FHLB Dividends: Low-Hanging Fruit for Reconfiguring FHLB Lending
A core part of the financial stability toolkit in any country's financial system is the central bank's lender of last resort tools—namely, the liquidity backstops that the central bank stands ready to provide to the banking system to weather temporary disruptions to funding markets access. Such backstops are typically provided both to bridge temporary disruptions in individual banks' access to funding, and more extended disruptions to funding markets. In the United States, the lender of last resort tool is the Fed's discount window. Yet, for a number of reasons well documented elsewhere, there is substantial stigma associated with borrowing from the discount window. Perhaps for this reason, over time, some market participants have come to view the Federal Home Loan Banks (FHLBs) as an alternative to the Fed's discount window for backstop liquidity needs, even terming them "lender of next-to-last-resort"—despite their limited ability to play this role relative to the Fed. The FHLBs are government-sponsored enterprises (GSEs) that borrow in debt markets primarily to make loans—termed “advances”—against real estate and government collateral to their nearly 6,500 members, which include credit unions, thrifts, insurance companies, community development financial institutions (CDFIs), and effectively the entire banking system. (For this article, we focus on the FHLBs’ lending to banks.) In addition to benefiting from discount window stigma, the FHLB System reinforces discount window stigma due to its ability to pass on subsidized funding to its members.
There is little public data on the interest rates the FHLBs charge, and what is public often appears to be comparable to or more expensive than the Fed. Yet, the FHLBs partially reimburse their borrowing members by distributing the bulk of their dividends based on the amount a member borrowed in a given quarter—a refund that is actively marketed to members. We demonstrate this below and share examples of FHLB marketing materials. This dividend structure reduces the “all-in” cost for FHLB borrowers, changing its comparison to the discount window’s primary credit rate and incentivizing banks to structurally rely more on the FHLB System than is optimal from a financial stability perspective.
On November 7, the Federal Housing Finance Authority (FHFA) released a comprehensive review of the FHLB System, in anticipation of the System’s centennial in 2032. The report examines the current FHLB System against its core missions, offers a vision for a modernization, and reflects on lessons from the March 2023 banking stress, noting that:
The ongoing market stress highlighted the need for a clearer distinction between the appropriate role of the FHLBanks, which provide funding to support their members’ liquidity needs across the economic cycle, and that of the Federal Reserve, which maintains the primary financing facility for troubled institutions with immediate, emergency liquidity needs.
FHFA regulation already makes clear that the mission of the FHLB System is to provide financial services to its members to “assist and enhance” the financing of housing and community lending. Given the FHFA’s clear desire to focus on realigning the FHLB System towards its core missions and away from its tendency to function as a quasi-central bank, now is the time to redirect its favorable advance rates and dividends towards the mission and away from simply subsidizing borrowing.
Roles of the Discount Window and the FHLB System
Importantly, the FHFA report pushes back against the problematic concept of FHLBs playing the role of lender of next-to-last resort, noting [emphasis ours throughout] that:
The role of the FHLBanks in providing secured advances must be distinguished from the Federal Reserve’s financing facilities… Due to operational and financing limitations of the market intermediation process, the FHLBanks cannot functionally serve as lender of last resort, particularly for large, troubled members that can have significant borrowing needs over a short period of time.
And further,
The reliance of some large, troubled members on the FHLBanks, rather than the Federal Reserve, for liquidity during periods of significant financial stress may be inconsistent with the relative responsibilities of the FHLBanks and the Federal Reserve.
The Federal Reserve has a statutory mandate to serve as lender of last resort to the banking system, in connection with its ability to create bank reserves, which enables it to meet highly elastic demand for liquidity in times of financial market stress. By contrast, the FHFA report notes that the FHLBs’ lending capacity is constrained by their ability to raise debt financing in the capital markets, and that the FHLBs are limited in their ability “to meet sizable liquidity requests late in the day or after debt markets close.” Yet, in March 2023, some troubled banks “were effectively using the FHLBs as their lender of last resort.”
As described in a prior Yale Program on Financial Stability article, some banks were completely unprepared to borrow from the discount window as growing deposit outflows increased their needs for liquidity in the spring of 2023. The discount window can provide credit against a much broader universe of collateral than the FHLBs, which can only lend against housing-related collateral and government securities.
Banks’ operational readiness to use the discount window and to move collateral from the FHLBs to their regional Federal Reserve Bank was certainly an issue in the spring of 2023. In its report, the FHFA says it (1) will provide guidance to the FHLBs to work with members to ensure all have established the capability to borrow from the discount window, and (2) expects all FHLBs to establish agreements with their local Reserve Bank to support timely pledges to the discount window. But this does not go far enough to realign the relative roles of the discount window and FHLB advances.
Discount Window Stigma Relative to FHLBs
It is well known that discount window borrowing continues to carry stigma. One driver of this stigma is the requirement, established by the Dodd-Frank Act of 2010, that the Fed must disclose all details of all discount window loans made, including borrower names—albeit with a two-year lag. The FHLB System, despite its public status, faces no such requirement—which likely appeals to potential borrowers. Moreover, the Fed discloses total discount window borrowing weekly, and the market can break this down by Fed district—which helps it parse potential suspects for having taken a loan. The FHLBs disclose each district’s advances only at a quarterly cadence.
However, some stigma also derives from the advantageous terms at which banks can borrow from the FHLBs relative to the Fed. The lack of transparency to the public about the terms of FHLB lending has likely shielded this driver of discount window reticence and stigma from broader public scrutiny; the FHLB System is less transparent than the Fed about the economics of its lending. The Fed publishes its primary and secondary credit rates in real time, and also makes historical data on rates publicly available to download. By contrast, most FHLBs make rate data available only to member banks; only FHLB Des Moines and FHLB Pittsburgh provide historical time series on advance rates (with the latter’s data only going back to 2020). FHLB Boston allows the public to query advance rates for a specific historical date, and some FHLBs publish the current day’s advance rates.
But the time series data that is published by FHLBs is on a gross basis and does not reflect the equity payouts that the FHLBs make to their borrowers. As discussed below, these dividend payments serve to reduce the borrower’s effective funding cost, frequently to a rate below that on equivalent-maturity discount window loans. While the Fed and other bank supervisors are increasing their efforts with banks to encourage discount window preparedness and usage, supervisory efforts cannot address the lower cost of FHLB advances.
FHLB System Preferential Treatment
The low cost of FHLB credit stems from the FHLBs’ preferential status in debt markets as government-sponsored enterprises (GSEs). As the FHFA report notes, this status allows the FHLBs to issue “debt in the capital markets at rates only slightly higher than those on comparable Treasury instruments. This ability arises from the joint and several nature of consolidated obligations, as well as the perception that the federal government would provide support in the event of a default by the FHLBanks.”
This GSE status gives the FHLB significant advantages as an issuer of debt. It makes FHLB debt eligible for purchase by government money market mutual funds and provides for favorable treatment in banks’ capital and liquidity requirements. GSE status also makes FHLB debt eligible for purchase by the Federal Reserve in its monetary policy operations. Indeed, in September 2008, when solvency issues at GSEs Fannie Mae and Freddie Mac led to market pressure on FHLB debt as well, the Fed announced its intent to purchase FHLB debt under its standard monetary policy authority, ultimately buying $14.5 billion. The New York Fed reported that, as a result of this announcement, “discount note yields declined up to 60 basis points ahead of the first operation and spreads to comparable U.S. Treasury securities narrowed.”[1] The FHLBs are also exempt from most taxes, and investors in their debt are exempt from state and local taxes on the interest income.
The FHLB System leverages its advantages as a debt issuer mainly to fund advances to members, but FHLBs also use the favorable access to debt markets to supplement their income (and enhance their dividends to borrowers, discussed below) by funding a relatively stable-sized portfolio of long-term investment securities ($187 billion as of 2023-Q3) and mortgage loans ($60 billion).
FHLB advance pricing faces few statutory constraints; FHFA regulation just requires that the FHLBs not price their advances below the marginal cost to the Bank of raising matching funds, plus the administrative costs of the advance. The FHFA report notes that, “Minimizing their debt issuance cost allows the FHLBs to pass benefits to members in the form of favorable advance pricing (relative to other funding sources) since the debt issuance cost is the core driver of advance pricing.” However, it does not note that this benefit accrues mainly to members who borrow from the FHLB system—nor does the report describe how this benefit is often effected through dividends paid disproportionately to those members.
FHLB Self-Capitalization Model
Unlike the Fed, the FHLB System is exposed to default risk and, as a result, subject to capital requirements. The FHLBs are required by statute to maintain 4% capital to assets, in addition to other risk-adjusted ratios. While FHLB System members are required to hold nominal amounts of “membership stock,” the FHLBs mainly capitalize their lending activity through activity-based stock. This is a unique structure that allows the FHLBs to “self-capitalize” as activity grows.
How does “activity-based stock” work in practice? When an FHLB member borrows from an FHLB, they must invest a percentage of the loan in FHLB equity; this is called “activity-based” stock. The exact percentage varies slightly across FHLBs but typically amounts to 4-5% of the loan. While the borrower pays interest on the full amount of the loan, the amount invested is not simply lost liquidity; it receives a dividend like any equity investment.[2]
FHLB Dividends Incentivize Borrowing
The FHLBs’ dividend rates on activity-based stock consistently exceed the stated rate charged on advances—meaning borrowers effectively receive a discount to the stated advance rate. Since FHLBs primarily capitalize through activity-based stock—which is issued to those who take advances—the FHLBs pay the bulk of their dividends to those who borrow the most from the FHLB System. Moreover, several FHLBs also explicitly pay much higher dividends on activity-based stock than membership stock—further tilting the payouts toward members that are active borrowers.
Indeed, a 2019 FHLB Des Moines presentation for members showed that while the membership stock dividend was simply in line with benchmark short-term interest rates, the activity-based stock dividend had historically received an additional 250 to 350 basis points (bps). The presentation specifically stated that the activity stock dividend level was “a deliberate measure to pay shareholders for using [the FHLB System] cooperative” [red markup ours]:
While public information about FHLB advances is relatively sparse, it’s clear that FHLB members are aware of the effect of FHLBs’ dividends on their total cost of borrowing. Some of the marketing of this benefit is in the public domain. For instance, a page on the FHLB Chicago website from 2019 shows how the dividend paid to activity-based stock (“B1” in the picture) consistently exceeds the dividend paid to general membership capital (“B2”). It also charts the size of the effective discount on the stated advance rate as a result of dividends over prior quarters; the discount hovers between 7 and 14 bps:
Another page shows the Q3-2023 calculation, offering a 13-bps reduction to the stated advance rate. Similarly, while FHLB Chicago—like most of the FHLBs—does not publicly provide historical advance rates, its homepage does provide a daily term sheet. In addition to effective advance rates, this sheet provides members the “all-in” rate they can expect to pay after adjusting for the expected dividend income. Here’s a snapshot from that sheet (full file here) on December 20, 2023 (red markup ours):
As of the date of this term sheet, the Fed’s discount window was charging 5.5% for loans of up to 90 days. Meanwhile, the FHLB was charging a headline rate of 5.49% for 3-month loans, and the post-dividend rate shown led members to expect another 13 bps refunded via dividends.[3]
FHLB Pittsburgh similarly advertises the benefit of the dividend and provides members with a demonstration of the refund’s impact on its website. As of this writing, the example is for a 1-year advance taken on October 23, 2023. It shows that while the stated rate on the advance was 5.72%, expected dividends mean a member can expect to effectively pay 5.39%—a 33-bps discount:
(Notably, however, the above calculation divides the interest cost by the full advance amount of $1 million rather than the actual liquidity provided to the borrower, which would be $0.96 million. After that adjustment, the rebate is only 11 bps.)
In a document on its webpage dated June 30, 2020, FHLB Atlanta illustrates a 21-bps discount on a 1-year advance as a result of dividends paid to borrowers:
FHLB Des Moines offers members a dividend discount calculator, and a 2019 presentation for members demonstrates a 19-bps discount on a 1-year advance:
FHLB Dividends: A Market Incentive to Avoid the Discount Window
While publicly available time series data from the FHLBs is limited, we can apply the type of calculation advertised by the FHLBs shown above to time series we obtained from the FHLB Des Moines and FHLB Pittsburgh websites. Below, we plot the Fed’s primary credit rate, the stated FHLB advance rates for a comparable maturity to the discount window, and the “all-in” FHLB advance rates that reflect the actual cost of borrowing after accounting for dividends paid to borrowers. (The precise calculation methodology is described in the endnotes.[4])
FHLB Des Moines data, which is available back through 2000, shows that the all-in cost of the FHLB advance has typically been markedly lower than the cost of primary credit. The exceptions occurred when FHLB advances became more expensive than the Fed in the latter half of 2008 and throughout 2009, as well as in 2022 and 2023—before recently falling below the primary credit rate again:
FHLBank Pittsburgh data, only available back through 2020, shows that while its stated advance rate has typically listed at a slight premium to the Fed’s primary credit rate, the all-in borrowing cost was lower than the primary credit rate for most of 2020 and all of 2021. The all-in FHLB cost then moved to a notable premium throughout 2022 and 2023, before tightening again recently:
At the outset of the pandemic, the Fed lowered the spread charged by the discount window over the upper bound of the fed fund funds rate (its monetary policy rate) to zero,[5] where it has remained. Despite this reduction in discount window pricing, which left the stated advance rates of both FHLB Des Moines and FHLB Pittsburgh higher than the Fed’s primary credit rate, the after-rebate cost at both FHLBs remained cheaper through 2020 and 2021. Moreover, it’s notable that both FHLBs went from then providing favorable rates to charging a premium throughout 2022 and 2023—just as demand for liquidity was picking up. A similar effect can be observed for FHLB Des Moines during the Global Financial Crisis.[6]
This shift in FHLB lending posture should further underline to banks and regulators the advantages of having collateral pre-positioned at the Fed. (At a minimum, establishing arrangements to readily move collateral from the FHLBs to the Fed and being operationally prepared to use the discount window would be directionally helpful—as the FHFA report notes.) If FHLB advance rates are consistently below the Fed’s primary credit rate—particularly during less volatile times, when banks are pre-positioning their collateral—the FHLB System may find it difficult to extract itself from playing a central role in banks’ contingency liquidity management, a role that the FHFA makes clear in its report is the Fed’s, not the FHLBs’.
Reforming the FHLB Dividend Construct
In thinking about the current way in which FHLB credit is priced, and the adverse impact it appears to have on discount window stigma, two statements from the recent FHFA report demonstrate a clear desire of the FHFA to reconfigure the incentives of the FHLB System, particularly given its public mission:
The FHLBanks receive certain advantages from their status as GSEs, and a greater portion of these benefits should be passed through to consumers and communities. FHFA supports the view that the unique advantages of the GSE status come with corresponding responsibilities for the FHLBanks to support their public mission.
And—
Congress has vested the FHLBanks with certain market advantages that enable them to provide low-cost wholesale funding and other services to their members. Therefore, it is reasonable to expect that the benefits of FHLBank membership should accrue predominantly to institutions that demonstrate a meaningful commitment to supporting the housing and community development mission of the FHLBanks.
While the FHFA Report did not specifically mention the borrower-favoring dividends or the effect thereof on banks’ contingency funding preferences, it did say the following:
To encourage members to increase their support for the FHLBank mission, FHFA plans to undertake a rulemaking and issue related guidance, as appropriate, whereby each FHLBank would provide added benefits—such as discounted advance rates or differential dividends on capital stock—to members with a strong and demonstrable connection to the FHLBanks’ overall mission.
In evaluating any reforms to how the FHLBs lend, the advance rates and the dividend rate must be considered together, as the all-in cost of borrowing reflects both. As the GAO said in 2003, “FHLBank officials said that there can be a connection between the advance interest rate that they charge and the dividend rate that they pay. That is, an FHLBank choosing to pay a relatively high dividend may need to charge a relatively high advance rate to earn sufficient profits to cover the costs associated with the dividend. In contrast, other FHLBanks may decide to provide value to their members through lowering their advance rates, which may mean a decrease in their dividend.”
If FHLB credit remains structurally competitive with—and often cheaper than—the Fed’s primary credit, discount window stigma will remain. As such, the starting place for implementing reforms desirable both for financial stability and FHFA mission alignment should be for the FHFA to promulgate FHLB regulations calling for the following:
- FHLBs’ stated advance rates must be set at some minimum premium to the Fed’s primary credit rate;
- Dividend payments and share classes shall distinguish between degrees of mission-consistent activity—that is, looking beyond simply borrowing from the FHLB System and to the nature of the borrowing member’s activity and its degree of alignment with advancing housing goals and community lending;
- The “all-in” cost of borrowing after accounting for dividends shall not fall below the primary credit rate except in cases of lending that is directly mission-consistent.
Now is the time to revisit the FHLBs’ dividend practices, both to better align FHLB lending with the System’s mission of funding housing and community development, and to reduce the incentives for banks to prefer FHLB credit over the Fed as a contingent source of liquidity. Addressing both these goals could mean altering the FHLB rate structure and creating a new class of “activity stock” that would be held against lending that supports housing and community development—and paying FHLB dividends principally on that stock.
The authors thank Vincient Arnold for his fantastic research support.
[1] Sitting between the relatively low legal hurdles of the Fed’s “standard” monetary policy authorities—of Section 14 of the Federal Reserve Act—and the relatively high legal hurdles of the Section 13(3) emergency lending authority is the Fed’s Section 13(13) authority. This authority allows the Fed to provide emergency loans against Treasuries and agency obligations—the latter of which the FHLBs issue by definition. This authority was used to provide standing credit lines to Fannie Mae and Freddie Mac in July 2008 (see p. 1 here); it could, in theory, do the same for the FHLBs—which represents another government safety net feature bestowed by the GSE status.
See also the Fed’s study of alternative assets, inclusive of FHLB debt, for effecting monetary policy when the U.S. was running budget surpluses (and thus running down the supply of Treasury securities that are purchased and sold to implement policy): Alternative Instruments for Open Market and Discount Window Operations (2002).
[2] This equity investment receives a risk weight of 20% for regulatory capital purposes. This contrasts with a 300% risk weight for publicly traded equities and 400% for non-public equity stakes.
[3] While we note the 3-month tenor because that’s what’s currently on offer from the discount window (statute limits the discount window to a maximum tenor of four months), one can see here that the entire FHLB-Chicago term structure is below the discount window rate—and even more so after accounting for the dividend refund.
[4] To provide a more accurate picture of the cost of borrowing from FHLBs, we have calculated an “all-in” cost measure, which accounts for the rebate from activity stock dividends paid to borrowers as part of the FHLB borrowing process. For purposes of calculating all-in rates on FHLB advances, we have broadly adopted the methodology in Ashcraft, Bech, and Frame (2008, Appendix), which captures (1) the cost of the advance rate plus (2) the cost of funding any haircut and the required capital stock purchase less (3) the capital stock dividend. We have made three methodological changes. First, we have assumed haircuts are nonbinding. This is to analyze the more “normal” positioning of these parallel borrowing mechanisms. To the extent FHLB haircuts do bind before the discount window, FHLB advances would become more expensive—unless market funding was cheaper than the FHLB, in which case the borrower would likely not be using an FHLB advance—or the borrower would likely be forced to (try to) move to the discount window. Second, to account for changes in discount window lending durations over time, where Ashcraft et al. consistently use a 30-day FHLB advance rate, we use advance rates aiming at matching that of the discount window at that time (e.g., if on a given date the discount window lent for a maximum of 90 days, we use 90-day FHLB advance rates). Due to an absence of data, when the discount window lent overnight, we use one-week-duration FHLB advance rates, the shortest duration rates available for the entirety of our FHLB Des Moines dataset; for FHLB Pittsburgh, we have one-day-duration data available. Third, similarly, where Ashcraft et al. used LIBOR to proxy the cost of the capital stock purchase, we have used AA-rated financial commercial paper of a duration matching that of the discount window (and therefore also the FHLB advance data) at that time (e.g., if on a given date the discount window lent for a maximum of 90 days, we use 90-day AA financial commercial paper rates); we use the 30-day measure for the brief period where the discount window’s maximum duration was 28 days. Where interest rate data was unavailable for a particular day, we used the previous day’s data point. For Q4-2023 data, we assume the FHLB dividends will be the same as the previous quarter. As is the case with the FHLB-provided calculation examples and Ashcraft et al., there is no adjustment made for the precise timing of the quarterly dividend payment. All interest rate data is from the Federal Reserve. Data on advance rates and dividends are from the applicable FHLBanks. Duration changes for the discount window can be found here and here. We have set the activity-based capital stock purchase requirement to 4.5% throughout for FHLB Des Moines, and 4.0% for Pittsburgh, consistent with publicly available data and the FHLBs’ capital requirements.
[5] Prior to the Global Financial Crisis, the discount window charged a 100-bps premium to the fed funds rate; the Fed lowered this premium to 25 bps during the crisis. Post-crisis, the Fed restored the premium to 50 bps above what was then the upper bound of the fed funds target range (a 25-bps range in which the Fed targeted the middle). This spread has been 0 since being lowered on March 15, 2020. For background on these discount window adjustments, see here and here.
[6] Likewise for the FHLB Des Moines in 2008 and 2009. The FHLBs also manage their collateral policy procyclically, increasing their haircuts during episodes of volatility. By contrast, the Fed manages its collateral policy so as to avoid the need to raise discount window haircuts during episodes of market stress.