Understanding discount window usage: Key drivers and early warning signals

When cash flow gets tight or markets experience turbulence, banks, credit unions and other depository institutions have a reliable option: the Federal Reserve’s discount window. This lending program plays a significant role in supporting liquidity in the banking system and maintaining financial stability, as well as supporting the effective implementation of monetary policy. The discount window allows banks to readily access funding so they can function smoothly and continue to provide a steady flow of credit to households and businesses, even during times of market stress.

What makes the discount window particularly useful is its accessibility. Depository institutions can set up access and take out test loans—typically smaller loans of $10,000 or less—from the discount window to check their operational readiness for borrowing. Banks are not subject to any penalties for borrowing from the discount window, and the amount of loans a bank can borrow is primarily determined by how much collateral—such as securities and loans—they can pledge. Bank liquidity management is an evolving practice, and in this study we take a closer look at some key drivers of discount window borrowing since the Fed began releasing public data (with a two-year lag) with the passage of the Dodd-Frank Wall Street Reform and Consumer Protection Act in 2010.

Discount window activity increasing over time

Since the Great Financial Crisis, the number of banking institutions in the U.S. has declined substantially, driven primarily by the consolidation through mergers, acquisitions and forced or voluntary market exits. At the same time, the share of banks with discount window access grew steadily until 2023 (Chart 1), despite the discount window stigma—a concern that borrowing banks might signal a financial weakness as this lending facility is perceived to be the lender of last resort. In the period following the banking turmoil in first quarter 2023, the figures for discount window activity—measured by the number of banks with access and borrowing—has also risen significantly.

We classify banks as having discount window access if they conducted a test or borrowed at least once over the sample period of third quarter 2010 to first quarter 2024. Borrower banks are defined as institutions that took a loan from the discount window for amounts exceeding $10,000, at least once during this period. The analysis is based on the Federal Reserve’s discount window lending and Bank Term Funding Program (BTFP) data sources, and it excludes credit unions.

Over the period of 2010-2024, the Eleventh District banks with discount window access borrowed more actively from the facility compared to their U.S. peers (Chart 2).

Discount window borrowers’ capital ratios and reserves dropped since 2022

In the United States, the Dodd-Frank Act strengthened the minimum risk-based capital requirements to ensure that banks hold sufficient capital to absorb losses during a financial stress. Since then, U.S. banks have generally held higher levels of capital. However, data show that banks who borrow from the discount window have relatively lower capital ratios than non-borrowers (Chart 3). Furthermore, borrower banks’ adjusted tangible common equity (TCE) ratios, which account for the unrealized losses from banks’ hold-to-maturity (HTM) securities, show a sharper drop as the Federal Reserve raised interest rates in 2022 and unrealized losses started to rise. On the asset side of the balance sheet, banks who borrow from the discount window tend to hold lower levels of reserves—a bank’s most liquid assets, held as cash or deposits at the Federal Reserve and used to manage banks’ daily payment flows—as a fraction of assets since 2022 (Chart 4).

Balance sheet composition of the discount window borrowers

Table 1 compares the average balance-sheet composition of banks that borrowed from the discount window with those that did not take a loan during the period of 2010-2024. Banks who borrow from the discount window are, on average, larger than non-borrowing banks. Banks borrowing at the discount window tend to have more commercial and industrial loans, illiquid securities and non-core funding. commercial and industrial loans are typically secured by business assets and projected cash flows rather than more stable or liquid forms of collateral. They can be pledged as collateral to the discount window, unlike the Federal Home Loan Banks. Banks with higher commercial and industrial concentration may find the discount window a better fit for their contingent liquidity needs for this reason. At the same time, borrower banks tend to have lower shares of core deposits and capital buffer.

Some key drivers predict when banks will need the discount window

We can try to estimate the likelihood of a bank borrowing from the discount window in the future, as a function of that bank’s financial characteristics. We use Call Report data from banks in 2010-2024, focusing on banks that already have access to the discount window—those banks that tested or borrowed at least once over the sample period. Table 2 presents the key drivers predicting discount window borrowing among these banks. The results are consistent with an earlier Fed study, which examines the determinants of bank borrowing from the discount window during relatively calm financial periods of 2010-2015.

The results of our model show that banks with lower levels of reserve-to-asset ratios are more likely to borrow from the discount window, even after controlling for various other bank characteristics. Also, banks with more illiquid balance sheets—measured by share of commercial and industrial loans, illiquid securities and non-core funding, relative to their core deposits and insured deposits—are more likely to borrow from the facility. Discount window borrowing is more likely for banks with higher shares of Fed funds borrowing, repo funding and Federal Home Loan Bank advances, suggesting that these funding sources may be complements rather than substitutes. In addition, banks are more likely to borrow as their adjusted tangible common equity ratio declines. Banks with larger asset size or growth are also more likely to borrow from the discount window.

Another factor impacting the likelihood of discount window borrowing is a bank’s geographic location. Even after controlling for financial measures of bank-specific risk, a bank’s designated Federal Reserve district, which is mainly determined by the location of its head office, plays a role in predicting the discount window activity. That means there may be regional differences in how banks perceive the benefit of using the discount window. Eleventh District banks are more likely to borrow from the discount window after controlling for their observable characteristics.

We also examine the impact of quarterly growth in GDP and Fed assets on discount window activity. Borrowing generally declines when GDP growth is positive. However, the Fed has expanded its balance sheet to respond to periods of financial stress, and an increase in Fed assets correlates with an increase in discount window borrowing next quarter. Lastly, the results indicate that banks who used the discount window last quarter are more likely to use it again this quarter, but there is a seasonal pattern as well: borrowing is less likely during the first quarter of each year.

We can use the regression model from Table 2 to predict overall discount window borrowing activity, which we measure as the year-over-year change in the number of banks borrowing over the sample period. A comparison of the model-predicted activity against the actual activity is provided in Chart 5. Even though this model was only trained with data up to 2022, it predicted a substantial rise in discount window borrowing going into 2023 and could have provided an early warning before the 2023 banking turmoil.

As we have explored throughout this study, the discount window facility serves as a valuable source of funding when banks’ liquidity needs arise, helping to support overall financial stability. Our analysis reveals that discount window usage is systematically related to fundamental bank characteristics, such as banks’ reserve holdings, asset liquidity and several other key drivers. These findings help us better understand how and why banks turn to the discount window, which can help us build more effective tools and approaches to anticipate borrowing activity in the future.