Central banks play a critical role in maintaining financial stability by lending to commercial banks, both in normal times and during periods of stress. However, the regulatory framework for limiting liquidity risk, including the Basel III liquidity coverage ratio (LCR), is premised on a view that central banks lending to banks should be rare. The Basel Committee’s emphasis on preventing central banks from becoming the “lender of first resort” has contributed to a broader issue: the stigma surrounding the use of central bank lending facilities, such as the Federal Reserve’s discount window, which ultimately undermines the stability these facilities are designed to support. This article explores the tension between regulatory approaches and central bank practices, arguing that bank regulators should align their expectations with clear and reliable lending commitments made by central banks, in order to enhance both bank safety and overall financial stability.
Many central banks provide ongoing funding to banks in both normal and stress times, playing a role similar to the Federal Home Loan Banks in the United States, and there has been a slew of recent announcements by central banks that they will be ramping up lending even further.[1] But when the Basel Committee finalized the liquidity coverage ratio in 2013, its new international liquidity requirement, the Chair of the Committee appeared unaware of the range of central bank practice. Basel Committee Chair Stefan Ingves 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.’”[2] That is, a committee of bank examiners and regulators took upon itself the duty of defining how central banks should lend to commercial banks. Chair Ingves’s stated objective was at odds with the lending principles at most central banks. Moreover, the approach reduces rather than enhances bank safety and soundness and financial stability by disincentivizing banks from being prepared to use the discount window.
While Ingves was talking about the central bank as a lender of last resort (LOLR), not ordinary central bank lending, in practice there is no meaningful distinction between the two roles. For example, the Fed’s primary tool for addressing liquidity strains is primary credit, which is also the facility that supports monetary policy implementation. Indeed, using the term “lender of last resort” as a synonym for central bank lending likely adds to the severe stigma associated with discount window lending in the United States.[3]
Unfortunately, the attitude toward central bank lending enshrined in the LCR permeates bank regulation and examination in the United States, with the result that banks are unwilling, and in many cases unprepared, to use the discount window. A better approach than regulators and examiners dictating what sort of Fed lending is appropriate is for the Fed to establish unambiguous terms on its lending programs, and bank regulators and supervisors to recognize lending on those terms as supportive of safety and soundness and financial stability.
In particular, the Fed as central bank could state definitively that a commercial bank or credit union that has access to primary credit can rely on being able to borrow for at least 31 days. As discussed below, the Fed has already stated that primary credit is a no-questions-asked facility; that banks can borrow loans with up to 90-days initial maturity, repayable and renewable on request; and that Reserve Banks have waived their right to demand early repayment of a primary credit loan.
If the Fed did so, regulators and examiners should then recognize a bank’s borrowing capacity in liquidity requirements. Of course, if the Fed remains vague about its reliability as a source of funding, U.S. regulations and examinations should recognize that the Fed may not be reliable.
Surveillance, Stigma and Mixed Messaging
A common misconception of U.S. researchers and regulatory policymakers is that there is a global consensus that central bank lending to commercial banks is socially costly and should be rare.[4] The misconception has arisen in part because the Fed has been schizophrenic about lending since the 1920s. The Fed would like banks to borrow from the discount window when “appropriate” but would not like banks to use the Fed as an ongoing source of funding. Over the ensuing decades, the Fed has bounced between “come hither and borrow” and “stay away.”[5]
In the 1920s, after many banks that borrowed frequently from the discount window failed, the Fed sought to develop a stigma against borrowing at the discount window. The Fed sought to do so by leveraging the “tradition against borrowing” – any borrowing (not just borrowing from the Fed) apart from taking deposits – that then prevailed among banks. The Fed once again sought to increase stigma in the mid-1950s when there was an unwanted increase in discount window borrowing. The Fed did so by subjecting borrowers to “surveillance” and, if borrowing was deemed inappropriate, “counseling” or “discipline” would follow. The Fed reiterated efforts to curtail borrowing after the Global Financial Crisis when it concluded that banks, especially U.S. branches of foreign banks, had borrowed too much. Indeed, in the June 2016 FOMC meeting, John Williams, the President of the San Francisco Fed, stated:
It’s not about who’s causing the stigma. It’s us, right? We’re the central bank, and the policies that we set and the decisions we make are what cause banks not to want to come to the discount window…so I think that when we say we want to reduce stigma, we might look maybe a little more inwardly and think harder about why it is that we developed that situation in the first place.[6]
Conversely, in 1972, 2003 and currently, the Fed has sought to increase banks’ willingness to borrow from the discount window. From the mid-1960s through 2002, the discount rate was below market rates, so the Fed developed rules to ration banks’ use of discount window credit. The Fed concluded that ambiguity about those rules was contributing to stigma, making the discount window a less effective monetary policy tool; so, in 1972 they revised Regulation A (which sets the discount window lending terms) in an attempt to make each bank’s “borrowing privilege” clearer. Stigma persisted, however, and in 2003 the Fed decided the rules were adding to rather than reducing stigma, and therefore sought to eliminate the rules created in 1972 and make the discount window a no-questions-asked facility. In order to encourage banks to only use the discount window as a backstop rather than a source of ongoing funding, the Fed moved the discount rate to 100 basis points about the target federal funds rate. The 2003 revision was slowly working, but stigma returned with a vengeance when banks that borrowed during the GFC were accused by the public of receiving bailouts and pilloried by Congress.
Currently, after the failures of Silicon Valley Bank and Signature demonstrated the importance of banks’ being prepared to borrow from the discount window, the Fed has returned to taking active steps to encourage banks to be willing to borrow.[7]
Global Perspective
Most other central banks have a starkly different perspective on lending to banks, given how they implement monetary policy. All central banks must have assets equal at least to the amount of their currency outstanding to perform their function. Prior to the creation of the ECB, the principal asset of the Bundesbank was loans to commercial banks because it judged that purchasing government debt was inflationary (link). The ECB inherited that tradition, and prior to the European banking crisis also conducted monetary policy through loans to banks rather than asset purchases. Currently, the ECB, BoE, and RBA are all letting their Covid-era asset purchases run down and are transitioning to frameworks in which loans to banks are their primary assets.
Another source of confusion is an incorrect assumption that there is a meaningful distinction between regular central bank lending and lender-of-last-resort lending. Most central bank lending to resolve liquidity strains is done by normal standing facilities. In the United States, for example, primary credit (what most people mean by the “discount window”) is limited to financially sound banks, while unsound banks may get secondary credit. However, secondary credit is almost never extended. It was primary credit, for example, that funded Silvergate Bank’s orderly winddown in November and December 2022.[8]
Even in the case of emergency credit programs, however, central banks should be transparent about when they would lend, and liquidity regulation and examination should take those terms as given. For example, while the Fed has opened emergency lending programs for nonbanks in some circumstances, it has made it clear that no nonbank should count on it doing so. Consequently, regulatory and supervisory assessments of the liquidity of a financial institution that does not have access to the Fed’s regular discount window should, of course, not assume that the institution will have access.
Constructive Ambiguity is Dead
Bank regulators may also be taking their cue from outdated papers in the economic literature that argued that central banks should control the potential moral hazard caused by lending through “constructive ambiguity” about their willingness to lend.[9] The view did not survive the Global Financial Crisis, however. Constructive ambiguity is dead, and the consensus view among economists is that central banks should state their lending stance clearly and in advance.[10]
The issue of how central banks should design their lending programs is critical for increasing financial stability and implementing monetary policy, but it is not an issue addressed here. The point here is that when central banks provide clear terms on their lending programs, examiners and bank regulations should take those terms as given rather than deciding to invalidate them. By contrast, if the central bank is ambiguous about how it wishes banks to make use of its lending programs, examiners and bank regulations should treat the central bank as an unreliable source of funding.
How U.S. Regulations Treat Borrowing from the Fed
As noted, in 2003, the Federal Reserve substantially revised Regulation A to make primary credit a no-questions-asked facility. That is, the Fed indicated that banks could borrow for whatever reason they wished, including, for example, to on-lend the funds into the federal funds market, which at that time was a vibrant interbank market, unlike today. To increase banks’ willingness to borrow, the Fed not only continued to keep borrowing secret but also stated that it would not inform examiners of borrowing unless there was a specific safety and soundness concern. To educate examiners that banks should be free to borrow, the Fed, OCC, FDIC, OTS and NCUA published a joint SR letter that stated that primary credit was a valuable tool for banks’ liquidity risk management.[11]
The Federal Reserve’s and the other banking agencies’ stance toward the discount window shifted after the Global Financial Crisis, even though the Fed did not change Regulation A. The agencies helped design the liquidity coverage ratio, and then they officially adopted the requirement as a rule. The LCR requires banks to hold “high-quality liquid assets” (HQLA) in an amount sufficient to meet an estimate of net cash outflows under stress. Even if banks have collateral pre-positioned at the discount window where they are supposedly allowed to borrow on a no-questions-asked basis, banks are not allowed to count anticipated borrowing from the Fed as a cash inflow under the LCR. Nor does collateral that a bank has pre-pledged to the discount window count as HQLA unless the collateral already counts as HQLA (such as Treasury and agency securities).
Large banks operating in the U.S. are also required to conduct internal liquidity stress tests at the overnight, 30-day, 90-day and one-year horizon on at least a monthly basis. Examiners pressure banks to calibrate the stress tests so that they align with the LCR’s assumptions. Like the LCR, banks are not allowed to anticipate borrowing from the discount window at the overnight or 30-day horizon, although they are allowed to do so at the 90-day and one-year horizon.
Thus, banks are required to be able to meet their liquidity needs under stress for 30 days before anticipating borrowing from the discount window. That’s perverse, because the discount window is intended primarily to meet banks’ immediate liquidity needs but not be an ongoing source of funding. Indeed, the Acting Comptroller Michael Hsu and the Group of Thirty have recommended that banks be required to establish discount window borrowing capacity to meet their immediate funding needs.[12] Relatedly, the liquidity requirements associated with bank resolution state that a recapitalized bridge bank that qualifies for primary credit is allowed to anticipate borrowing from the discount window for a few days but no longer.[13] Similarly, the Fed requested comment on tightened liquidity requirements on the branches and agencies of foreign banks in large part because the banks borrowed for longer than the Fed would have liked from the discount window and the Term Auction Facility during the GFC, even though the Fed was encouraging banks to use the window and created the TAF precisely so that banks would be willing to borrow.[14]
No doubt reflecting these mixed messages, when the OCC revised its examination manuals in 2023, the section on liquidity initially stated that the discount window was an unreliable source of funding. The manual stated:
The discount window is available to relieve liquidity strains for individual banks as well as the banking system, but the Federal Reserve Banks are not required to lend through the discount window and may turn banks away. [emphasis added][15]
While that language was quickly removed, the manual still says that examiners should “apply additional scrutiny” to banks that “place significant reliance” on the discount window, and the OCC pointedly points out that “…Federal Reserve Banks are not required to lend through the discount window.”
A survey of BPI bank treasurers found that examiners, including Fed examiners, discouraged use of the discount window.
All the treasurers noted that if they borrowed, they would immediately inform their bank supervisors, and in most cases this notification would occur prior to borrowing. One bank relayed that the notification of their supervisor would be automatic, because borrowing is one of their “early warning indicators.” Another emphasized that if she did not tell her examiner, the examiner would be “furious.” Several banks noted that while discount window staff at their Reserve Bank and officials at the Board wanted them to be willing to borrow, they believe it would be viewed quite differently by their examiners, including those from the Federal Reserve. All stated that even if the supervisor indicated in the moment that it was OK to borrow (such as happened in spring 2020), borrowing could be viewed as a red flag that would result in additional supervisory scrutiny down the road. One bank noted that its internal guidelines state that if the bank borrowed from the discount window, it would be required to immediately provide the Fed a repayment plan because that is what the bank had once been told by a Fed examiner, possibly in reference to borrowing for more than one night.[16]
Doing it Right: The Bank of England
In 2012, the Bank of England commissioned a review of its lending operations.[17] The report, which was written by Bill Winters, stated that the best way to reduce stigma was to “regularize” use of the Bank’s liquidity facilities. The report also recommended that the Bank should “avoid constructive ambiguity” and provide “even greater clarity upfront about the terms on which it would expect to provide liquidity insurance in a range of circumstances.” Winters also recommended that the Bank coordinate with domestic and international authorities so that “policy actions the Bank pursues to encourage acceptance of Bank liquidity facilities are not effectively offset or rendered redundant by the actions of others.”
The Bank followed many of Winters’s recommendations. In 2019, the PRA (the UK bank regulator) observed that the Bank had increased the availability, flexibility, and certainty of its lending facilities. In response, the PRA revised its guidance. The old guidance stated that banks were expected to first use their HQLA before turning the discount window. The new guidance stated that the facilities are intended to be “open for business” and there was no presumptive order in which a bank should choose to borrow from the BoE or draw down its liquidity buffers. The guidance specifically states that the PRA does not expect banks to use the discount window only as a “last resort.”[18]
On Aug. 4, 2022, the Bank of England announced a new Short-term Repo Facility. The facility provides banks seven-day collateralized loans at Bank Rate (the BoE’s deposit rate) against gilt collateral.[19] The Bank stated that the SRF should be used freely from the point of instruction. Simultaneously, the Bank’s Prudential Regulatory Authority issued a statement reiterating how the Bank intended the facility to be used, and that the PRA would conform its supervision of bank’s use of the facility to those intentions.
The Bank intends that the STR should be used freely from the point of introduction, as a way for counterparties to access reserves as necessary. The PRA would judge use of the STR as routine participation in sterling money markets and intends that it should be seen as such by bank boards and overseas regulators.[20]
On Dec. 9, 2024, the Bank published a discussion paper explaining that it would be transitioning to a new implementation framework in which collateralized loans to banks would be its primary asset.[21] Index Long-Term Repo operations will, along with the STR, constitute the majority of BoE assets. The ILTR facility provides banks six-month collateralized loans against three types of collateral. Simultaneously, the PRA again issued a statement reiterating the Bank’s intention for the facility and that supervision take that intention as given:
The Bank intends that the ILTR should be used freely as a way for counterparties to access reserves. The PRA would judge usage of the ILTR to be routine sterling liquidity management and intends that it is also seen as such by firms’ boards, credit rating agencies and overseas regulators.[22]
Moreover, when the PRA examines banks’ liquidity, it recognizes that the Bank’s lending facilities will be a source of funding.[23] In conformance with Basel III, however, the PRA requires banks to satisfy the LCR and the LCR does not recognize banks’ capacity to borrow from their central bank’s discount window.
Liquidity Coverage Ratio
Currently the Basel LCR states:
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 honour credit facilities, or may decide to incur the legal and reputational risk involved in not honouring the commitment, in order to conserve their own liquidity or reduce their exposure to that bank.[24]
Clearly, such concerns do not apply to ordinary central bank liquidity facilities. There is no safety and soundness risk to a bank borrowing from its central bank’s ordinary lending facilities. Indeed, it is almost universally recognized that safety and soundness as well as financial stability is enhanced when banks are able and willing to borrow from their central banks.[25]
The LCR does allow banks to count the capacity to borrow from the central bank as a source of liquidity in one circumstance: when the central bank offers “restricted use committed credit facilities.” However, the BCBS provides extremely prescriptive rules about when RCLFS can count, including precisely at exactly what highly punitive fees the lines should be sold.[26] Indeed, the BCBS stated that the restrictions placed on the facilities were intended to ensure that central banks should “remain the lenders of last resort.”[27]
By failing to recognize the liquidity benefits of being able to tap normal central bank liquidity facilities in those jurisdictions where such credit is reliably and readily available, the LCR is an inaccurate measure of a bank’s liquidity risk. The measure could be improved considerably by taking a jurisdiction’s central bank’s lending facilities as given. For example, the LCR could state that banks can count as a cash inflow borrowing from their central banks against prepositioned non-HQLA collateral as long as the central bank has clearly committed that such funding would be available to the bank for at least 30 days in the stress circumstances that the LCR envisions.
Recommendations and Conclusion
The Federal Reserve should state unambiguously that a bank that qualifies for primary credit may count on borrowing for at least 31 days. Consistent with that statement, the U.S. banking agencies should simultaneously announce that they view such borrowing as normal liquidity management and that a bank that is prepared to borrow is safer than one that isn’t. Doing so would enhance bank safety and soundness and improve financial stability by encouraging banks to be prepared to borrow while simultaneously making the agencies’ liquidity assessments more accurate. It would also take the agencies (including the Fed’s regulatory arm) out of the role of dictating to the Fed’s discount window lending arm how it should do its business.
The change would also increase U.S. GDP by allowing banks to devote more of their balance sheets to lending to households and businesses, pledging those loans to the discount window, rather than investing in a vast amount of unproductive deposits at Federal Reserve Banks. Indeed, the BIS has estimated that the financial stability benefits of complying with the liquidity requirements in Basel III are substantially outweighed by the cost in terms of lost GDP day-in and day-out.[28]
In recognition of these benefits, at the end of 2024, Senator Warner of Virginia introduced a bill that would require the U.S. banking agencies to recognize the liquidity benefits of a bank’s preparedness to borrow from the Fed. The bill states “…regulations and supervisory standards of the agency with regard to liquidity provision shall give appropriate, positive consideration to…the ability of the depository institution to access liquidity through advances under this section, including the pre-pledged collateral of the depository institution.”[29]
The Fed took a large step toward such transparency with its 2003 revisions to Regulation A which removed most of the rules governing when a solvent bank was allowed to borrow. Nevertheless, its FAQs still state “…a Federal Reserve Bank has no obligation to make, increase, renew, or extend any loan or advance to any institution”.[30] As noted, that qualification is echoed in the OCC examiner guidance.
The Fed took a further step toward an affirmative commitment to lend to eligible banks with pre-positioned collateral on March 15, 2020 when Reserve Banks waived their right to require repayment on demand for a term primary credit loan with a remaining maturity of more than 30 calendar days.[31] Economically, committing to not call a callable term loan even if the borrower becomes ineligible for primary credit is the same as committing to roll over an overnight loan for the same length of time. A similar step would be for the Fed to indicate that a bank would receive 31 days’ notice before losing primary credit eligibility. The banking agencies could then adjust liquidity regulation and examination accordingly.
[1] The European Central Bank, the Reserve Bank of Australia and the Bank of England have all announced that the share of their assets that will be loans to commercial banks will be increasing and significant going forward.
[2] Basel Committee on Banking Supervision 2013a.
[3] See Madigan and Nelson 2023.
[4] For example, in a speech in 2013 while a Fed governor, Jeremy Stein, now at Harvard, stated that the case for having liquidity regulations rests on the premise that the provision of contingency funding by a central bank to a commercial bank in a crisis “… is socially costly, so that it is an explicit objective of policy to economize on its use in such circumstances.” (Stein 2013). In a 2015 Fed working paper, Mark Carlson, Burcu Duygan-Bump and I stated that liquidity regulations were needed in part to limit the potential costs including moral hazard of central bank lending and “…increase the likelihood that when central bank lending does occur, it is in response to marketwide, pure liquidity situation.” (Carlson, Duygan-Bump, and Nelson 2015)
[5] Nelson 2021.
[6] Federal Open Market Committee (2016) p. 39.
[7] See, for example, Barr (2024).
[8] Nelson (2024).
[9] See, for example, Enoch, Stella, and Khamis (1997).
[10] See Domanski, Moessner and Nelson (2014) and Tucker (2014).
[11] Board of Governors of the Federal Reserve System (2003).
[12] Hsu (2024) and Group of Thirty (2024).
[13] Federal Reserve System (2021).
[14] Federal Reserve System (2019).
[15] The language appeared on page 14 of the Comptroller’s Handbook, Liquidity. Office of the Comptroller of the Currency (2023).
[16] Nelson and Waxman (2021).
[17] Winters (2012).
[18] Bank of England (2019).
[19] Bank of England (2022a).
[20] Bank of England (2022b).
[21] Bank of England (2024a).
[22] Bank of England (2024b).
[23] https://bpi.com/wp-content/uploads/2022/12/Against-What-Liquidity-Risks-Should-a-Bank-Self-Insure.pdf
[24] Basel Committee on Banking Supervision (2013b) paragraph 149.
[25] https://bpi.com/wp-content/uploads/2022/01/A-Major-Limit-on-the-Feds-Crisis-Toolkit-Shame.pdf
[26] Basel Committee on Banking Supervision (2014a).
[27] Basel Committee on Banking Supervision (2014b).
[28] Covas and Nelson (2017).
[29] Warner (2024).
[30] Federal Reserve System (2024). See the answer to FAQ “The Federal Reserve describes the primary credit program as a ‘no questions asked’ program with minimum administration. What does that mean?”
[31] Board of Governors of the Federal Reserve System (2024). See footnote 6.
