Why Does the Fed Really Use SPVs?
The Federal Reserve “borrowed a tool from financial engineers whose handiwork had led to the Great Panic. It created a ‘special purpose vehicle’ . . ." – David Wessel, In FED We Trust (2009)
It was March 2008, and the Fed was helping rescue Bear Stearns by facilitating its sale to JPMorgan (JPM). To get it done, JPM hived off $30 billion of Bear Stearns’s assets in a bankruptcy-remote special purpose vehicle (SPV)—$29 billion of which would be financed by the Fed, the remainder by junior funding from JPM. This would become known as Maiden Lane (named for the street next to New York Fed). It would be the first of three Maiden Lanes (with the second and third established as part of the AIG rescue); it was the first of five SPVs set up by the Fed during the Global Financial Crisis (GFC)—of the 7 SPV programs and 11 actual SPVs authorized. Since the start of the GFC through today, the Fed has provided emergency liquidity from 10 different SPVs.
Reputable discussion of the Fed’s use of SPVs—from journalistic coverage to legal articles to governmental reports—often take as received wisdom that the Fed sets up these vehicles to evade its legal limitations. However, the Fed itself has never said this anywhere. Indeed, the Fed’s communications surrounding its SPV interventions have mentioned the creation of SPVs without offering a justification.
A comprehensive look at the Fed’s history with SPVs reveals the oft-postulated story of legal evasion does not hold up. For one, the Fed’s history with SPV use and non-use is not consistent with that story. Additionally, formerly confidential internal documents from the time of the GFC lay out the Fed’s thinking and make clear it analyzed the use of SPVs within the bounds of its authorities. Moreover, several interviews conducted by the Yale Program on Financial Stability (YPFS) with past crisis fighters reveal more strategic reasons for using SPVs. This article lays out the legal underpinnings as well as the benefits and costs of using SPVs when fighting crises in the US.
The Mythology of the SPV Legal Runaround
The common-but-erroneous story goes like this: Because the Fed’s emergency liquidity authority provided by Section 13(3) of the Federal Reserve Act only authorizes lending, the Fed must set up a SPV if it wants to purchase any assets it can’t under its normal monetary policy authority. According to this line of thinking, the Fed then simply lends to the SPV, which executes the purchases on behalf of the Fed—effectively enabling the Fed to do an end-run around its statutory limits by still technically lending. For example, a government-produced report updated during the pandemic wrote the following in a section titled “Lending to Itself?”:
During the [GFC], the Fed structured many of its transactions under Section 13(3) as loans to an LLC or SPV that it created and controlled. It did so to comply with the Section 13(3) requirement that assistance take the form of a loan.
However, the chart below—showing every Section 13(3) program the Fed has authorized since the start of the GFC—already begins to tell a different story.
For one, you can see the Fed has also used SPVs for programs that only do lending—not just for asset purchase programs. In each case of the Fed establishing an SPV for a lending program—TALF, TALF II, and the PPPLF—it did not have recourse beyond the collateral. Yet, even that is not a throughline; the Fed also established nonrecourse lending programs—the Bear bridge loan, AMLF, and MMLF—without using an SPV.
Often, for legal reasons or additional risk protection/tolerance, the Fed’s 13(3) programs will leverage a layer of junior loss protection. As shown in the chart, this has taken various forms—equity, junior debt, or various forms of loss guarantees—however, they are near economic equivalents from the Fed’s perspective. In addition to funding from the TARP legislation during the GFC, the Fed often utilized private sector funding to provide this protection. Since then, during the COVID-19 pandemic and the Bank Crisis of 2023, the Fed’s junior protection has come exclusively from Treasury (UST) funds—either from the standing funds available in the Treasury’s Exchange Stabilization Fund (ESF) or funds appropriated by the Coronavirus Aid, Relief, and Economic Security (CARES) Act in March 2020.
The presence of this subordinated funding also does not explain the use of an SPV. As the table shows, there are several cases where the Fed receives subordinated protection but does not use an SPV.[1]
The Legal Limitations
Establishing Businesses?
The SPVs are structured as limited liability companies (LLC)—a common corporate form. As discussed below, using SPVs provides the Fed with some benefits for an intervention’s operations, transparency, and—in certain conditions—efficacy. Moreover, as alluded to above and shown below, an SPV is not legally required for the Fed to be able purchase assets under Section 13(3). The Fed has viewed the establishment of SPVs as falling under its “incidental” authorities. The regional Federal Reserve Banks are chartered as banks and, under Section 4(4) of the Federal Reserve Act, authorized to perform “all powers specifically granted by the provisions of this Act and such incidental powers as shall be necessary to carry on the business of banking within the limitations prescribed by this Act” [emphasis added throughout]. Establishment of SPVs is normal in the business of banking and, in this case, functions as incidental to the Fed’s exercise of its Section 13(3) authorities—which do not require the Fed to utilize such a structure.
Section 13(3) Loans vs. Purchases
Section 13(3) of the Federal Reserve Act does not say the Fed can only lend under its emergency authority. Rather, the law says the Fed may “discount.” Section 13(3) reads[2], in part:
In unusual and exigent circumstances, the Board of Governors of the Federal Reserve System, by the affirmative vote of not less than five members, may authorize any Federal reserve bank […] to discount for any participant in any program or facility with broad-based eligibility, notes, drafts, and bills of exchange when such notes, drafts, and bills of exchange are indorsed or otherwise secured to the satisfaction of the Federal Reserve bank …
It’s possible that the confusion stems from the word “discount”—an archaic banking term now most frequently applied to the discount window, a lending facility. But, historically, banks would “discount” paper by purchasing the short-term loan from a counterparty with the intent of collecting the principal and any interest (imputed or otherwise). It is economically no different from a loan to the ultimate obligor, and the purchase price need not have been at a discount to fair or par value.
Indeed, a then-confidential Fed legal memo from April 2008 describes the term as the Fed has long viewed it, citing past Fed publications as precedent (“IPC” stands for “individual, partnership, or corporation”):
A “discount” of a note for a counterparty under section 13(3) encompasses a broad range of transactions, including a simple advance to the counterparty on a note newly issued or made by the counterparty and a purchase of one or more third-party notes held by the counterparty. Specifically, the Board consistently has viewed the term “discount” under section 13(3) as including a Reserve Bank extension of credit to an IPC (a loan to an IPC by a Reserve Bank on the borrowing IPC’s own note) as well as a purchase by a Reserve Bank of third-party notes held by an IPC.
This memo laid out the legal reasoning behind the Fed’s first assistance program that made use of an SPV—the Maiden Lane transaction to assist JPMorgan in rescuing Bear Stearns.
Similarly, the Congressional Oversight Panel—citing discussion with Fed staff and a Federal Reserve Bulletin from 1958—wrote in 2010 report that “the term ‘discount’ has been interpreted broadly to refer to any purchase of paper (or essentially any advance of funds in return for a note) with previously computed interest.”[3]
What Can the Fed/SPV hold?
“It's easy to think of the SPV as the recipient of the loan and that the Fed is using 13(3) to lend to the SPV. But if you follow that logic, you get to two bad places. One is the Fed lending to the Fed, which is a bad place to be. And the second is then the SPV can hold anything, including real estate, including equity securities...”
– Scott Alvarez, 2004-2017 Fed Board General Counsel—from a 2022 interview with YPFS
Section 13(3) provides that the Fed may discount “notes, drafts, and bills of exchange” (henceforth, “notes”). While the statute no longer[4] has any limits on the types/issuers of notes, the restriction to notes limits the Fed to discounting instruments of credit. As the April 2008 memo notes, “a note is any written promise to pay a stated amount of money with or without interest or other charges.”
To be clear, that does not limit what the Fed may take as collateral for a loan—the loan is the note—but it inhibits what the Fed can purchase under its Section 13(3) authority. So, while equities or real estate (to keep Alvarez’s examples) may serve as collateral for Fed lending, the Fed cannot purchase them directly—neither on balance sheet nor through an SPV.
That is, the Fed can buy your car loan, but not your car.
In the Fed’s SPVs, one can “look through” the SPV and see that, in all cases, the Fed would have been able to do that liquidity provision directly on its balance sheet. In all the lending facilities, the loan itself functions as the “note,” while all the asset purchase facilities have only bought instruments of credit. As Alvarez characterized it when discussing the pandemic response:
“You could say in those situations if you needed to, ‘The SPV is the Federal Reserve. It's just a separate form of the Federal Reserve. And we are doing what we could do which is discount notes for individuals, partnerships, and corporations in a broad-based facility.’”
The Fed’s initial legal analysis, laid out in that April 2008 memo, makes exactly that case—that even looking through the SPV does not affect the Fed’s authority to take on the JPM assistance transaction. The Fed first analyzes the transaction in the more straightforward way that the Fed is discounting a direct obligation the SPV:
As explained above, for purposes of section 13(3), a discount of a note includes a purchase of an IPC’s own note, and a note is a promise to pay a sum of money. The FRBNY proposes to pay $29 billion to discount a note of the LLC…
Before going on:
In the alternative, if the FRBNY special facility were characterized as an acquisition by the FRBNY of the assets of the LLC or of Bear Stearns, the FRBNY special facility would still be a discount of notes of an IPC permitted under section 13(3). As discussed above, the Board consistently has viewed the term “discount” as including a purchase by a Reserve Bank of third-party notes held by an IPC. The assets of the LLC will consist of third-party notes that are eligible for discount under section 13(3).
Additional Distance from the Fed
With the Section 13(3) facilities that used SPVs to assist individual institutions—Maiden Lanes I, II, and III for supporting the JPM acquisition of Bear Stearns and the restructuring of AIG’s balance sheet—the Fed sought an extra layer of legal protection against the possible accusation of lending to itself, according to Alvarez. Despite that each facility ultimately only took on notes[5], the Fed asked for material junior protection from JPM and AIG. (This subordinated funding from JPM and AIG was also there to help the Fed reach its statutory requirement of being satisfactorily secured.) As Alvarez put it regarding Maiden Lane:
“The worry we had was setting up the SPV at that point was, ‘Well, the Fed had the paperwork done on the SPV; the Fed was going to be one of the managing partners. Was that going to open us to the argument that this was a subsidiary of the Federal Reserve Bank of New York? And so [FRBNY] was lending to itself.’ That's not an argument you want to have to worry about. So, JPMorgan put in what we considered to be the equity of that vehicle—a billion dollars, real serious money. And it was going to take the first set of losses. So, that was a serious injection, very much like a JPMorgan subsidiary.”[6]
Yet, part of the impetus for these transactions was to get the assets that ended up in the Maiden Lane SPVs off the JPM and AIG balance sheets—which was indeed what happened—as only a purchase by separate entity (as opposed to a collateralized loan) could effect. And while the large, equity-like funding helped further establish the SPVs as a separate entity from the Fed, the Fed still reported the SPVs on its consolidated balance sheet. As Alvarez said:
“The accountants, when they looked at that, they said, ‘Well, you lawyers can think of that as a subsidiary of JPMorgan; we don't know if it's a subsidiary of anybody. We don't know, but we're going to say JPMorgan doesn't have to report those assets, and the Fed does.’ So, that's fine. We didn't care much about the accounting. We didn't like it much. But we were okay with it.”
Tom Baxter, then the general counsel of the FRBNY, similarly expressed:
“When we created Maiden Lane I to facilitate the JPMC-Bear merger, this lawyer was thinking the SPV would stand with its own balance sheet, independent of the Federal Reserve. The accounting professionals taught me that, because of post-Enron reforms directed toward SPV accounting, the SPV needed to be reflected on the Federal Reserve’s balance sheet.”
The Fed has followed this practice with all its Section 13(3) programs that used SPVs.
The Dodd-Frank Act has since forbidden the Fed from crafting a Section 13(3) intervention for a specific institution (all these interventions must have “broad-based” availability). Yet, the strategy from these Maiden Lane facilities of getting additional separation from the Fed via significant outside funding may still be important going forward—for a situation in which the financial system needs substantial Fed purchasing of non-note collateral. To the extent such discounting can be done in an SPV that is sufficiently capitalized/owned by non-Fed parties, the Fed may be able to lend substantially into such a purchase vehicle.
Too Many SPVs or Not Enough?
The Fed has viewed use of the SPV structure as providing management, accounting, and legal advantages to an intervention—especially when the Fed operates multiple Section 13(3) programs in parallel. Each intervention has its own specific terms, timeline, capital structure, and management team. The management teams may also be in geographically separate reserve banks, depending on which one is administering a given intervention. The SPV structure simplifies the reporting of income and the management of any sales of assets discounted by the facility.
Moreover, the degree of corporate separation from both the Fed and other Section 13(3) interventions provided by an SPV structure may protect those other entities in the event of a lawsuit. As Alvarez put it:
“Then there's this corporate separation thing that lawyers like to worry about where, suppose one facility loses money and another facility makes money. And you have to go after somebody. If they challenge one facility, that doesn't damage—or it doesn't represent a challenge against any other facility. They're sealed off from each other. Profits of one facility don't go to pay off losses in another facility. The government, in the end, takes all of those together. But as a corporate legal matter, you don't have to worry about fighting off different people who want access to different levels of profit.”
The Forgone Super-SPV
As noted above, the Fed often receives subordinated loss protection of some kind when effecting Section 13(3) facilities, and this funding enables the Fed to do more liquidity provision than it could do in the absence of such support. Primarily, it allows the Fed to take on more asset risk. Section 13(3) requires that any assistance provided under the statute be “secured to the satisfaction” of the Fed, a standard which the Fed has consistently interpreted as mandating that it reasonably expect to be fully repaid—essentially, a confident expectation at the time of assistance of at least a 0% return on investment. Thus, to the extent the Fed receives junior protection from another funding source, it can move down the risk curve and provide more assistance.
The Fed could economize on this external funding by pooling it into one combined emergency liquidity SPV, essentially getting downside protection from the benefit of diversification. With additional protection against loss in any individual program from pooling all Section 13(3) programs, the Fed could stretch each external dollar of loss protection yet further—getting even greater enhancement to its ability to disburse liquidity.
Indeed, as pandemic-era Head of Markets at the New York Fed Daleep Singh told YPFS:
“Oh, we would've preferred to have one big pool, one SPV in which we had discretion to allocate the equity backing where it was most needed.”
But, the provider of the junior funding—in his case, the US Treasury—had some competing interests:
“Treasury was very much opposed to that, and they wanted a clean, transparent way of knowing exactly what portion of their equity injections was being used to support which markets and to what extent, under what terms. They thought that was really important for Congress to know.”
Michael Held, FRBNY’s general counsel from that time, said that by not pooling, “you do sacrifice a little bit more agility and flexibility.” However, he noted the Fed similarly had incentives cleanly report each program’s details, saying:
“You make the accounting and the transparency more complicated if it's all in one SPV, so I think the idea was that you're just being a little bit clearer on how the money is being used by creating separate facilities.”
This issue of disclosure has been a key consideration in the use of SPVs.
Simplifying Disclosure
Section 13(3) of the Federal Reserve Act requires the Fed Board to provide congressional committees with monthly reports regarding outstanding Section 13(3) credit, and the Fed typically elects to publish these reports on its website. The reports to the congressional committees require the Fed to disclose virtually all relevant details about the underlying liquidity provision, but the Fed chair can elect to keep identifying details of individual liquidity recipients confidential. In such a case, the confidentiality could only be in place for a maximum of one year after the closure of the assistance program—the maximum delay permissible under Section 11(s) of the Federal Reserve Act.[7] The Fed also publicly reports total dollar amounts for each program on a weekly basis as part of the usual publication of its overall balance sheet.
The Fed has provided separate annual financial statements for the SPV LLCs, and these statements are each independently audited by an outside accounting firm. While the Fed already provides a great deal of transparency on all its Section 13(3) interventions as noted above, the audited statements provide greater detail and transparency.
Indeed, with respect to why the Fed utilizes the SPV structure, Held said,
“It helps with the accounting, and the transparency, and keeping track of exactly how the public money is being used. That's number one.”
He went on:
“I think where there's no taxpayer money—don't get me wrong: all Fed money, I recognize, is money from the public—but the money coming from Treasury, that's where you really want to make sure there's clarity around how the funds are being used. And it makes for better, clearer accounting when it's separate facilities. You have separate financial statements; people would be looking for those. It's something that it's easy to point to with respect to each facility. So, it just provides a bit of clarity around what we're doing.”
Yet, former Boston Fed President Eric Rosengren told YPFS in an interview (publication forthcoming): “I do think it would make more sense to have the same level of transparency, but not the SPV structure,” to the extent using an SPV risks hampering the market impact.
Considering the Market Impact
As shown in the above chart, the AMLF and its pandemic-era successor facility the MMLF—which both lent nonrecourse to banks to purchase assets from money markets funds—did not utilize an SPV structure. (Both facilities were run out of the Boston Fed.) Rosengren noted that using an SPV would have dangerously delayed disbursing the announced liquidity provision. That is, it was important to start actually purchasing assets as quickly as possible, as opposed to just relying on the positive market effects of the announcement until the SPV could be set up and begin operating. Said Rosengren:
“We directly purchased the paper; it was not done through an SPV structure. The SPV structure is much more cumbersome to negotiate. It takes about six weeks of legal work. If you really have a crisis where you need to immediately get funds into the market, an SPV structure pretty much eliminates the possibility of doing that.”
He caveated his remarks, noting that the Fed does often benefit from strong announcement effects from establishing an intervention. That is, the market frequently stabilizes significantly when the Fed announces an intervention, even if the implementing SPV won’t be operational for some time. That said, the luxury of the automatic market reaction is not a given, depending on the severity and specifics of the situation and the facility. As Rosengren described:
“The New York Fed had a bunch of SPV structures, but they didn't start purchasing assets for four or five weeks, sometimes longer. The good news is that the announcement that the Fed was willing to buy assets brought in the spreads. If it was a situation where it was important to actually purchase the assets rather than stating intent to purchase the assets, those SPV structures would have seriously hampered how we could have responded.”
However, that the market now (at least until market memory fades) has “experience” with the Fed rolling out SPV structures may actually enhance the SPV structures’ effectiveness and close the gap with the nimbleness of non-SPV structures. As Rosengren noted about the lag between announcement and actual intervention caused by the time it takes to actually set up the SPV:
“I think in part because financial markets have been trained through the [GFC] that when the Fed said it was going to purchase an asset, it brought the [credit] spreads in on the announcement, that made [the lag] less of a serious problem.”
Held similarly talked about the market’s relative comfort with the structures now. Indeed, when the Fed made decisions during the pandemic on which facilities would be run out of SPVs and which weren’t, Held noted that one of the considerations was making the contemporary iterations of GFC-era facilities look like their familiar predecessors:
“For many of the old facilities, there was a little bit of wanting to replicate so that we don't cause unnecessary confusion in the market about why we're doing it differently. It wasn't like a need to do it in some instances, but there was a feeling that it worked before, and why do something different, particularly for the same types of facilities we did previously?”
Yet, the issue of speed and the Fed’s inability to simply rely on announcement effects arose again in the Bank Crisis of 2023. On the first weekend of the crisis, the Fed rolled out the Bank Term Funding Program—which provides loans to banks against the par value of some of their assets—without establishing an SPV. In a June 2023 speech, FRBNY General Counsel Richard Ostrander explained why, saying:
“There was not enough time to set up special purpose vehicles as the Fed had done for some of the pandemic programs. The only way to have the program up and running so quickly was to leverage our discount window facilities.”
Thankfully for effective crisis fighting, the Fed has never legally needed that setup time. The Fed’s emergency liquidity provision authority does not change based on whether the Fed uses an SPV or not. And, while SPVs offer advantages with respect to transparency, legal separation, operations, and other benefits around the edges of intervening, it is ultimately the unique details and demands of a given intervention that either permit the Fed or prevent it from utilizing an SPV structure.
[1] In post-GFC cases, the junior protection has always been structured as an equity injection when an SPV is present and as credit protection when one is not. Similarly, in GFC cases of the junior protection taking the form of credit protection, it was for non-SPV interventions. Crisis fighters questioned by YPFS consistently didn’t have an explanation for this pattern—and often showed little interest in the distinction given the economic equivalence of a US Treasury-provided guarantee versus a US Treasury-provided equity injection.
While it seems clear from this pattern that direct funds injections (equity or subordinated debt) are paired with SPVs, there was also partial funds transfer in the case of the MMLF. The MMLF received $10 billion of credit protection from the Treasury’s ESF and did not use an SPV; however, the ESF did transfer an initial $1.5 billion to the program. Notably, the pandemic-era Fed programs that received an equity injection from the Treasury (all with SPVs), received the injection only 15% in reserves—with the rest in nonmarketable Treasury securities. Thus, the actual amount of “cash” transferred to the intervention was the same as it would have been in the case of an equity injection into an SPV.
[2] The requirement to be “broad-based” was added by the Dodd-Frank Act of 2010, but the rest of the quoted bit was the same during the GFC. Prior to the amendment, the statute provided for discounting for any “individual, partnership, or corporation.”
[3] The Federal Reserve Bulletin from 1958 says, “the judicial interpretations of the word ‘discount’ show that the term is used very broadly. In practice the term ‘bank discount’ is applied broadly to transactions by which a bank computes interest in advance so that there is the possibility of compound interest, and it seems that any purchase of paper is a ‘discount’ in that sense since it permits such advance computation and compounding.”
[4] In 1991, following the 1987 market volatility, Congress removed the last restriction on the types of notes the Fed could discount under its emergency interventions. Section 13(3) had until then forbade the Fed from providing lender of last resort services against “investment securities.” In the Senate Banking Committee report accompanying the legislation (the otherwise mostly unrelated FDIC Improvement Act), the committee said of the restriction removal:
This clarifies that access to liquidity in special circumstances can be made available directly to a securities dealer to help preserve market liquidity and avoid market disruption. [. . .] With the increasing interdependence of our financial markets, it is essential that the Federal Reserve System have authority and flexibility to respond promptly and effectively in unusual and exigent circumstances that might disrupt the financial system and markets.
[5] Maiden Lane also took on the derivative hedges associated with the mortgage assets it assumed from JPMorgan (Bear Stearns). As with the creation of SPVs, the Fed saw the discounting of the hedges as “incidental” to the rest of the 13(3) intervention, and thus falling under its Section 4(4) authority. As the Fed wrote in a footnote of the memo mentioned in the text of this article:
A small amount of assets of the LLC that are not notes, drafts, or bills of exchange (for example, cash and hedging instruments) may be discounted by the FRBNY under the incidental powers provision of the Federal Reserve Act. In addition to the express powers of the Federal Reserve Banks set forth in the Federal Reserve Act, the Act provides that each Federal Reserve Bank has the authority to exercise “such incidental powers as shall be necessary to carry on the business of banking within the limitations prescribed by this Act.”
[6] He said similarly of Maiden Lanes II and III: “Both of those were designed in the same way with the same logic: that AIG has to put in a big slug of money that is equivalent to equity. And it's going to take the first loss on all of that, just like shareholders of a corporation would take the first loss. And so, in that way, we are thinking of that not as the Fed lending to the Fed, but the Fed lending really to AIG through this vehicle.”
[7] If the Fed chair elects to keep borrowers’ identifying details confidential, Section 13(3) stipulates that the information shall still be made available only to the “Chairpersons or Ranking Members” of the Senate Banking Committee and House Financial Services Committee.