We authored this post jointly with our friend and colleague, Lawrence J. White, Robert Kavesh Professor of Economics at the NYU Stern School of Business.
Some government financial institutions strengthen the system; others do not. In the United States, as the lender of last resort (LOLR), the Federal Reserve plays a critical role in stabilizing the financial system. Unfortunately, their LOLR job is made harder by the presence of the Federal Home Loan Bank (FHLB) system—a government-sponsored enterprise (GSE) that acts as a lender of next-to-last resort, keeping failing institutions alive and increasing the ultimate costs of their resolution.
We saw this dangerous pattern clearly over the past year when loans (“advances”) from Federal Home Loan Banks (FHLBs) helped postpone the inevitable regulatory reckoning for Silicon Valley Bank (SVB), Signature Bank, and First Republic Bank (see Cecchetti, Schoenholtz and White, Chapter 9 in Acharya et. al. SVB and Beyond: The Banking Stress of 2023). Not only did the FHLBs have a starring role in the recent banking turmoil, but they also played an important (albeit less publicized) part in the 2007-09 financial crisis by providing substantial funding to Washington Mutual, Countrywide, and Wachovia—all of which ultimately failed.
Put simply, in their current incarnation, the FHLBs make loans to illiquid intermediaries that are highly vulnerable to runs. However, from a public policy perspective, FHLB advances have extremely undesirable properties. First, in addition to being overcollateralized, these loans are senior to other claims on the borrowing banks—including those of the Federal Deposit Insurance Corporation (FDIC) and the Federal Reserve: If the borrower defaults, the FHLB lender has a “super-lien” (just in case the overcollateralization is insufficient to satisfy the FHLB’s claim). Second, there is little timely disclosure about the identity of the borrowers or the amount that they borrow. All we have are quarterly filings in which the banks can disclose the amount they borrowed, while the FHLBs reveal their top 10 borrowers. Third, reflecting the super-lien, the overcollateralization of their advances, and the implicit federal guarantee of their liabilities, the FHLBs appear unconcerned about the viability of their borrowers: They are willing to provide speedy, low-cost funding to failing institutions—something we assume private lenders would not do. (For more on the distortions and incentive problems associated with the FHLBs, see Judge.)
It was not always so. Created by federal law in 1932, the FHLB System was designed to provide wholesale lending in support of residential mortgage finance and community investment. As of March 31, 2023, the system had 6,484 members. Like other GSEs, including Fannie Mae and Freddie Mac, investors view FHLB System debt as having the implicit support of the federal government. As a result, the FHLBs can borrow at rates that are only modestly above Treasury yields. Having evolved over time, the System today is a nationwide set of 11 wholesale cooperative banks that jointly raise funds in debt markets and use the proceeds to lend to members: including commercial banks, savings banks, credit unions, and insurance companies. (For more detail see Frame, and White and Frame.)
At the beginning of 2022, total FHLB System advances were $335 billion—well below the average of $555 billion during the previous decade. But during 2022 and over the first quarter of 2023, advances roughly tripled, reaching $1,044.6 billion by March 31, 2023. In the following chart, we plot FHLB System lending since 2003 (in red). For comparison, we also show Federal Reserve lending (in black). As Ashcraft, Bech, and Frame and Judge emphasize, and the chart highlights, the System’s members today primarily use its advances as a tool for liquidity management during periods of stress. Not only that, but the increases during stress periods are several times larger than the rise in Federal Reserve lending. This pattern suggests that borrowers strongly prefer advances from the FHLBs over discount loans from the central bank that usually are accompanied by greater supervisory scrutiny.
FHLB Advances to FHLB Members and Federal Reserve Lending to Banks, (Billions of U.S Dollars) Q1 2003–Q1 2023
Sources: Board of Governors, Financial Accounts of the United States, BOGZ1FL403069330Q and H.4.1.
The following table shows FHLB lending at the end of 2022 and during the first quarter of 2023 as reported in the system’s quarterly disclosures. For each of these dates, we report the 10 largest recipients of FHLB advances.
Top Ten Holding Company Recipients of FHLB Advances (Billions of U.S. Dollars), Year-end 2022 and March 31, 2023
*NY Community Bank owns Flagstar, which assumed substantially all deposits of Signature Bridge Bank.
Sources: Advances are from Federal Home Loan Banks (2023) Table 12 and Federal Home Loan Banks (2023) Table 7. For total assets, with the exception of First Republic and TD Bank, the information is for March 31, 2023, and comes from 10-Q filings. For First Republic, the information is from “FDIC’s Supervision of First Republic Bank,” September 8, 2023. For TD Bank, the information is from “TD Bank Group Reports First Quarter 2023 Results: Report to Shareholders,” and is for January 31, 2023 (converted from Canadian to U.S. dollars using that day’s exchange rate).
We draw four conclusions from this table. First, total advances for the top ten rise sharply from year-end 2022 to the end of the first quarter of 2023 (from $218.8 billion to $326.0 billion). Second, lending is quite concentrated and rising. The top ten borrowers accounted for 26.5% of advances at the end of 2022. And this rose to 31% three months later. Third, banks with assets in the range of $500 billion to $700 billion are quite prominent—especially at end of the first quarter of 2023, when they constituted the top four and account for 17.2% of total FHLB advances. Of the eight banks on the year-end 2022 list, seven were still there at the end of the first quarter of 2023 (SVB isn't!). (See our earlier post on the failure of SVB.)
Perhaps most disturbing is that the FHLBs seem content to lend large amounts to troubled banks. Had SVB, Signature, and First Republic instead been forced to face market discipline during 2022, their borrowing cost would have surely been far higher. Thus motivated to address their losses at an earlier stage, they might have survived—or at least could have been absorbed by other banks at lower public cost than what we saw. Yet, from the FHLB perspective, the combination of overcollateralization and the super-lien made these loans extremely safe—even though the borrowers faced an elevated risk of failure.
Clearly FHLB lending does not diminish the risk to the banking system as a whole. When some of the borrowers do eventually fail, other lenders must wait in line behind the FHLBs to get paid. Short of an unprecedented systemic calamity, it is impossible to imagine circumstances where the FHLBs are not repaid fully, but it is not difficult to imagine wider systemic stress and spillover costs. Indeed, this is one way to interpret the numerous bank failures of 2008-09: the FDIC imposed a special premium on surviving commercial banks to replenish its depleted Deposit Insurance Fund that in part went to repay the FHLBs.
It also is easy to understand why, when large unrealized losses eroded (or even wiped out) their net worth during 2022, SVB, Signature Bank, and First Republic Bank turned to the FHLBs to stay afloat. In addition to avoiding greater supervisory scrutiny, FHLB advances allowed these banks to delay asset sales that would have forced balance-sheet recognition of their losses and compelled them to increase their regulatory capital ratios (either by raising equity or shedding assets). Instead, the banks gambled for resurrection on the back of mispriced government-sponsored financing. In the end, the gamble failed. Indeed, SVB’s March 8, 2023, attempt to raise capital came so late that uninsured depositors ran, and the bank failed within two days. The panic then quickly spread to other mid-sized banks with similar vulnerabilities.
Furthermore, there are strong indications that the FHLB system facilitates regulatory arbitrage. As Anadu and Baklanova discuss, and we explained in an earlier post, FHLB liabilities are treated as high-quality liquid assets (HQLA) for banks in meeting their liquidity requirements and are eligible holdings for government-only money market funds (MMFs).
These special features of their liabilities mean that the FHLBs now serve as a conduit through which the MMFs can finance U.S. banks in normal times. But matters are worse during periods of stress because, as deposits move out of vulnerable banks into government MMFs, the FHLB System ends up recycling the funds: MMFs use the inflowing funds that had fled the banks to acquire the liabilities of the FHLBs; in turn, the FHLBs use the proceeds to lend to the vulnerable banks, filling their panic-driven funding gap.
This stress-driven version of regulatory arbitrage appears to have been especially important in March 2023. During that month, commercial bank deposits fell by $307 billion, while borrowing and other liabilities rose by $510 billion: Overall, the banking system balance sheet actually grew. In the same period, government MMF shares rose by $442 billion. While we do not have monthly data for the FHLB System, we know that during the first quarter of 2023, FHLB advances rose by $216 billion, while FHLB bond liabilities increased by $312 billion. Hence, significant increases in government MMF shares essentially funded FHLB advances which, in turn, made up for much of the lost bank deposits: The FHLB System used its government imprimatur to provide low-cost funding to the banks, displacing the Fed as LOLR.
This all leads us to conclude that in its current guise the FHLB System is counterproductive. To counter the most damaging aspects of the regulatory arbitrage, we should eliminate or sharply scale back the System’s ability to serve as a lender to stressed banks.
For the most part, banks and other intermediaries rely on market sources of liquidity that impose a healthy discipline on the borrowers, helping to limit the risks that they take. However, in periods of financial stress, the market supply of liquidity can become dangerously scarce, which justifies the existence of a central bank LOLR. Put simply, the LOLR addresses a well-known externality: that individual bank runs (or failures) can turn into systemwide panics and fire sales that threaten the payment system and/or the supply of credit to healthy borrowers. In contrast, we know of no such theoretical or practical foundation that can justify the creation of a GSE that effectively supplants the LOLR, substituting for market sources of liquidity when that supply is costly.
Nor do we see any externality that rationalizes the existence of such a lender. As it currently operates, the FHLB System delays and undermines market discipline. It expands the supply of low-cost, federally subsidized credit to severely troubled, and potentially insolvent, banks. It also undermines supervisory discipline—especially that of the LOLR. For example, an effective LOLR must commit not to lend to insolvent banks: In addition to subordinating other lenders, such lending would make other recipients of LOLR loans—institutions that are solvent, but temporarily illiquid—suspect of insolvency. Moreover, lending to insolvent banks would not put an end to financial fragility. More likely, it delays and raises the costs of resolution.
Unless someone produces a good justification, policymakers should eliminate the FHLB’s role as U.S. lender of second-to-last resort. We also see little justification for yet another GSE or Federal agency to support residential housing, on top of the Federal Housing Administration (FHA), Fannie Mae, Freddie Mac, Ginnie Mae, and the Veterans’ Administration. However, it is likely too much to ask that the entire FHLB System be shut down.
Instead, we offer three concrete proposals for reform:
Eliminate the super-lien. As it stands, the FHLBs are tempted to lend to zombie banks in the form of a stealth, but temporary, bailout. This delays resolution, increasing public costs.
Require that Federal Reserve supervisors approve their banks’ borrowings (advances) from the FHLBs beyond a normal level: The conditions for advances should be at least as stringent as those for discount loans.
Require immediate public disclosure by each FHLB of its advances (or, at least of advances beyond some size threshold) and of the lending conditions (including the collateralization). In that way, other creditors would learn quickly about the strains that advance recipients may face, helping to focus counterparty and regulatory scrutiny where it belongs. Because this information is materially relevant for bank investors, policymakers also should require the borrowing banks themselves to make this disclosure in their quarterly filings, rather than just in their annual reports.
Some of these proposals—certainly the first—would require legislative action, but we hope that the FHLBs and bank supervisors would quickly implement what is feasible on a voluntary basis.
As it stands, the FHLB System is a destabilizing force in the U.S. financial system. By providing subsidized loans during period of stress, it undermines the LOLR. Fixing this would improve financial resilience by creating greater incentives for banks to manage risk.
This post is a revised version of Chapter 9 in SVB and Beyond: The Banking Stress of 2023.
Acknowledgement: Without implicating her, we are grateful to Professor Kathryn Judge (Columbia Law School) for her guidance regarding the legal aspects of FHLB reform.
Erratum 2023-12-06: In the bottom panel of the table, we have made modest corrections to several entries under “Total Assets" and “Advances as a Pct. of Assets.” These corrections have no impact on our analysis.