserve. (2005). The Federal Reserve System: Purposes and Functions. Washington, DC: Author.
https://www.federalreserve.gov/aboutthefed/files/pf_1.pdf
Board of Governors of the Federal Reserve (2008). Interest on Reserves: A Preliminary Analysis of Basic Options. Washington, DC: Author.
https://www.federalreserve.gov/monetarypolicy/files/FOMC20080411memo01.pdf
Board of Governors of the Federal Reserve. (2019, January 30). Statement Regarding Monetary Policy Implementation and Balance Sheet Normalization. Washington, DC: Author.
https://www.federalreserve.gov/newsevents/pressreleases/monetary20190130c.htm
Board of Governors of the Federal Reserve. (2022, May). Open Market Operations During 2021. Washington, DC: Author.
https://www.newyorkfed.org/medialibrary/media/markets/omo/omo2021-pdf.pdf
Carlson, Mark, Duygan-Bump, Burcu, and Nelson, William (2015, February). Why Do We Need Both Liquidity Regulations and a Lender of Last Resort? A Perspective from Federal Reserve Lending during the 2007-09 U.S. Financial Crisis. Finance and Economics Discussion Series (FEDS) 2015-011. Washington: Board of Governors of the Federal Reserve System. http://dx.doi.org/10.17016/FEDS.2015.011.
Covas, Francisco, Collard, Katie, and Waxman, Brett (2022, August 9) The Reverse Repo Bank. Washington, DC: Bank Policy Institute.https://bpi.com/the-reverse-repo-bank/
Diamond, Douglas W., and Dybvig, Philip H. (1983, June). Bank runs, deposit insurance, and liquidity. Journal of Political Economy, University of Chicago Press, Vol. 91(3), pages 401–419.
Bank Runs, Deposit Insurance, and Liquidity | Journal of Political Economy: Vol 91, No 3 (uchicago.edu)
Ervin, D. Wilson. (2017, December 12). The Risky Business of Ring-Fencing. Available at SSRN: https://ssrn.com/abstract=3085649 or http://dx.doi.org/10.2139/ssrn.3085649
Gale, Douglas, and Bhattacharya, Sudipto. (1987). Preference shocks, liquidity and central bank policy. In W. Barnett and K. Singleton (Eds.), New Approaches to Monetary Economics. New York: Cambridge University Press.
Preference shocks, liquidity, and central bank policy (Chapter 4) – New Approaches to Monetary Economics (cambridge.org)
Nelson, Bill. (2017, June 7). Unlocking the Liquidity Coverage Ratio. Washington, DC: Bank Policy Institute.Unlocking the Liquidity Coverage Ratio – Bank Policy Institute (bpi.com)
Nelson, Bill. (2019, March 21). A Modest Change to the LCR that could Substantially Improve Financial Stability. Washington, DC: Bank Policy Institute.A Modest Change to the LCR That Could Substantially Improve Financial Stability – Bank Policy Institute (bpi.com)
Nelson, Bill. (2021, August 10). Discount Window Stigma: We Have Met the Enemy, and He Is Us. Washington, DC: Bank Policy Institute. Discount Window Stigma: We Have Met the Enemy, and He Is Us – Bank Policy Institute (bpi.com)
Nelson, Bill. (2022, September 19). Bank Examiner Preferences Are Obstructing Monetary Policy. Washington, DC: Bank Policy Institute. Bank Examiner Preferences Are Obstructing Monetary Policy – Bank Policy Institute (bpi.com)
Nelson, Bill. (2022, December 1). Against What Liquidity Risks Should a Bank Self-Insure? Washington, DC: Bank Policy Institute. Against What Liquidity Risks Should a Bank Self-insure? – Bank Policy Institute (bpi.com)
Nelson, Bill, and Covas, Francisco. (2017, March 27). Foundational Basel Committee Study Estimates the Costs of NSFR Exceed its Benefits by Nearly $1 Trillion. Washington, DC: Bank Policy Institute.Foundational Basel Committee Study Estimates the Costs of NSFR Exceed its Benefits by Nearly $1 Trillion – Bank Policy Institute (bpi.com)
Nelson, Bill, and Parkinson, Pat. (2022, January 25). A Major Limit on the Fed’s Crisis Toolkit: Shame. Washington, DC: Bank Policy Institute.A Major Limit on the Fed’s Crisis Toolkit: Shame – Bank Policy Institute (bpi.com)
Nelson, Bill, and Waxman, Brett. (2021, October 12). Bank Treasurers’ Views on Liquidity Requirements and the Discount Window. Washington, DC: Bank Policy Institute.Bank Treasurers’ Views on Liquidity Requirements and the Discount Window – Bank Policy Institute (bpi.com)
Waxman, Brett. (2022, June 30). BPI Comment Letter on Bank of England Discussion Paper 1/22. Washington, DC: Bank Policy Institute.BPI-Comment-Letter-Bank-of-England-Discussion-Paper-1-22-6.30.2022-Final.pdf
[1] To be clear, banks’ demand for liquidity is calculated as net cash outflows, projected outflows minus projected inflows which equal fractions of some maturing loans and securities. But the projected inflows do not include tapping external sources of funding maintained for liquidity contingencies.
[2] See “Why Bank Capital Matters,” remarks by Vice Chair Michael S. Barr at the American Enterprise Institute, December 1, 2022.
[3] Commercial Bank Examination Manual, November 2003, Section 4020.1, p. 2.
[4] Commercial Bank Examination Manual, November 2003, Section 4020.1, pp. 1-2.
[5] Reappraisal of the Federal Reserve Discount Mechanisms, 1971.
[6] Commercial Bank Examination Manual, November 2003, Section 4020.1, p. 2.
[7] See The Federal Reserve System, Purposes and Functions, 2005.
[8] Only a loan from the Fed can increase the total supply of reserves as opposed to, say, a loan from a Federal Home Loan Bank.
[9] Paragraph 58, page 13. https://www.bis.org/publ/bcbs69.pdf
[10] See “Understanding the Fed’s Implementation Framework Debate,” November 26, 2018.
https://bpi.com/wp-content/uploads/2018/11/Understanding_the_Fed%E2%80%99s_implementation_framework_debate_Review05.pdf
[11] The $2.3 trillion figure is the minimum level of reserve balances projected by the Federal Reserve Bank of New York in “Open Market Operations During 2021,” May 2022. https://www.newyorkfed.org/medialibrary/media/markets/omo/omo2021-pdf.pdf
[12] Paragraph 149. https://www.bis.org/publ/bcbs238.pdf
[13] See “Bank Treasurers’ Views on Liquidity Requirements and the Discount Window.”
[14] Banks are, however, allowed to include deposits at other banks (“cash”) as HQLA in their ILSTs. 12 CFR 252.35(b)(3)(i)(A).
[15] When a bank makes a loan or accepts a deposit, it is now required to hold additional HQLA as well. Because the bank must fund the HQLA in part with capital, and capital is expensive, providing credit to businesses and households and accepting deposits is more costly. The need to fund the HQLA in part with capital owes to leverage ratio requirements, stress tests, and GSIB surcharges, see Collard, Covas, and Waxman (2022) “The Reverse Repo Bank.”
[16] Indeed, the problem was recognized 101 years ago in 1922 by Dennis Robertson in his book Money. As Robertson wrote concerning the Fed’s reserve requirement: “it resembles the procedure of a certain municipality which tried to guard against a shortage of cabs by ordaining that there should always be at least one cab on the ranks.” Robertson also observed, “An iron ration which you must not touch even in the throes of starvation is something of mockery.”
[17] For a discussion of the relationship between the HQLA usability problem and the discount window stigma problem and some proposals to address the usability problem, see “Unlocking the Liquidity Coverage Ratio”. For another idea for increasing usability, see “A Modest Change to the LCR That Could Substantially Improve Financial Stability.”
[18] Financial Stability Report, Bank of England, July 2019, p. 53.
[19] Bank of England News Release, February 3, 2021.
