Much ado was made during the Banking Crisis of 2023 when the previously empty line item “Other Credit Extensions” became the largest lending category on the Federal Reserve’s balance sheet. The line item represented the Fed loans to banks that had subsequently been put into Federal Deposit Insurance Corporation (FDIC) receivership—inclusive of newlending to the failed banks that were temporarily reopened as FDIC-run bridge banks.
While the “ado” about these loans largely faded after they were paid off on November 30, 2023,a recent speech by FDIC Vice Chair Travis Hill provided the first official discussion of this lending. Hill observed [emphasis added throughout]:
The failures of SVB and Signature (and subsequently, but to a lesser extent, First Republic) placed substantial liquidity demands on the FDIC. The FDIC initially met these demands through borrowings from the Federal Reserve and did not pay off the borrowings in full until nearly nine months later. The unprecedented nature of these borrowings, and the substantial cost incurred, have raised a number of questions.
Two things were particularly “unprecedented” about this funding mechanism. As Vice Chair Hill noted, the FDIC did not immediately pay back the Fed upon the closing of the Silicon Valley Bank (SVB) and Signature Bank bridge banks or of First Republic Bank. The FDIC instead sold many of the assets collateralizing the Fed loans, swapping an FDIC corporate guarantee for the collateralization and keeping the Fed loans outstanding.
Second, and not discussed by Hill or elsewhere, the bridge banks’ extensive new discount window borrowings included uncollateralized borrowing. The FDIC made this “naked” borrowing possible by providing its corporate guarantee—backed by the full faith and credit of the United States—to the Fed as security.
The initial use of Fed liquidity for the bridge banks seems to have been driven by a desire to preserve liquidity in the face of uncertainty. Notwithstanding the other potential motivations for making expansive use of this funding—and the legal uncertainties around it (discussed below)—the FDIC had implemented a novel mechanism to supplement its liquidity in crises. However, the “substantial cost incurred” must also be considered, particularly when thinking about how this mechanism fits within the FDIC’s legal authorities.
What Happened?
When resolving a bank, the FDIC is normally legally required to choose the resolution option that results in the least cost to the Deposit Insurance Fund (DIF). The “systemic risk exception,” however, provides that if the Treasury secretary, in consultation with the president, and a supermajority of the FDIC and Fed boards agree that following the least-cost resolution requirement “would have serious adverse effects on economic conditions or financial stability,” the FDIC need not execute a least-cost resolution.
On March 12, 2023, the federal banking agencies jointly announced the invocation of the systemic risk exception. Thus, the FDIC was able to consider financial stability when resolving SVB and Signature and take pursuant actions that wouldn’t have presented the least cost to the DIF. (A least-cost resolution, in this case, would have involved imposing losses on uninsured depositors.) Also on March 12, the Fed invoked its Section 13(3) emergency liquidity provision authority and announced the Bank Term Funding Program (BTFP)—under which the Fed would lend to banks against the parvalue of their government securities.[1]
Additional FDIC announcements explained that the failed Silicon Valley Bank and Signature Bank would see all their uninsured deposits protected; all deposits would be transferred—along with the banks’ assets and some liabilities—to Silicon Valley Bridge Bank (SVBB) and Signature Bridge Bank (SBB). Following this announcement, the line item on the Fed’s weekly balance sheet called Other Credit Extensions suddenly shot up from zero:
Figure 1 – Discount window primary credit vs. “Other Credit Extensions” vs. the Bank Term Funding Program in 2023, weekly. Sources: Federal Reserve Board, FRED
The Fed added a footnote to its balance sheet saying[2] that Other Credit Extensions,
includes loans that were extended to depository institutions that were subsequently placed into Federal Deposit Insurance Corporation (FDIC) receivership, including depository institutions established by the FDIC. The Federal Reserve Banks' loans to these depository institutions are secured by pledged collateral and the FDIC provides repayment guarantees.
In a June 2023 notice[3] on this line item, the Fed further clarified,
In 2023, the Federal Reserve extended loans under its ordinary discount window lending authority (section 10B of the Federal Reserve Act), as well as, in one instance, under its emergency lending authority through theBank Term Funding Program(BTFP) (established under section 13(3) of the Federal Reserve Act), to depository institutions that were later placed into FDIC receivership. Discount window loans were extended to Silicon Valley Bank, Signature Bank, and First Republic Bank, and BTFP loans were extended to First Republic Bank, in each case, before the depository institutions were placed into receivership.
However, banks have failed with outstanding discount window loans before, and such lending has never shown up as Other Credit Extensions. (See Box: A Brief History of “Other Credit Extensions.”) For instance, Washington Mutual (WaMu) failed in September 2008 with $3 billion of outstanding discount window borrowing.[4]
The June 2023 Fed notice added that Signature’s discount window loans were transferred to its bridge bank,[5] and that both SBB and SVBB then took on new discount window borrowings:
The Signature Bank discount window loans that were outstanding when the depository institution was placed into receivership were assumed by Signature Bridge Bank, N.A., an OCC-chartered insured depository institution. In addition, new 10B discount window loans were extended to Signature Bridge Bank, N.A. and Silicon Valley Bridge Bank, N.A., both OCC-chartered insured depository institutions that were eligible discount window borrowers under the Federal Reserve Act.
Moreover, the Fed added that the discount window loans (and, in the case of First Republic, the BTFP loans) remained outstanding and were not assumed by the failed/bridge banks’ acquirers, but were instead left behind in the FDIC receiverships of the failed banks:
Upon the two bridge depository institutions and First Republic Bank being placed into receivership, the discount window and BTFP loans were not assumed by the acquiring depository institutions. In each case, the outstanding discount window and BTFP loans are being repaid from the proceeds of the sales to the acquiring depository institutions, if any, and the recovery on the collateral that was left behind in the receivership, supported by the FDIC guarantees of repayment.
Despite the sales of these franchises within weeks of their initial failure—SBB on March 19, SVBB on March 26, and the immediate sale of First Republic to JPMorgan Chase on May 1 (in other words, without establishing a bridge bank)—this Other Credit Extensions category remained substantial until the end of November 2023.
The weekly data (see Figure 1) shows that Other Credit Extensions peaked at $228 billion on the first reporting date following the First Republic failure (May 3).
Where Does the FDIC Normally Get Its Money?
While FDIC obligations have the “full faith and credit” status of US Treasuries, the FDIC is not funded by congressional appropriation; it is funded primarily by the fees it assesses on the banking industry. The Deposit Insurance Fund (p. 30) ended 2022 with a balance $128 billion and 2023-Q1 with $116 billion.
The FDIC also has a $100 billion line of credit with the US Treasury and an additional $100 billion line with the Treasury’s Federal Financing Bank (which is also funded by the issuance of Treasury debt).[6] The FDIC can also borrow from the Federal Home Loan Banks (FHLBs) and from banks themselves. It has no specific statutory authority to borrow from the Fed. Total FDIC borrowing is subject to a formulaic, statutory cap:
Figure 2 – Screenshot from the FDIC’s 2023 Annual Report describing its Maximum Obligation Limitation (MOL). Source:FDIC
FDIC guarantees would apply towards this cap on an expected cost basis, rather than by the total nominal amount covered. (FDIC guarantees would not count against the national debt ceiling.)
FDIC funding during the GFC
The FDIC borrowed during its cleanup of the Savings & Loan Crisis of the 1980s and 1990s. Yet, when the DIF balance went negative during the Global Financial Crisis (GFC), the FDIC didn’t do any borrowing, in part because of Fed liquidity provision—from policies unrelated to the FDIC or the DIF.
As shown in Figure 3, the DIF balance went negative in 2009. The “balance,” however, adjusts for expected losses; it is not an accounting of the FDIC’s liquidity. In fact, the DIF stayed very liquid without any FDIC borrowing.
As Figure 3 shows, the FDIC imposed a small special assessment of $5.5 billion on banks in June 2009. However, it later increased its liquidity through an elegant mechanism: a prepayment from banks of their regular FDIC insurance premiums. In December 2009, it raised more than $45 billion by requiring banks to pay their normal quarterly payment, plus the next 12 quarters’ worth, so the DIF could stay liquid despite a temporarily negative balance:
Figure 3 – DIF balance vs. DIF liquidity during the GFC. Source:FDIC
The FDIC wanted to avoid drawing on its credit lines for political reasons during the GFC. As Diane Ellis, then an FDIC director, explained in 2013:
During the most recent crisis, the banking industry had been beneficiaries of extraordinary government assistance, and the industry and public were suffering from what was termed bailout fatigue. It was believed that drawing on a borrowing line with the Treasury, which is backed by the U.S. taxpayer, would exacerbate matters, even though the borrowing would be only for liquidity purposes and would be repaid with interest.
The FDIC had also wanted to avoid imposing another special assessment because that would be booked as a large expensefor the already-fragile banking system. As the FDIC later said, “In the second quarter of 2009, when the[$5.5 billion] special assessment was charged, FDIC-insured commercial banks and savings institutions reported an aggregate net loss of $3.7 billion.” A prepayment of their regular assessments, by contrast, had the benefit of enabling banks to book it as an asset, while booking the expense slowly over time.[7] This is where Fed liquidity proved helpful—if entirely indirectly.
At this chronic, less acute stage of the crisis in late 2009, liquidity was no longer in short supply. The Fed had injected then-record amounts of liquidity through quantitative easing (QE), and the FDIC thus determined that banks were essentially sitting on excess liquidity. In this sense, the FDIC’s liquidity could be viewed as being supported by the Fed—but only fortuitously.[8]
The 2023 Unprecedenteds
1) Instead of immediately paying down the Fed loans to free up the backing collateral, the FDIC swapped its guarantee for the collateral encumbered at the Fed.
The public data on Fed lending is difficult to interpret in the Silicon Valley Bridge Bank and Signature Bridge Bank cases because the buyer of each franchise took only part of the balance sheet—and only loans from the financial asset pool. (See NYCB’s purchase of SBB and First Citizens Bank’s purchase of SVBB.) That is, we don’t know which assets of SVBB and SBB were posted at the Fed’s discount window, which takes both loans and securities.[9]
The First Republic data is clearer. First Republic was able to borrow from the BTFP (the other two banks failed before it was established) and the discount window; plus, JPMorgan took “substantially all of the assets” immediately upon First Republic’s failure. So, we are able glean in the public data that the FDIC kept most of First Republic’s Fed loans outstanding despite selling the underlying collateral.
In its last pre-failure earnings release, First Republic shared having about $14 billion of BTFP borrowing and $64 billion from the discount window as of March 31, 2023. In its order taking possession of the bank, the California state regulator said First Republic had a total $93.2 billion in borrowings from the Federal Reserve as of close of business on April 28 (the Friday before its weekend failure). In the Wednesday snapshots of the Fed’s balance sheet that bookend the date of First Republic’s failure (May 1), Other Credit Extensions increased by $58 billion, suggesting First Republic’s total Fed borrowing still stood at least at that latter number (in other words, the FDIC likely paid down, as an upper bound, $45 billion of the Fed borrowings):
Figure 4 – Discount window primary credit vs. “Other Credit Extensions” vs. the Bank Term Funding Program surrounding the time of First Republic’s failure and sale to JPMorgan, weekly. Sources: Federal Reserve Board, FRED
In addition to Other Credit Extensions not falling back to its pre-First-Republic-failure balance for a couple months, First Republic’s BTFP loans remained outstanding; this was visible by comparing two different types of Fed reports. Supplemental to the Fed’s Wednesday balance sheet reports, the Fed also reports month-end balances of the BTFP to Congress. For several months starting in May, these reports showed BTFP totals consistently approximately $14 billion above what the Fed reported in its weekly balances.
For instance, May 31, 2023, was both a month-end and a Wednesday; the Fed’s weekly balance sheet reported a total of $93 billion outstanding at the BTFP, while its report to Congress said $107 billion. On August 30, the Fed balance sheet listed $107 billion of BTFP loans outstanding, while the August 31 report to Congress listed $121 billion. This difference was the outstanding First Republic BTFP loans.
What does this mean? Well, certainly in the case of First Republic, effectively all the collateral that was posted to the Fed was sold to JPMorgan; there was thus no collateral left at the Fed associated with First Republic. So, the FDIC kept for itself approximately $58 billion of the Fed loans to First Republic, by replacing First Republic’s collateral (which the FDIC had sold to JPMorgan), with an FDIC guarantee of repayment.
There may have also been other loan balances like this from the bridge banks’ discount window borrowings; as noted, however, that data is less decipherable.
2) The bridge banks’ new discount window borrowing was partially uncollateralized.
SVB’s and Signature Bank’s struggles to borrow from the discount window in their final hours have now been well documented by official reports. Regulators closed SVB on March 10 with no discount window loans outstanding[10] and Signature on March 12 with $5.6 billion outstanding. When both reopened as bridge banks on March 13, Signature’s discount window loans transferred to its bridge bank, SBB. As nationally chartered bridge banks, SVBB and SBB were able to borrow from the discount window in the normal course of operation. SBB’s ultimate resolution came on March 19 and SVBB’s on March 26, when the deposit franchises were sold to other banks and the bridge banks put into receivership.
FDIC Chair Martin Gruenberg, responding to congressional questions for the record, wrote (p. 189) that SVBB had $126.5 billion of discount window borrowing outstanding by the time of resolution, and SBB had $53.6 billion—a substantial increase in borrowing in two weeks operating for SVBB and one week for SBB.[11]
He also revealed that SVBB’s discount window borrowing appeared to have little margin and SBB’s appeared to be substantially underwater. The SVBB borrowings of $126.5 from the discount window were against collateral with book value of $152.2 billion and fair value of $127.5 billion—a haircut of less than 1%. SBB’s outstanding $53.6 billion of discount window borrowing at the time of failure was backed by collateral with a book value of $65.3 and fair value of just $42.3 billion—a “haircut” of negative 27%.
This should serve as a good proxy for how much non-collateral security—in other words, FDIC guarantees—the Fed received. (It seems extremely unlikely that this kind of valuation gap would emerge within a week, not least because the discount window margin calls daily.) For the avoidance of doubt, it’s clear the Fed was not unsecured; it had the FDIC guarantee. Rather, the loans were simply partially uncollateralized.
Thus, the valuation details show that even before the FDIC sold the failed bank franchises, it was creating liquidity for itself at the discount window by simply providing FDIC guarantees as security. This goes a step further than Unprecedented #1 above, which demonstrated that the FDIC in some cases appeared to be creating liquidity by selling the underlying collateral of the Fed loans without immediately paying back the loans. This additional information suggests that, irrespective of the failed banks’ collateral in the bridge banks, the FDIC was using the bridge banks for new, uncollateralizedborrowing, creating liquidity with just the FDIC’s guarantees as security.
So, Why the New Liquidity Mechanism?
Why wouldn’t the FDIC have paid back the Fed immediately upon putting banks into resolution, like it normally does?
Uncertainty
One reason for the FDIC’s sourcing of liquidity through the failed banks in 2023 was the substantial uncertainty over the ultimate duration and size of the banking stress—and thus how much liquidity the FDIC would ultimately need—especially given the failed institutions’ relative size (Figure 5):
(PPTX | 135.94 KiB)
Figure 5 – Total assets of failed banks vs. total number of failed banks, by year. Source:FDIC
FDIC Chair Gruenberg wrote in his response to congressional questions for the record that,
Since these bridge banks were open and operating national banks that had access to the Federal Reserve's discount window, it was preferable to use this liquidity option to conserve the funds in the DIF for any potential subsequent resolution activity and/or other needs.
This answer directly addresses the FDIC's additional Fed borrowing while the bridge banks were open. It does not address any liquidity creation through loans beyond the value of the collateral or maintaining the loans after the underlying collateral is sold, though a similar motivation presumably held.
Moreover, because of the scale of the discount window borrowing of SVBB and SBB, and the fact that their acquirers left many assets behind, the FDIC may have wished to conserve the discount window liquidity until the underlying assets were sold. The FDIC got only $10.6 billion of cash in the First Republic sale and no cash inflow in the SVB and Signature sales. As noted, however, that explanation does not carry over to the First Republic case—where effectively all the assets were assumed by JPMorgan—nor to the uncollateralized borrowing activity of the bridge banks.
The debt ceiling
Another reason the FDIC may have resorted to this novel method of sourcing liquidity over its direct methods of borrowing was the then-binding limits on outstanding government debt. The US statutory debt ceiling became binding in January 2023, preventing any additive federal debt issuance. The Treasury was approaching the so-called “X-date”—at which it would not be able to pay its bills without further borrowing—potentially as early as June 2023. While the congressional standoff over the debt ceiling continued, the FDIC was constrained in its ability to call on Treasury or issue its own debt—both of which would count against the public debt limit.[12] In June 2023, however, Congress and the president suspended the debt ceiling until 2025, alleviating this constraint.
Borrowing aside, though, the FDIC can call on the existing funds of the DIF—which are invested in Treasury securities—without any net impact on the national debt. Vice Chair Hill said, however, that “there were significant challenges and frictions” in drawing on DIF funds “because of sensitivities related to the debt ceiling.”
When asked in a congressional hearing (starting around 4:12:00) if the FDIC was discouraged from drawing on its Treasury credit line owing to debt ceiling concerns, FDIC Chair Gruenberg opened his answer with, “I think there were discussions in regard to that….” However, in a written response to follow-up questions, he said (p. 188), “The FDIC was not denied nor discouraged from accessing its line of credit.”
Cost?
Another reason for tapping the Fed might be if the funding were somehow subsidized. As mentioned above and discussed more below, though, it was relatively expensive compared to the FDIC’s other contingent funding options, at least from a pure accounting perspective—that is, not considering political capital or other optics and stigma issues. For instance, FDIC Vice Chair Hill said, “I suspect we may overestimate the impact that the DIF’s temporary liquidity position has on depositor confidence,” suggesting that some at the FDIC view drawing on the DIF as a suboptimal choice for the FDIC to source its liquidity.
The “Substantial Cost Incurred” and Tensions with the FDIC’s Authority
The Fed’s June 2023 notice on Other Credit Extensions said that it began charging a 100-basis-point (bp) upcharge on the outstanding discount window and BTFP loans once the bridge banks and First Republic went into receivership. Vice Chair Hill’s recent speech revealed the behind-the-scenes causes[13] for that upcharge:
When the bridge banks were dissolved, and when First Republic was sold, the Federal Reserve viewed the advances as in default, which, under the Federal Reserve’s Operating Circular, results in a 500 basis point increase in the interest rate charged. After some negotiation, the Federal Reserve reduced the penalty rate to 100 basis points.
That the Fed continued to charge a penalty rate despite the strength of the FDIC guarantee clearly shows some continued discomfort with the loans on the Fed’s part. As Hill went on to note, FDIC guarantees are of the same credit quality as US Treasuries:
Yet, the FDIC fully guaranteed the borrowings, and, as banks are required by law to tell their customers, the FDIC is backed by the full faith and credit of the U.S. government.
He also provided a conservative estimate of the additional cost resulting from the penalty surcharge:
The 100-basis-point penalty rate was extremely costly given the volumes involved and the time it took to pay it off, ultimately costing the FDIC roughly $1 billion. In effect, this was a shadow tax on the banking industry, primarily charged through the special assessment, that in the short run decreases the Federal Reserve’s losses and in the long run increases the “earnings” the Federal Reserve sends to Treasury.
And as he also said, the FDIC’s alternative funding sources discussed above wouldn’t have come with that penalty rate:
The most obvious places the FDIC could turn are the Federal Financing Bank (FFB) and the Treasury Department, which would have charged a cheaper rate than the Federal Reserve’s penalty rate. […] Nonetheless, the FDIC chose to continue paying the Federal Reserve’s penalty rate for months even after the debt ceiling was raised, rather than borrow from the FFB or Treasury.
Regulators invoked the systemic risk exception to establish the bridge banks and protect SVB’s and Signature’s uninsured depositors. Notably, they did notinvoke the systemic risk exception as part of First Republic’s resolution, and, as noted, the FDIC paid the 100-bp penalty on some of the Fed’s previous loans to First Republic. It’s unclear how this borrowing choice fit with the “least-cost” resolution standard.[14]
Some Early Implications
As discussed, the 100-bp Fed premium relative to other FDIC funding options perhaps complicates any use of this mechanism within the confines of the least-cost test. That is, bringing FDIC guarantees to the discount window may be legally appropriate only in the case of invocation of the systemic risk exception—despite its use in the First Republic case. Moreover, as noted, even the 100-bp surcharge was the result of a negotiation (the contractual language demanded a 500-bp surcharge); there’s no guarantee that the Fed would be willing to engage in such negotiation in the future.
That said, and while a full discussion of the possibilities under this novel funding mechanism is beyond the scope of this article—as it would entail exploring several hypotheticals—several particularly plausible implications rise to the surface.
As we saw already, there were implications for expanding the universe of the FDIC’s available funding in crisis—particularly for identifying an option that avoids the federal debt ceiling constraint. By contrast, so-called Title II resolution—a post-GFC framework for non-bankruptcy-code, FDIC resolution of systemically important financial companies, which utilizes the creation of bridge financial institutions—uses the Orderly Liquidation Fund (OLF) for providing receivership liquidity at the holding company level. While the OLF losses would ultimately be recouped by assessments on large banks, it is initially funded by the Treasury; thus, there is a temporary debt ceiling impact. Using the depository institution subsidiary to directly obtain further funding from the discount window in such instances may thus be preferable.[15]
It’s also possible that, given the strength of the FDIC guarantee, other lenders might be more comfortable than the Fed was. While the Dodd-Frank Act of 2010 curtailed the use FDIC assistance to open banks under the systemic risk exception, such assistance remains possible (12 USC §1823(c)) for the FDIC if it fits within the limitations of the least-cost resolution test. For instance, while the Fed continued lending to First Republic through April 28 (the Friday before its weekend failure), the FHFA advised the Federal Home Loan Bank of San Francisco on March 17 to not provide any new loans to First Republic. This decision might have changed if First Republic had been backed by the full faith and credit of the United States through an FDIC guarantee.
Similarly, the reason First Republic ultimately ran out of Fed funding is that the supervisory downgrades it received on April 28 moved it from the Fed’s standard discount window facility (primary credit) to so-called secondary credit—which comes with higher collateral haircuts, among other stricter terms. To the extent the FDIC can enable new lending from the Fed by providing its guarantee, such supervisory downgrades need not bind.
Certainly, to the extent there is due or undue reticence toward accessing both the standing DIF funds and the FDIC’s backup credit lines—and to the extent the debt ceiling binds—the FDIC may feel its hands are tied. Thus, the Fed’s discount window and other lending facilities may, in a twist of irony, represent the least stigmatized option for the FDIC to preserve its liquidity in future crises.
