When engaging in 13(3) lending, the Federal Reserve must be “secured to [its] satisfaction,” which it has generally (and reasonably) taken to mean that its ex ante expectation should be that it will be fully repaid.i However, the Fed also interprets the legislative history of 13(3) as suggesting that this standard cannot be met with an entirely unsecured loan—a loan in which the Fed takes no collateral, third-party guarantee, or other protection of some kind. It seems at least partially due to this view that the Fed used available Treasury funds to support most of its emergency lending responses to COVID-19. However, if the Fed feels unsecured loans meet its repayment standard, it is not clear that it must avoid unsecured lending altogether. This is especially the case in the context of modern 13(3) which, through its statutory requirement to be “broad-based,” provides more security for repayment than individual unsecured loans—a pooling effect the Fed itself acknowledged during the Global Financial Crisis.
The Fed and Section 13(3) collateral
A confidential Fed memo from March 2009 released by the Financial Crisis Inquiry Commission (FCIC) reveals that Fed lawyers believed there was no clear precedent governing whether the Fed could make an unsecured loan. They said only that legislative history likely suggested a loan could not be “entirely unsecured.” The Legal Division noted the issue was “unnecessary to decide at this time”:
In the early days of the COVID-19 crisis in 2020, the Fed announced several 13(3) facilities to stem pandemic-induced market volatility. Before the CARES Act became law, the Treasury committed $50 billion of existing Exchange Stabilization Fund (ESF) funds to these facilities. These funds provided credit protection to the Fed lending facilities. Later, the Treasury replaced $30 billion of that funding with CARES Act funds, while non-CARES ESF funds remained committed to the Commercial Paper Funding Facility (CPFF) and Money Market Mutual Fund Liquidity Facility (MMLF). The CPFF bought commercial paper from highly rated firms, while the MMLF engaged in nonrecourse lending to banks, backed by commercial paper and government securities the banks purchased from money market funds.
The 13(3) programs that received equity funding from the Treasury under the CARES Act purchased corporate and municipal debt and extended credit against asset-backed securities (the Corporate Credit Facility (CCF), Main Street Lending Program (MSLP), Municipal Liquidity Facility (MLF), and Term Asset-backed Securities Loan Facility (TALF)).
The CPFF, CCF, MSLP, and MLF all saw the Fed extending unsecured credit, while the MMLF and TALF lent against unsecured collateral without recourse to the direct borrower—functionally an unsecured loan to the end-borrower. Thus, the Fed’s operating interpretation regarding unsecured loans likely required supplementary protection for all of these programs, irrespective of their financial risks.
The 2008 Experience
The Fed used similar structures during the Global Financial Crisis (GFC). The Fed sought additional funds as collateral—from the private sector and the Treasury, respectively—when constructing the GFC-era iterations of the CPFF and TALF. The structure of each is revealing.
The structure and legal interpretations around the 2008 CPFF point to several inconsistencies in the Fed’s view that it should avoid lending that is entirely unsecured. The 2008 CPFF, like its 2020 counterpart, bought highly rated issues of both asset-backed commercial paper (ABCP) and unsecured commercial paper. As per the aforementioned Fed memo, which outlines the legal support for the CPFF, Fed analysis determined that ABCP “itself is ‘secured’ for section 13(3) purposes: the collateral would be the assets that constitute the corpus of the asset securitization special purpose vehicle that issues the ABCP.”
In the case of unsecured commercial paper lending, however, the Fed required additional security. When the Treasury withheld from the CPFF a previously planned injection of TARP funds (p. 344-345), the Fed instead charged unsecured borrowers an up-front credit surcharge of 1% of the loan value. These funds were pooled to function as loss reserves and provide the protection necessary for the Fed to meet its security standard; as noted, the Fed thought it likely could not lend on a wholly unsecured basis.
The memo notes that, when making a 13(3) loan, the Fed can accept “collateral of any value, including collateral that at the time of the extension of credit may have a current market value that is less than the amount of credit extended, provided that the credit is secured to the Reserve Bank’s satisfaction.” In an interview with the FCIC, former Fed General Counsel Scott Alvarez provided further clarity on this point (around the 1:16 mark). He noted that other factors are required if lending against collateral worth less than the loan, such as appreciating collateral or lending with recourse when you expected the borrower to have “sufficient financial resources to repay the loan without resorting to the collateral.”
In the event, the Fed extended unsecured commercial paper credit against protection (in the form of up-front insurance fees) that amounted to 1% of the loan value. The Fed characterized these fees as available collectively to cover losses from any individual loan. Alvarez also stated to the FCIC that, “we never did lend…against assets that had a value that was lower at the time of the loan than the amount of the loan…including the CPFF.”
From a collective interpretation of the above, it’s clear that the Fed accepted that the other characteristics of the loans—such as the high ratings of, and Fed recourse to, its borrowers—were sufficient to back 99% of the value of each unsecured loan made by the facility. Surely that logic could be extended to fully unsecured lending, especially in cases where the unsecured loan, via recourse to a strong firm’s balance sheet, is actually safer than a collateralized alternative. As witnessed during the GFC, which began in earnest in the ABCP market, it’s clear that can easily be the case.
While the memo said “ABCP itself” is secured, ABCP lending doesn’t actually have any extra protection as a result of being collateralized. It is issued by conduits that have no capacity to pay beyond the cash flows on the assets serving as collateral. That is, there is no recourse for the Fed outside of the assets themselves; the loan is thus equivalent to an unsecured loan to the ABCP conduit.
The Fed’s Lorie Logan, Bill Nelson, and Pat Parkinson acknowledged as much (p. 90) when discussing another Fed facility that dealt in ABCP, the AMLF—the 2008 precursor to the MMLF. Like the MMLF, the AMLF lent on a nonrecourse basis to banks to purchase commercial paper from money market funds. The AMLF, however, lacked Treasury support and only lent against ABCP. The Fed officials noted,
“ABCP is generally seen as riskier than other types of commercial paper. . . [It] is not backed by a general promise to pay. . . The Fed’s choice to restrict the assets purchased under the program to ABCP can be seen as more about making it clear that the Fed’s loans were backed by specific assets as collateral. . .”
Indeed, even after including the credit surcharge on the unsecured paper, the total cost of credit that the Fed charged on unsecured CPFF loans was still lower than that on ABCP by 100 basis points. Specific collateral or not, the Fed implicitly understood that, in this case, its unsecured lending actually provided it greater security. The legislative history that the Fed’s Legal Division pointed to—which demonstrated that Congress expected the Fed would be protected when extending credit under Section 13(3)—was in this case better served by unsecured lending.
Also in 2008, the Treasury committed $20 billion of TARP funds to support the GFC-era iteration of TALF, which officials designed to provide up to $200 billion of non-recourse loans against specified asset classes. In an April 1, 2009 response to Congressional Oversight Panel questions on the TALF, Fed Chair Ben Bernanke and New York Fed President Bill Dudley said,
“…even if the economy evolves in a manner significantly worse than we currently expect, all credit costs will be more than covered by the haircuts and the excess interest rate spread paid by investors, resulting in no credit losses for the Treasury or Federal Reserve.”
This reveals that the Treasury investment was not for purposes of absorbing expected losses. The Fed could have taken over the Treasury’s subordinated position and still met its requirement of reasonably expecting to be repaid. The 2020 TALF would adopt a near-identical structure.
Redux: the COVID-era facilities
Given the Fed’s practice of avoiding unsecured lending, Treasury’s investments in the Fed’s pandemic-driven unsecured lending facilities could not be structured so as to be a total loss in expectation. While some critics of the COVID-era programs pushed for such, if the Fed had expected the Treasury investment to be wiped out, it would have been engaging in de facto unsecured lending.
That’s not to say that the Treasury/Congress can’t inject equity for the purpose of taking intentional losses, or that the Treasury wasn’t perhaps too concerned with avoiding losses in these programs. All else equal, loss-bearing injections from Treasury can enable Fed lending to be more aggressive.
However, it is the case that, under the Fed’s interpretation that additional protection is required in the case of unsecured lending, the Fed must still have that additional security after deducting from the Treasury funds any expected losses. That is, to the extent Treasury equity is serving as the required protection against an otherwise-unsecured loan, the Fed cannot ex ante expect it to be wiped out.
And if the Fed reasonably expects the Treasury investment to be repaid, the Fed is actually overcollateralized relative to its standard of needing to reasonably expect full repayment of its own loans.
This interpretation is further reinforced by the fact that the Congressional Budget Office (CBO) scored the Treasury’s CARES Act-funded investments in the Federal Reserve facilities as having a $0 impact on the federal deficit. The CBO noted, “Based in part on information from the [Fed] Board of Governors, CBO estimates that the income and costs will roughly offset each other.” Once again, the Fed could have met its “secured to the satisfaction” requirement that it expect to be repaid even if it were to take over the Treasury’s first-loss position. Its taking of Treasury funds for the facilities as designed left it overcollateralized.
The non-CARES facilities with Treasury funding tell a similar story. When the Fed restarted the AMLF in 2020, this time as the MMLF, it looked almost the same as last time. The difference was that the MMLF’s eligible collateral expanded to include unsecured CP and Treasuries/agencies. Despite unsecured CP being safer than ABCP as discussed above, the Fed this time took $10 billion of credit support from the Treasury for the MMLF, after using no public (or private) equity funding for the 2008 AMLF.
The 2020 reintroduction of the CPFF was a similar story; the facility no longer collected an up-front surcharge on unsecured paper—but used $10 billion of Treasury support.
Neither of these facilities had a fixed cap on the amount of lending they could do. It seems clear in these instances that the Treasury support was not based on some calculation of expected losses or necessary taxpayer protection, but rather was token security to enable the facilities’ unsecured lending. (The collateralized lending of the Primary Dealer Credit Facility, by contrast, took no Treasury funding.)
This structure allowed the Fed to again avoid confronting the issue of its undetermined authority to engage in unsecured loans. This exercise ultimately made the Fed’s commitment to emergency lending more fragile as the facilities were susceptible to Treasury’s and/or Congress’s early withdrawal of this support. This early withdrawal occurred with the CARES-funded facilities in November and December of last year. Thus, the Fed got the worst of both worlds by taking Treasury funds: no added tolerance for risk and early withdrawal of support.
Dodd-Frank drives it home
In the GFC, the Fed analysis deemed that 13(3) “likely” required some additional form of protection from losses when engaging in otherwise-unsecured lending. Irrespective of the tensions inherent in that conclusion illustrated above, the legislative history that they cited as the source of their reticence is not as clearly applicable to modern Section 13(3) lending. There is new legislative history.
For its near-80-year history leading up to the Dodd-Frank Act in 2010, Section 13(3) of the Federal Reserve Act referred to lending to any “individual, partnership, or corporation.” Following Dodd-Frank, the Fed can only set up 13(3) facilities with “broad-based eligibility.” It’s thus not clear that legislative intent from a time when the Fed could use 13(3) to lend to just one borrower is applicable to the modern 13(3). It indeed likely would have been inappropriate for the Fed to extend 11th-hour loans to Bear Stearns or Lehman Brothers on an unsecured basis, but individual loans (and loans designed to help a specific company avoid bankruptcy) are forbidden under the post-Dodd-Frank 13(3).
As Bernanke and Dudley noted, the collective interest revenue in a 13(3) facility is available to offset any losses on individual loans. The Fed’s Legal Division stated much the same with respect to the 2008 CPFF, saying, “any earnings on the CP in the CPFF. . . would serve as a source of funds to repay the CPFF SPV’s extension of credit from the Reserve Bank if losses result from the CP.” Moreover, as former General Counsel Alvarez said in an interview with the Yale Program on Financial Stability, “Of course, we also had other assets in the CPFF that we could liquidate in order to pay off any losses.”
As discussed earlier, many factors could leave a pre-Dodd-Frank 13(3) loan requiring very little collateral in the eyes of the Fed, even despite the Fed view that some additional protection was always necessary for the Fed to engage in unsecured lending. A post-Dodd-Frank 13(3) loan is inherently more secured by being part of a portfolio, and thus can require even less additional protection than before. A “broad-based” pool of loans can achieve the same level of security as an individual, partially-collateralized unsecured loan—provided, of course, that the Fed is still secured to its satisfaction.
The CARES Act facilities were a political lightning rod, with Treasury announcing in November it would pull back the supporting funds—against the Fed’s wishes to keep the programs open. The Fed, however, viewed itself as having no authority over the Treasury funds. Congress directly rescinded the support in December and raised the bar for how easily the Treasury can invest in Fed emergency lending facilities going forward.
Yet, given the above conclusions, the Fed may not need Treasury protection as readily as it has thought. In the future, if Treasury wants to invest in Fed facilities for the purpose of expected losses, the Fed should at least keep an open mind, as it would allow the Fed to lend more aggressively in pursuit of its mandate to stabilize the economy.
However, to the extent the Treasury funds are only there to enable 13(3) facilities to engage in unsecured lending and the Fed reasonably expects the Treasury to be repaid, the Fed should reconsider. The Fed can do that lending on its own.
Steven Kelly is a Research Associate at the Yale Program on Financial Stability