Learning from banks’ 2023 borrowing from the Fed

By Jeff Huther and Jigar Gohel
ABA DataBank

The failure of Silicon Valley Bank and Signature Bank in March 2023 led to record use of the Federal Reserve’s discount window, the creation of a temporary lending program (previously discussed), and the failure of First Republic Bank in the following quarter. The Fed recently released loan-level discount window data from July to September 2023. With this release, we have a relatively complete picture of the Fed’s lending in response to the bank failures of 2023. From the data, it is clear that:

• This is clearer with the new transactional data than with the weekly totals available in 2023.
• Excluding FRC, most discount window activity in the first three quarters of 2023 was a normal mix of funding out of need and for testing purposes.
• FRC’s loans from the window shifted from overnight to term borrowing over the course of 2023 Q1-Q2 until its failure.

The overall conclusion from the 2023 data is that banks continue to use the discount window in various ways — including as a source of defensive liquidity amid stress elsewhere in the industry. This is confirmed by the fact that banks’ borrowing from the discount window following the SVB failure reflected wider liquidity concerns only for a short period of time and that the number of need-based borrowers settled at a higher level over the course of 2023.

Discount window lending in 2023

Discount window usage hit a record high immediately following SVB’s failure (based on data back to 2002). While the activity could have been interpreted at the time as a sign of bank fragility or excessive government intervention, the transaction-level data released this year show that the bulk of the borrowing was due to First Republic Bank (FRC, Figure 1).

Figure 1:  Discount window usage in spring 2023

Source:  Federal Reserve discount window

Within three weeks of the Silicon Valley and Signature Bank failures, use of the window had returned to historical averages for all banks excluding FRC.

‘Need’ or ‘test’ borrowing from the discount window

One of the challenges in analyzing loan-level discount window data is that there is no definitive way to distinguish between banks borrowing out of need and banks testing their borrowing lines. There are, however, clues that suggest much of the borrowing, even at the height of concerns about bank health in the spring of 2023, was for testing purposes.

To separate activity at the window into “test” and “need” categories, we applied the criteria shown in Figure 2 to the transaction-level data. We ordered the classifications based on simple, intuitive criteria anchored on data observations. For example, a bank with Other Outstanding Loans >0, more than one discount window borrowing in the quarter, with terms greater than three days and a “non-generic” loan amount would be labeled as a “need” borrower. A similar bank but with a “generic” loan amount would be labeled as a “test” borrower.

Figure 2:  Criteria for need or test loans

Application of our criteria, shown in Figure 3, shows that most borrowers, even at the height of the pandemic, have used the discount window to test their lines rather than to meet funding needs (the red line, for the number of distinct test borrowers, in Figure 3 has always been above the blue line, representing the number of distinct need borrowers).

Figure 3:  Need vs. test usage of the discount window, quarterly totals

Sources:  Federal Reserve and call reports

The need category was unusually low throughout 2021, likely reflecting fiscal and monetary stimulus policies. The rise in need borrowers in 2022 was presumably a result of the Fed’s sharp increases in interest rates.  The last data available, usage from July to September 2023, suggests that the effects of the higher rates and more need borrowers persisted well beyond the rate hikes and SVB’s failure. The persistent seasonality (low in the first three months of each year and generally peaking in the second half of each year) likely reflects underlying seasonality in economic activity.

One note of caution about interpretation of test and need results is that our standard indicators of testing are inadequate in times of widespread concerns about bank health. When supervisors fear contagion, they may “encourage” banks to use the discount window regardless of underlying need. While this may seem contrary to normal concerns about discount window stigma, it muddies the waters, making it difficult to tell whether an individual bank is facing funding pressures or taking a defensive liquidity stance encouraged by supervisors. We saw this approach most clearly during the pandemic, when the largest banks borrowed the same amounts ($500 million, $1 billion, $5 billion, etc.) for the same terms (90 days) despite a lack of need. The Fed also removed the 50 basis point penalty in March 2020, bringing the primary credit rate in line with top of the fed funds target range to reduce stigma and encourage use.

In sum, the transaction-level data show that a small number of banks use the discount window as a regular source of backstop liquidity and a somewhat larger number of banks that test their lines with the Fed through periodic borrowing in small amounts (where “small” depends on the size of the bank).

First Republic deep dive

FRC’s use of the discount window in 2023 is of particular interest because it borrowed out of need, it was open with investors that it was borrowing from the Fed (directly taking on the issue of stigma) and it ultimately went into resolution with outstanding loans to the Fed and its Federal Home Loan Bank.

FRC borrowed heavily from the discount window in the immediate aftermath of SVB’s failure, relying on overnight loans through the end of March 2023 (see Figure 4). On quarter end, it shifted from overnight borrowing from the Fed to a term loan of 90 days.

Figure 4:  First Republic’s daily discount window balances in spring 2023

Source:  Federal Reserve

By the third week of April 2023, FRC’s needs increased and, on April 21, it prepaid its loan to the Fed (likely because of a revaluation of FRC’s condition and underlying collateral) and took out a new loan with a mid-May maturity date. That loan turned out to be insufficient and, following a poorly received Q1 earnings release on April 24, First Republic supplemented its term loan with overnight borrowing that grew until the FDIC put it into receivership on May 1. At that time, the $61 billion term loan due in mid-May was redeemed and then re-issued to September 15.

At the time of First Republic’s failure, it owed the Fed $79 billion and the FDIC estimated the cost of the failure to be $13 billion, the expense being borne entirely by banks through the FDIC’s Deposit Insurance Fund. The Fed was repaid in full.

Conclusion

Different banks use the discount window in different ways. A small number of banks see the window as one part of their regular funding sources. Another group of banks clearly view the discount window as a contingency source of funding, regularly testing their lines (other banks have tested their lines only once over the years and presumably concluded that was sufficient). FRC was unusual because it turned to the discount window for large-scale term borrowing and it failed with outstanding loans from the Fed, although the Fed did not bear any losses from FRC’s failure.  

 Jeff Huther recently retired as VP for banking and economic research in ABA’s Office of the Chief Economist. Jigar Gohel is an economic research associate at ABA. 

Box: The role of FHLB funding in 2023