Incorporating Discount-Window Borrowing Capacity into a Liquidity Requirement

The banking crisis in March 2023 highlighted the shortcomings of the discount window’s current operations. Silicon Valley Bank and Signature Bank were not prepared to borrow from the Fed. This has led to calls for requiring banks to pre-position collateral at the Fed and test their arrangements to ensure that they can access the window during the time of financial stress.[1]

In this blog post, we examine the liquidity of Category III and IV banks after accounting for their borrowing capacity at the discount window.[2] Information on discount-window borrowing capacity is provided on the banks’ 10-Ks. We focus on large regional banks because in their annual filings, most Category I and II banks do not separately report their borrowing capacity at the discount window versus their borrowing capacity at Federal Home Loan Banks. Given that the suggested metric is focused on a bank’s ability to borrow from the discount window, having information on two different types of borrowing capacity is crucial. In addition to reporting the levels, we present the sum of cash assets and unused borrowing capacity at the discount window as a percentage of uninsured deposits. Then, we compare these percentages in December 2022 to their levels in December 2023 to see how they changed after the March 2023 banking turmoil.

Although we normalize these contingency funding sources by uninsured deposits, it is important to note that the behavior of uninsured deposits may differ greatly based on depositor profiles. In particular, deposits of customers that have a longer-term and multifaceted relationship with the bank, and deposits that are needed for the depositors’ ongoing operations, are less likely to flee.

We find that the large regional banks are well prepared to manage deposit withdrawal risk by cash assets and the ability to borrow from the discount window. The large regional banks had large cash assets and the capacity to borrow from the discount window before the 2023 banking crisis. Nonetheless, these banks increased their holdings of cash assets and borrowing capacity at the discount window after March 2023. As a result, large regional banks are now better able to meet deposit outflows.

Table 1. Ultra-short-term Liquidity Ratio of Failed Institutions, December 2022

Source: SEC Form 10K filings, FR Y-9C, Call Report.

In Table 1, we examine the liquidity ratios of the three banks that failed in 2023. It shows cash, unused borrowing capacity at the discount window and unused borrowing capacity at the FHLBs, both in total (Panel A) and as a percentage of uninsured deposits (Panel B). Although having cash assets and adequate borrowing capacity is important for all banks, it was crucial for these banks, which relied on a high level of uninsured deposits. The failed banks tended to be illiquid, holding only 7 percent of cash assets against uninsured deposits. In addition, they did not have large borrowing capacity at the discount window. This suggests that the failed banks would not have been able to meet deposit withdrawals by accessing the discount window even if it had functioned smoothly. Their borrowing capacity at the discount window was also lower than their borrowing capacity at their regional FHLBs: 14 percent of uninsured deposits at the discount window, versus 27 percent of uninsured deposits at the relevant FHLB.

Table 2. Ultra-short-term Liquidity Ratios of Category III and IV Banks, December 2022

Source: SEC Form 10K filings, FR Y -9C, Call Report.

In Table 2, we examine similar liquidity ratios for several Category III and IV financial institutions in 2022. These are large regional banks that operate under a traditional banking model with a wide network of branches, and they rely on deposit funding. We find that these institutions were better prepared to meet deposit withdrawals than the failed institutions. While the failed banks held only 7 percent in cash relative to uninsured deposits, these banks had average cash assets equal to 19 percent of uninsured deposits. In addition, the borrowing capacity as a percentage of uninsured deposits for the surviving banks stands at 25 percent, compared to 8 percent at the failed banks. These banks’ capacity to borrow from FHLB was smaller than that of the failed banks, equal to 19 percent compared with 27 percent of failed banks’ uninsured deposits.

The diminished capacity at the FHLB likely reflects both industry and supervisory concern arising from trouble that SVB experienced in attempting to borrow from its regional FHLB (which was unprepared to lend in a sufficient amount) and transferring collateral from the FHLB to its Federal Reserve Bank. As the next table shows, significant capacity has shifted from the FHLBs to the Federal Reserve Banks over the past year.

Table 3. Liquidity Ratios of Category III and IV Banks, December 2023

Source: SEC Form 10K filings, FR Y-9C, Call Report.

Table 3 presents information on the contingent liquidity capacity of these same Category III and IV banks in December 2023. We note that the banks had more borrowing capacity from the Federal Reserve System in December 2023 compared with December 2022, because they could borrow through the Bank Term Funding Program. However, we constructed the liquidity metrics looking solely at discount-window borrowing capacity, because the BTFP is a temporary program (and stopped extending new loans on March 11, 2024). Most banks also did not disclose their borrowing capacity at the BTFP. Nonetheless, this shows that the banks increased their ability to meet deposit withdrawals after the March 2023 crisis, as measured by cash reserves and discount-window borrowing capacity. On average, the banks had cash assets equal to 26 percent of uninsured deposits. In addition, borrowing capacity was also 39 percent of uninsured deposits. The percentage of cash and discount borrowing capacity against uninsured deposits increased from 44 percent in 2022 to 65 percent in 2023.

Regulators have emphasized the importance of overcoming issues with both preparedness and stigma related to using the discount window. However, based on banks’ 10-Ks, Category III and IV banks were prepared to tap the discount window before the March 2023 crisis and significantly increased their readiness afterward. As a result, as banks have started improving their operational readiness to borrow, it is important that regulators ease the stigma attached to discount-window borrowing.

Regulators can reduce this stigma by offering incentives to use discount-window borrowing more often. For instance, Susan McLaughlin, who led the discount-window operations at the Federal Reserve Bank of New York, suggested that the regulators should revise supervisory and regulatory practices and related language to support the banks’ use of the discount window.[3] This is because the current liquidity rules require banks to hold high-quality liquid assets to handle a sudden withdrawal of deposits, even though the regulators are encouraging banks to use the discount window when they need it. By definition, and perhaps paradoxically, “high-quality liquid assets” excludes discount window borrowing capacity.

Moreover, internal liquidity stress testing and resolution funding plan requirements generally don’t allow banks to count discount-window borrowing capacity as a source of contingent liquidity, although the discount window is part of such institutions’ contingency funding plans. The regulators would be able to improve bank readiness to tap the discount window and reduce the stigma attached to discount-window borrowing by recognizing both cash assets and discount-window borrowing capacity as means to meet deposit outflows and encouraging banks to use these during times of financial stress.[4]

And finally, because the new liquidity metric focuses on discount-window borrowing capacity and the size of uninsured deposits, it would be important for regulators to have a clearer understanding of uninsured deposits and design a metric more suitable for the broader banking industry.

The depositor profiles of SVB and SB differed from those of other banks (see the link). Most banks have much more diversified client bases and hold a large portion of operational deposits, which are placed to fulfill specific business purposes (e.g., meeting payroll or paying suppliers). However, SVB and SB were in an unusual position, because they resembled the small banks of the 1930s that operated without deposit insurance. Both SVB and SB had depositor bases that looked local: their depositors reportedly knew each other personally and interacted frequently. SVB’s depositors were connected through venture capital networks, and SB’s depositors through law firm networks. In addition, the majority of SVB’s and SB’s depositors could be considered financially naïve, at least relative to the financial firms that drive the majority of uninsured deposits in the United States today.[5]

For these reasons, while they design a new metric, banking regulators should consider the differences in the characteristics of uninsured deposits that were held at SVB and SB compared with those currently at other banks.

Appendix A. Liquidity of Other Category III and IV Financial Institutions in 2022

Source: SEC Form 10K filings, FR Y-9C, Call Report.

Note: Charles Schwab reported the face value of collaterals pledged at the discount window. Since these collaterals are in the form of credit card loans, we applied a 74-percent haircut to construct its borrowing capacity at the discount window.

Appendix B. Liquidity of Other Category III and IV Financial Institutions in 2023

Source: SEC Form 10K filings, FR Y-9C, Call Report.

Note: Charles Schwab reported the face value of collaterals pledged at the discount window. Since these collaterals are in the form of credit card loans, we applied a 74-percent haircut to construct its borrowing capacity at the discount window.


[1] For instance, Michael Hsu, the acting Comptroller of the Currency, suggested a new ultra-short-term liquidity requirement that accounts for the banks’ readiness to access the Federal Reserve’s discount window. Under Hsu’s suggested rule, banks would be measured by their exposure to uninsured deposits and receive credit for the reserves they hold, as well as for collateral pre-positioned at the discount window.

[2] In Appendices A and B, we present information on other Category III and IV financial institutions that are excluded from this blog: Charles Schwab, American Express, Synchrony Financial, Ally Financial and Discover Financial. These institutions are excluded from our analysis because they generally do not rely on deposits to fund their activities. U.S. Bank and 5/3 Bank are excluded because they did not separately report their borrowing capacity at the discount window versus their borrowing capacity at Federal Home Loan Banks. TD Bank USA and BMO US are excluded because they did not report their borrowing capacity.

[3] McLaughlin, Susan. “Lessons for the Discount Window from the March 2023 Bank Failures.” Yale Program on Financial Stability, Sept. 19, 2023. https://som.yale.edu/story/2023/lessons-discount-window-march-2023-bank-failures.

[4] Wessel, David. “How to Fix What Ails the Fed’s Discount Window.” Brookings, March 12, 2024. https://www.brookings.edu/articles/how-to-fix-what-ails-the-feds-discount-window/.

[5] Anderson, Haelim, and Adam Copeland, “Banks Runs and Information.” Liberty Street Economics, May 12, 2023. https://libertystreeteconomics.newyorkfed.org/2023/05/banks-runs-and-information/.