Lessons for the Discount Window from the March 2023 Bank Failures

by Susan McLaughlin, Executive Fellow, Yale Program on Financial Stability, susan.mclaughlin@yale.edu | September 19, 2023

In March 2023, the unprecedented pace of the runs on Silicon Valley Bank (SVB) and Signature Bank (Signature) took many by surprise. Much has already been done to identify ways to mitigate the risk of such runs and the threat they pose to financial stability. Indeed, in the US, several proposals to strengthen bank regulation and bank recovery and resolution planning have already been released for public comment, and work on possible modifications to the deposit insurance framework and to supervision of “novel activities” is already underway.

One part of the financial stability toolkit that has been less frequently mentioned is the Federal Reserve’s ability to act as lender of last resort to the banking system through the discount window. While lender of last resort tools cannot by themselves mitigate run risk—and are most powerful when well aligned with good bank regulatory and supervisory frameworks—they can help address banks’ short-term liquidity needs and mitigate bank run risk. The events of March 2023 provide an opportunity to reflect on ways to improve banks’ willingness and ability to use the discount window—which can be particularly important in a world where runs can occur over hours, rather than days. This note draws on observations about the SVB and Signature failures to offer some suggestions on steps that can be taken to reduce discount window stigma and improve operational readiness and agility—to meet the challenges presented by today’s fast-moving banking system.

What is the discount window?

A foundational responsibility of any central bank is to serve as the lender of last resort to the banking system to facilitate the ongoing flow of credit to the economy. In the US, Section 10B of the Federal Reserve Act authorizes the Fed to extend credit (on both an intraday and overnight basis) to US-chartered depository institutions (including commercial banks, thrift institutions and credit unions) and US branches of foreign banking organizations. (This note uses the term “banks” throughout to include all these types of institutions.)

The Fed’s discount window provides overnight and term credit to banks as needed to support the ongoing flow of credit to US households and businesses. The Fed operationalizes discount window lending through its twelve regional Federal Reserve Banks—which all rely on the same technology infrastructure, follow the same collateral policy and lending guidelines, and operate with a common set of lending documents.

A bank that requires a discount window loan borrows from its local Reserve Bank on a collateralized basis. A bank’s need for a discount window loan typically stems from a late-day funding need due to an unexpected withdrawal or a failure to receive expected funds. Sometimes, it occurs as a result of a bank’s temporary inability to repay its intraday borrowing from the Reserve Bank by the end of the day. However, the need for credit can arise at any point in the day.

There are three types of discount window credit: primary, secondary, and seasonal credit. Primary credit loans serve as the principal safety valve for ensuring adequate liquidity in the banking system and are available to banks that are in generally sound financial condition, as determined by their primary regulator. Secondary credit loans are available to banks not eligible for primary credit, and these loans come with a higher rate, higher haircuts, and a shorter term. Primary credit loans are supposed to be granted with “no questions asked”; secondary credit loans are subject to more restrictions on usage of the funds borrowed. Seasonal credit is available to banks with deposits of less than $500 million that can demonstrate liquidity needs of a seasonal nature. Primary credit typically makes up the bulk of discount window loans outstanding.

While the discount window is an important tool to ensure that banks have access to the liquidity they need to intermediate credit to support the economy, the events of March 2023 showed that stigma remains. As concerns about the health of many regional banks in the US prompted an acceleration in depositor withdrawals in early March, borrowing from Federal Home Loan Banks (FHLBs) increased sharply relative to discount window borrowing, and we subsequently learned that many banks were utterly unprepared to use the discount window as a source of backstop liquidity. There are several steps the Fed could take to reduce the stigma associated with primary credit and improve banks’ incentives to use it as a contingency funding source. These steps would strengthen the effectiveness of this tool in mitigating run risk for banks of all sizes.

It is useful to recall that for much of the Fed’s history before 2003, discount window credit was extended at a below-market rate—but only when the borrower had exhausted other sources of funding. This attestation, combined with the significant administrative oversight and monitoring that also accompanied the extension of discount window credit, served to stigmatize discount window borrowing. Historically, policymakers had concerns about the moral hazard risk associated with possible bank overuse of the window, as well as concerns about how to ensure that Fed credit was not fueling speculative activity. These concerns gave rise to a strategy of “constructive ambiguity” among some Fed policymakers—dating back to the Great Depression—regarding the discount window, in which there was no pre-commitment about its availability to any bank on any given day. This attitude is perhaps understandable in the period preceding the establishment of rigorous prudential supervision for all types of banking organizations operating in the US. However, it served to discourage use of the discount window and has created stigma that continues to the present day.

In 2003, the Fed endeavored to reboot and de-stigmatize the discount window by establishing two main tiers of credit: a primary tier that would be lent on a “no questions asked” basis to sound institutions at an above-market rate, and a secondary tier that would be lent to less-sound institutions on more restrictive terms. Fed policymakers have since 2003 attempted on many occasions to communicate to the public that primary credit is a legitimate source of funding that banks may use when needed with “no questions asked.” Indeed, since the onset of the COVID-19 pandemic, the Fed has redoubled its efforts to signal the message that primary credit should be used when needed; it extended the term of primary credit to 90 days and reduced the spread of the primary credit rate over the upper bound of the fed funds rate target range to zero.[1]

Yet, stigma is still present despite this communication. Why? Observers have pointed to the greater disclosure required under 2010’s Dodd-Frank Act, as well as the presence of cheaper alternatives to primary credit like the FHLBs.[2] But one key driver of stigma that gets less attention is the mixed messaging that has come from the Fed and other banking regulators on the topic of discount window usage. As discussed in more detail below, supervisory and regulatory practices, requirements and language are often at odds with the “no questions asked” policy stance on primary credit.

Steps to reduce stigma

Communicate clearly and regularly to the public that primary credit is a legitimate source of liquidity for banks to utilize when needed. A more comprehensive communication strategy that aligns words and actions is needed to combat discount window stigma and overcome the long memories of bank managers, market participants, reporters, and even bank regulatory staff themselves that “discount window borrowing” is to be questioned or avoided. When bank treasurers are grilled by their risk management committee or their regulators about why they borrowed from the window, this perpetuates stigma. When market commentary and press pieces speculate about which bank was behind the borrowing activity seen in the Fed’s weekly H.4.1 data release, this perpetuates stigma.[3] These dynamics will be difficult to disrupt without clear, intentional, and repeated communication from the Fed about its stance on primary credit. This communication must be mirrored in supervisory and regulatory practices and in adjustments to how primary credit is administered.
Recalibrate supervisory and regulatory practices and language to support the communication strategy about primary credit. There are some ways in which US bank supervisory and regulatory practices undermine the “no questions asked” message by disincentivizing primary credit borrowing. For example, collateral pre-positioned at the discount window is not included in the calculation of high-quality liquid assets (HQLA) for the liquidity coverage ratio (LCR) in the US. Supervisory guidance limits the ability of a banking organization to rely on primary credit as a source of liquidity for its depository institutions in its recovery and resolution plan (R&RP) to “only a few days,” even though the process of recovery or bridging to a resolution outcome could take longer than that.[4] And primary credit is not consistently emphasized by supervisors as a source of funding banks should incorporate into their contingency funding plans (CFPs). Recalibrating supervisory expectations to better align with the idea that primary credit is a legitimate source of short-term funding would help to improve the incentives for banks to use it.
Use supervisory authority to require banks to be prepared to use the discount window if needed, as a matter of prudent contingency liquidity planning. In the US, the onus is on banks to make the effort to obtain access to the discount window and conduct test borrowings periodically, ahead of a need to borrow. Many banks do this; however, the postmortem reports by the Fed on SVB and the FDIC on Signature make clear that neither bank was operationally prepared to use the discount window when they needed it. SVB had identified the discount window as a source of funding in its CFP, but it had not tested its discount window borrowing arrangements in the year prior to its failure and was not prepared to manage within the cutoff times for movement of collateral within its time zone. Signature’s CFP was overly reliant on the Federal Home Loan Bank of New York and did not include the Fed’s discount window. Signature had not tested its discount window borrowing arrangements for five years prior to its failure and tried repeatedly in its final days to pledge collateral that was, based on publicly available information, clearly ineligible.

A chorus of Fed speakers have since reinforced the need for banks to be operationally ready to use the discount window. And on July 28, the Fed and other US banking regulatory agencies took a step in this direction, by issuing amended supervisory guidance for Category II and III firms that emphasizes the discount window’s value as a contingency funding tool and encourages banks to take the steps to be operationally ready to borrow—including small-value test borrowing and pre-positioning collateral. But guidance is not a requirement, and still leaves some discretion to banks, which seems suboptimal from a systemic risk perspective.

Why not require any bank that has access to primary credit to include it as a source of funding in its CFP? Making this a requirement would increase the likelihood that banks are ready to borrow from the discount window when and if needed. Further, any bank that is relying on primary credit in its CFP could be required to do the test borrowing and collateral pre-positioning needed to ensure readiness to borrow. Testing will assure that banks are aware of the mechanics of the borrowing and collateral-pledging processes, operational cutoff times in their time zone, and collateral eligibility guidelines. Readiness to use the discount window should be a supervisory expectation for contingency liquidity risk management against which banks should be assessed and, if found to be noncompliant, required to take corrective action.

Consider ways to decouple administration of the primary and secondary credit programs to support the de-stigmatization of primary credit. The 2003 redesign of the discount window was intended to destigmatize primary credit borrowing—by distinguishing it from secondary credit borrowing by troubled banks. But the fact that primary credit is administered in tandem with secondary credit may in fact have contributed to stigma; much of the Fed’s communication on its lender of last resort tools speaks in terms of “the discount window” or simply “advances” in a manner that does not distinguish between the two programs. Consider the following language from section 14.1 of the Federal Reserve Operating Circular 10, which governs discount window lending:

The (Reserve) Bank is not obligated by the Lending Agreement or otherwise to make, increase, renew or extend any Advance to the Borrower.

Such language appears to reflect wariness about lending to unhealthy banks, which seems entirely appropriate messaging for secondary credit. But it is not consistent with the idea of a “no questions asked” primary credit backstop and reflects the lingering effects of the constructive ambiguity doctrine that has served to stigmatize discount window borrowing over time. Indeed, this language was embedded in the Federal Reserve Act itself, well before the 2003 redesign of the discount window.[5] Clarity on the central bank’s stance on primary credit could be supported by administering the primary and secondary credit programs separately and communicating about them more distinctly.[6]

Develop and communicate a longer-term approach to pricing primary credit that is intended to mitigate stigma. Part of the 2003 redesign of the discount window included pricing of primary and secondary credit at a penalty rate, defined as a spread above the fed funds target rate.[7] Initially, the spread for the primary credit rate was 100 basis points over fed funds; with the onset of the Global Financial Crisis, the Fed reduced the spread to 50 basis points in August 2007 and then to 25 basis points in March 2008 as dollar funding market stress intensified. The spread returned to 50 basis points in February 2010 when the crisis-era terms for primary credit lending were partially reversed. In March 2020, as part of a set of measures to support the flow of credit to households and business amid the onset of the global COVID-19 pandemic, the Fed reduced the spread to 0 basis points, where it has remained since.

This lineage raises interesting questions about the classical Bagehot idea of a penalty rate for lender of last resort facilities. A penalty rate is surely appropriate for standing facilities that lend to undercapitalized institutions, or that are established for crisis liquidity facilities that lend to a wide range of borrowers not usually eligible for central bank liquidity. But is a penalty rate consistent with the idea of short-term liquidity provision to sound banks with “no questions asked”? And if so, is there a level at which the penalty rate can be set that would minimize stigma? Should the penalty vary in stress and normal funding market environments? Consideration should also be given to the appropriate relationship of the primary credit rate to other sources of contingency funding available to banks. Does it make sense that banks can borrow more cheaply from the FHLBs, for example?[8] What is the appropriate relationship between the primary credit rate and the lending rates at FHLBs (net of dividends) and other sources of contingency funding for banks? Can that relationship be calibrated to minimize stigma? A framework for pricing primary credit that answers these questions would be an extremely useful component of a broader communication strategy about discount window borrowing.

Steps to improve operational readiness and agility

In a world of intraday runs, it seems worth considering ways to make it easier and faster for banks to use primary credit when they need it. There are several steps that could be taken to modify the processes for establishing access to, and using, primary credit to better align it with the speed at which liquidity needs can arise.

Establish access early in a bank’s life cycle. The discount window is an important financial stability tool to mitigate systemic risk transmission in the banking system. However, discount window access is not automatic; there are multiple hurdles a bank must clear before it can borrow from the discount window. First, a bank must affirmatively apply for a master account on the Federal Reserve’s books—or identify a correspondent bank with a master account through which to operate—as discount window borrowing and collateral pledging are executed via banks’ Fed accounts. Next, the bank must apply for discount window access, which requires the execution of several legal agreements and resolutions. Finally, the bank must pledge collateral.

This leads to the question: Why not make the establishment of Fed master accounts and discount window arrangements an upfront part of the process for chartering a bank or establishing a branch in the US? The Fed has historically been quite careful in granting master accounts to banks given the risks account activity can pose to a Reserve Bank if the bank is not in sound condition. But perhaps newly established banks without a credit risk management track record could receive a master account that only provides the ability to borrow at the discount window up to the margined value of collateral they have pledged and does not provide access to other Fed services, such as intraday credit.

Automate the loan approval process for primary credit loans. Once a bank has discount window access, the loan request and approval process are still quite manual. A bank requests the loan by telephone and waits for confirmation that it has pledged sufficient collateral value to support the amount requested. The Fed recently announced it is taking steps to automate the loan request process; another step that could be taken is to automate the loan approval process for primary credit loans. Automating the process of confirming that there is sufficient unencumbered collateral value to support the loan requested could speed the loan approval process modestly and, perhaps more importantly, provide another way to signal that primary credit is available with “no questions asked.” An ability to override the automated check manually could be created to deal with cases in which the condition of the bank is in question.

It is useful to note that a shift to this “automatic” access paradigm for primary credit would need to be accompanied by improvements in the quality and frequency of supervisory updates to the national supervisory ratings database. This is because counterparty risk management for discount window lending happens primarily at the supervisory level. All US banking supervisory agencies are responsible for monitoring and rating bank safety and soundness—and for providing ratings for the banks they oversee to the national database. These ratings feed into Reserve Banks’ assessments of borrower condition and drive determinations about primary and secondary credit eligibility (as well as intraday credit access). There is a long history of poor and/or stale assessments of bank condition (Indy Mac, Washington Mutual, etc.) before March 2023. On any given day, the boundary between illiquidity and solvency can be fuzzy; bank supervisors who are monitoring the bank’s condition on an ongoing basis are best placed to provide an assessment of it.

Explore ways to expand the feasibility of late-day pledges of collateral to the discount window. While all eligible types of securities collateral can be pledged on a same-day basis in theory, in practical terms the devil is in the details. The table below summarizes the information on cutoff times for securities pledges that is provided on the Fed’s Discount Window and Payment System Risk webpage, as of the date of this writing. Figure 1 shows that depending on the location of a bank’s collateral and the time zone in which the bank is operating, a late-day pledge may or may not be feasible in practical terms. Indeed, SVB ran up against this issue in March 2023.

Figure 1
[9] This assumes no request for extension is made by the pledging bank and granted by the Fed.

For this reason, the Fed has long urged banks to pre-position collateral at the discount window so it is there if needed. However, there may be circumstances in which a bank loses access to funding on a given day and may need to move collateral to the Fed to support borrowing. The ability to execute same-day pledges may have been less of an issue in the past when bank runs materialized over a period of days. But in a world where runs can occur within the space of hours, it is worth exploring opportunities to expand the windows within which pledges can be effected late in the US business day. Importantly, this is not a discount window infrastructure issue; rather, it reflects the design of the broader wholesale payments infrastructure, of which Fedwire and the custodial banks are part, and which was designed for a different financial market environment than exists today.

We also saw in March 2023 that the process of getting collateral from FHLBs to Federal Reserve Banks was time-consuming due to a combination of legal and operational issues. Indeed, in the case of Signature, special ad hoc legal arrangements were required to effect a timely pledge of collateral to the Federal Reserve Bank of New York in a manner that would adequately secure the Reserve Bank’s loan. To the extent that banks continue to pledge large amounts of Fed-eligible collateral to their local FHLB, it seems advisable to begin work now to address any outstanding legal or operational questions associated with collateral transferability between FHLBs and the Federal Reserve System—before the next liquidity event occurs.

Explore ways to incentivize or even require banks to pre-position a certain share of their loan portfolios as collateral pledges to their local Reserve Bank. Loans, which represent the majority of assets pledged to the Federal Reserve System and the bulk of assets pledged by smaller banks, can take even longer to pledge to a Reserve Bank than securities. Loans eligible to be pledged as collateral by banks are not typically traded and must be pledged through a held-in-custody arrangement. This can take weeks to set up properly to ensure the Fed has perfected its security interest in the collateral. So, pre-positioning is particularly critical for loans, which are pledged in large part by banks that are not large enough to have been deemed “systemically important” but whose activities can nonetheless give rise to financial instability in today’s environment.

Conclusion

While there are surely legal, technological, cultural, and risk management issues to overcome in moving to a more agile and truly “no questions asked” paradigm for primary credit lending to banks, the recommendations offered in this note would collectively help to reduce discount window stigma—and to avoid outcomes in which a bank needs liquidity before it has put the necessary operational arrangements in place. Timely discount window borrowing alone would not have saved SVB or Signature, given their risky business models, concentration and interest rate risks, and the associated solvency issues present in both of these institutions. But it might have slowed the run and mitigated contagion risk, while buying valuable time to execute resolutions that were less costly for the deposit insurance fund. And in the case of a solvent bank that is experiencing liquidity stress, timely access to discount window liquidity can make the difference between recovery and failure. The frictions in discount window access and usage described here seem to reflect the legacy of the constructive ambiguity doctrine, and they continue to contribute to the stigma associated with discount window borrowing. While these frictions were perhaps tolerable during runs that unfolded over a period of days, they are problematic in a March 2023 world, where runs can occur in a matter of hours.

[1] Section 10B of the Federal Reserve Act restricts discount window lending to a maximum tenor of four months; the Board of Governors of the Federal Reserve System has some discretion within that constraint to set the maximum term.

[2] Leaving aside the question of whether the large-scale lending by FHLBs to banks like SVB and Signature is consistent with the policy objective for which the FHLBs were established in 1932, the fact is that many banks prefer to borrow from their local FHLB rather than their Federal Reserve Bank, as it is often cheaper than the discount window and does not carry the same stigma.

[3] The Federal Reserve took an important step to mitigate this type of speculation in March 2020, by discontinuing the district level reporting of discount window borrowing on its weekly report on factors affecting reserve balances. See https://www.federalreserve.gov/releases/h41/20200319/.

[4] See also the joint Federal Reserve Board-FDIC February 2019 guidance for G-SIBs.

[5] The FDIC Improvement Act of 1991, passed in response to the S&L crisis, amended the Federal Reserve Act to add Section 10B(b)(4) which states: “A Federal Reserve bank shall have no obligation to make, increase, renew, or extend any advance or discount under this Act to any depository institution.”

[6] Other central banks generally do not have different tiers of access in their lender of last resort standing facilities; financial institutions are either eligible or not eligible.

[7] Since the FOMC’s adoption of a target range for the policy rate in December 2008, the top of the target rate has been the reference point for the spread.

[8] This question is admittedly beyond the Fed’s remit, but it is relevant to the current debate on FHLB reform.