United States: Bank of America Emergency Liquidity Program, 2009

As of September 30, 2008, BofA was the largest banking organization in the United States with $2.7 trillion in assets. Its primary bank subsidiary, Bank of America, N.A., was the second-largest bank in the country. The bank held more than 10% of US bank deposits, had total consolidated assets of $1.4 trillion, and was the largest issuer of insured deposits. BofA was a major supplier of credit to the US economy and, following its merger with Merrill Lynch, the largest securities broker in the world. In addition to deposits, the combined BofA organization had $426.3 billion of debt outstanding in commercial paper, other short-term debt, and long-term debt. BofA was also a counterparty to numerous financial institutions throughout the world (BofA 2009b; FDIC 2018; Fed 2009a).

According to Fed Chairman Bernanke, the Fed’s decision to assist BofA through its loan facility as part of the Asset Guarantee Program occurred “under highly unusual circumstances in the face of grave threats to [the] financial system and [the] economy” (US House Oversight Committee 2009b, 19). Against the backdrop of the Global Financial Crisis of 2007–2009 (GFC), BofA was negatively impacted by market conditions both directly and through Merrill Lynch upon its acquisition. On October 28, the Treasury had purchased $15 billion of BofA preferred stock using TARP funds (Fed 2009a). By year-end 2008, BofA’s low levels of tangible common equity had worried the market (FCIC 2011). At the beginning of 2009, BofA’s stock price had fallen 70% in 2008, and BofA was preparing to announce Q4 2008 results below market expectations (FDIC 2018). BofA’s stock price was declining, and the cost of insuring against the bank’s failure was rising (see Figure 3).

In December 2008, government officials were in communication with BofA about potential government assistance necessary to support BofA because of the firm’s assumption of losses from Merrill Lynch (FCIC n.d.). On January 16, 2009, BofA announced $1.8 billion in losses (not inclusive of losses resulting from its acquisition of Merrill Lynch)—its first quarterly loss in 17 years (Stempel 2009). The Fed said that, without a support package, the market reaction to BofA’s earnings could result in contagion:

The BofA program was modeled on the previous guarantee program that the authorities had created for Citigroup. Policymakers employed the ring-fencing strategy because it convinced markets that the government had taken extreme risk (“the left tail”) off of the bank’s balance sheet, was easily replicable, and would cost the government less than other strategies, such as further capital injections, asset purchases, or conservatorship. For a detailed discussion of the Citigroup Asset Guarantee Program, see Arnold (2025).

In response to market pressures on BofA and its systemic importance, the Fed approved a loan facility to BofA for 90% of the amount of mark-to-market and credit losses exceeding $21.1 billion accrued on a specified pool of assets (Fed 2009a). The Federal Reserve Bank of Richmond, which was to operationalize the loan facility, did not expect to lend under the BofA program given the size of the loss absorption provided by the Treasury and then the FDIC (GAO 2011).

However, the Fed’s participation was needed in the Asset Guarantee Program’s structure because it was the only agency that could provide a loan large enough to cover the entire asset pool. According to the Congressional Oversight Panel:

In other words, using TARP to cover the full amount of the Asset Guarantee Program would have burned through more than 15% of the entire $700 billion Congress had allocated to Treasury under the program.

On the other hand, the Fed's residual financing facility was not strictly necessary in a financial sense, according to Jason Cave, an official at the FDIC involved in the BofA AGP. In his view, the main value of the Fed's participation in the AGP was the announcement or signalling effect with the public and its sister agencies.

Regulators planned the assistance package to BofA in advance of its scheduled January 16 earnings call to preempt a negative market response (FDIC 2018).

Other Options

The government also considered as other options a government backstop of a capital raise, a government injection of common equity with limited control rights, and a capital offering with a ring-fence (instead of a government capital injection with a ring-fence) (BofA 2008; FCIC n.d.). FDIC regulators considered other options like a covered bond structure for BofA, but the Treasury and Fed thought that they would get more for each dollar of TARP funding by using a guarantee structure (Cave 2024).

Under the Asset Guarantee Program, the Treasury and FDIC agreed to guarantee ring-fence-eligible assets by taking second position on a pari passu basis, 75% to the Treasury and 25% to the FDIC. Under this arrangement, BofA would bear the first $10 billion in losses on the asset pool. After that, BofA would bear 10% of the second-loss layer, with and Treasury the FDIC providing the other 90% of the second-loss layer. While the Treasury and FDIC would absorb those losses on a pari passu basis, 75% of the loss guarantee would be covered by the Treasury and 25% by the FDIC. In compensation for their loss protection, BofA would issue the Treasury and FDIC $3 billion and $1 billion, respectively, in preferred stock bearing an 8% coupon. Treasury also received warrants to purchase common stock with an aggregate exercise value of 10% of the total amount of preferred issued (Fed 2009a).

At the same time as it agreed to the AGP, Treasury agreed to purchase $20 billion in senior preferred stock bearing an 8% coupon with TARP funds through the Targeted Investment Program (FDIC 2018). BofA received the TIP capital injection on January 16, 2009. As required by the TARP enabling legislation, the Treasury also received warrants for the purchase of common stock at a strike price of $13.30 per share, for an aggregate value of $2 billion (SIGTARP 2009). (For details on the TIP, see Hoffner and Arnold [2024].)

BofA also utilized several other broad-based crisis response programs. It had already borrowed substantial amounts from the Fed’s Term Auction Facility (an auction-based discount window liquidity facility), the Primary Dealer Credit Facility, the Term Securities Lending Facility, and the Commercial Paper Funding Facility (FCIC 2011; Fed 2016). It also issued $31.7 billion in FDIC-guaranteed senior debt under the Temporary Liquidity Guarantee Program. In addition to utilizing these crisis programs, BofA borrowed $92 billion from the Federal Home Loan Bank System, which comprises government-sponsored enterprises (FCIC 2011).

Pursuant to the term sheet, regulators would have agreed to assign the ring-fenced assets a favorable 20% risk weighting, the standard risk weighting for AAA-rated corporate credits under Basel capital standards (BIS 2004; Fed 2009a). Although the company did not release details, the troubled assets in the asset pool would likely have carried risk weightings of closer to 80% based on the information the bank did provide.^FNBofA said the Asset Guarantee Program lowered its risk-weighted assets by $70 billion. The 20% risk-weighting on the $118 billion asset pool is $24 billion. So, their original risk-weighted assets were roughly $94 billion ($70 billion $24 billion), or about 80% of $118 billion. The favorable regulatory treatment would have lowered BofA’s capital requirements on the ring-fenced assets by from about $7.5 billion to $1.9 billion.^FNThe standard 8% capital charge would result in a $7.5 billion capital requirement based on $94 billion in risk-weighted assets and $1.9 billion based on $24 billion in risk-weighted assets (O’Connor, Bush, and Mayo 2009). BofA said that the risk-weighting adjustment as a result of the Asset Guarantee Program would have lowered its risk weighted assets by $70 billion. The regulators’ decision to grant a 20% risk weight to the asset pool thus appears to represent significant regulatory forbearance—they were allowing BofA to hold just $1.9 billion in capital against an asset portfolio on which BofA remained exposed to the first $10 billion in losses. BofA said that the BofA package—including the $20 billion in TIP, the $4 billion fee paid in preferred shares, and the regulatory relief—would have improved its Tier 1 capital ratio on a pro forma basis from 8.7% to 10.7% (O’Connor, Bush, and Mayo 2009). 

The Fed’s role, through the loan facility, was not specifically necessary for this risk-weighted asset treatment. It was not strictly necessary in a financial sense either, according to Cave at the FDIC, since nobody expected the bank to exercise the loan. In Cave’s view, the main value of the Fed’s participation in the AGP was the announcement or signaling effect with the public and its sister agencies (Cave 2024).

Section 13(3) of the Federal Reserve Act (FRA, 12 U.S.C. § 343) provided the legal authority for the Fed to create and lend through the loan facility. The Fed’s Board of Governors voted to invoke Section 13(3) on January 15, 2009 (Fed 2009a). Section 13(3) permitted the Fed, “in unusual and exigent circumstances,” to provide liquidity “for any individual, partnership, or corporation,” provided the assistance was “secured to the satisfaction” of the Fed (US Congress 2008, sec. 13[3]). The invocation of Section 13(3) allows the Fed to provide liquidity more broadly than its monetary policy and discount window authorities allow (US Congress 2008, secs. 10B, 13, 14).^FNAs noted in Overview and Key Design Decision No. 10, Balance Sheet Protection, the asset pool contained a significant amount of derivative assets. Section 13(3) authorizes the Fed to provide liquidity only against note collateral—that is, instruments of credit—a classification that would typically not apply to derivatives. While the loan facility would technically be a loan to BofA, its nonrecourse status generally would negate that fact. However, this legal risk would be mitigated by the Fed’s recourse to BofA for interest and fees; the junior protection from BofA, Treasury, and the FDIC; and the Fed’s “incidental” authorities under Section 4(4) of the FRA, as it could be argued that taking the derivative collateral functioned as incidental to the Fed’s 13(3) authority. See (Alvarez 2022; Alvarez et al. 2008; Kelly 2023).

Fed legal staff and policymakers consistently interpreted Section 13(3)’s “secured to the satisfaction” requirement as meaning that the Fed had to be reasonably confident of full repayment. In other words, while the Fed might take some losses in 13(3) lending, in order to provide liquidity under Section 13(3), it had to make the ex ante determination that it would be repaid (Alvarez 2022). According to the GAO, the Fed didn’t expect it would be required to extend any credit under the facility, based on the analysis of a third-party vendor projecting that losses would not exceed $20 billion in a stressed scenario (less than the amount that BofA, the FDIC, and the Treasury had committed to cover before the Fed facility would be required).

Given that the BofA Asset Guarantee Program had a maturity of five to 10 years (see Key Design Decision No. 9, Loan Duration), it was possible that the Fed’s loan facility lending commitment could have been activated after the “unusual and exigent circumstances” required by Section 13(3) had passed (US Congress 2008a, sec. 13[3]). The Fed’s view on this, according to then–General Counsel Scott Alvarez, was that if the ring-fenced losses had penetrated through the junior layers to the point where the Fed had to lend into the ring-fence, that situation itself would satisfy the “unusual and exigent circumstances” requirement (Alvarez 2022).

The Fed’s loan facility, to the extent that it was a constituent part of the broader BofA package, was the product of “significant and intense” intragovernmental negotiations involving the Treasury, FDIC, and Office of the Comptroller of the Currency in January 2009 surrounding key terms of the package, including duration, scope, and fees (US House Oversight Committee 2009b, 19). According to Jason Cave, then at the FDIC, the BofA AGP was significantly more complicated to design than its Citi predecessor because of the opacity of the assets it had at the time (see Figure 8 at Appendix) (Cave 2024).

While the Fed never lent under the loan facility (see Key Design Decision No. 9, Loan Duration), the FRBR was the designated administrator of the facility (Fed 2009a).^FNBank of America is headquartered in Charlotte, North Carolina, which is in the FRBR’s district. The FRBR never finalized an agreement with BofA to provide the loan facility (GAO 2011). The GAO referred to the BofA lending commitment as an “agreement-in-principle” (GAO 2011, 185).

On January 9, the FRBR engaged PIMCO to provide valuation services in connection with the loan facility for a fee of $12 million. There was no competitive bidding process because of the time constraints. On January 16, the FRBR also engaged Ernst & Young to conduct due diligence in connection with the loan facility for a fee of $10.6 million (without a competitive bidding process). While the FRBR paid all vendor fees, BofA was required to reimburse the FRBR for those fees (GAO 2011).

BofA was eligible to draw on the loan facility if and when it accrued losses (unrealized mark-to-market losses and realized credit losses) on the asset pool exceeding $21.1 billion. BofA could terminate the loan facility commitment and prepay any outstanding loans at any time, subject to the Fed’s approval (Fed 2009b).

Section 129 of the EESA, passed on October 3, 2008, required the Fed to report to the Committee on Banking, Housing, and Urban Affairs of the Senate (Senate Banking Committee) and the Committee on Financial Services of the House of Representatives (House Financial Services Committee) on any use of its Section 13(3) authority within seven days of invoking that authority (see Key Design Decision No. 6, Communication and Disclosure) (US Congress 2008b, sec. 129[a]).

In July 2011, the GAO published a report on the Fed’s emergency programs, including an appendix specifically on the BofA loan facility (“agreement in principle”) (GAO 2011). The GAO’s findings were that:

• FRBR effectively implemented conflicts of interest controls in its BofA program, for example excluding BlackRock as a service provider due to BofA’s significant stake in BlackRock; and
• FRBR did not use a competitive procurement process to award contracts to the private sector entities hired to provide services (see Key Design Decision No. 4, Administration) (GAO 2011).

The GAO report did not include any further commentary on the Fed decision-making in its implementation of the loan facility (GAO 2011).

On June 11 and June 25, 2009, Chairman Bernanke testified before a congressional House committee about the Fed’s involvement in BofA’s acquisition of Merrill Lynch (US House Oversight Committee 2009a; US House Oversight Committee 2009b).

On January 16, 2009, the Fed issued a joint press release with the Treasury and FDIC announcing the BofA package. The press release said that, “if necessary, [the Fed stood] ready to backstop residual risk in the asset pool through a non-recourse loan” (Fed, FDIC, Treasury 2009). Policymakers timed this announcement to try to head off the adverse market reaction they expected to BofA’s announcement of its Merrill Lynch losses (FDIC 2018). The press release said that the USG’s support package to BofA was necessary to protect US taxpayers and strengthen the financial system (Fed, FDIC, Treasury 2009). On the same day, BofA held its fourth-quarter 2008 earnings call, in which then–CEO Ken Lewis referred to the Asset Guarantee Program (though he did not explicitly mention the Fed’s loan facility) and the assurance that it gave to BofA to move forward with the Merrill Lynch acquisition on its original terms (O’Connor, Bush, and Mayo 2009).

The Financial Times reported the BofA package after the early-morning announcement. The article described the BofA Asset Guarantee Program as similar to the Citigroup Asset Guarantee Program, which occurred on January 15, 2009 (see Arnold [2025]). Before the announcement, rumors had been spreading that a large US bank would be nationalized (Tucker 2009).

In its 2009 annual report, the Fed disclosed that in September 2009, BofA paid a fee to exit the arrangement (Fed 2010).

BofA announced its intention to terminate the AGP on May 7, the same day it announced its results in the Fed’s stress test. According to Joe Price, then the chief financial officer, in that statement:

In its 2009 annual report, BofA disclosed that it had reached an agreement to terminate the term sheet for the agreement-in-principle with the USG; it did not provide any details on the loan facility or the USG Asset Guarantee Program (BofA 2010).

Section 129 of the EESA required the Fed to report to the Senate Banking Committee and House Financial Services Committee on any use of its Section 13(3) authority within seven days of invoking that authority, including (1) its justification for invoking Section 13(3) and (2) the specific terms of any Fed actions, specifically, “the size and duration of the lending, available information concerning the value of any collateral held with respect to such a loan, the recipient of warrants or any other potential equity in exchange for the loan, and any expected cost to the taxpayers” (US Congress 2008b, sec. 129[a][2]). Pursuant to its Section 129 reporting requirements, the Fed issued a Section 129 report on its loan facility on January 15, 2009 (before its announcement to the public) (Fed 2009a). In its Section 129 Disclosure, the Fed said that the AGP—of which its loan facility was part—would “help restore confidence” in BofA and would “promote financial stability” (Fed 2009a, 3). The Fed also said that it did not expect its loan facility to result in any losses to the Fed or the taxpayer and presented other disclosures required by Section 129 (subsection [a][2]). The Section 129 Disclosure included the term sheet, which provided a broad description of the assets eligible for guarantee but stipulated that no assets would be assigned to the asset pool until BofA and the USG had agreed to effectuate the AGP. 

The Fed made its January 15, 2009–dated Section 129 Disclosure publicly available on February 18, 2009. The FDIC's January 15, 2009–dated memorandum was made publicly available upon the publication of the Financial Crisis Inquiry Commission Report in 2011.

The Fed approved the loan facility to BofA for 90% of mark-to-market and credit losses exceeding $21.1 billion accrued on a $118 billion specified pool of assets (Fed 2009a). The Fed funds liquidity provision through the creation of central bank reserves.

After the junior layers were exhausted ($10 billion BofA layer and $11.1 billion split 90% to the FDIC/Treasury and 10% to BofA), the Fed would have lent BofA 90% of $97 billion ($87.3 billion) to cover the remaining value of the asset pool, assuming the asset pool’s size didn’t change. BofA would have been responsible for reimbursing the Fed for 10% of any losses on the loan facility’s portion of the asset pool (GAO 2011).

The size of the loan or loans that the Fed would extend to BofA once losses passed the $18 billion threshold isn’t clear from the documentation. The documentation for the Citi package, in contrast, states clearly that the Fed would have immediately extended a loan equal to the entire remaining value of the asset pool as soon as its threshold was reached. While authorities typically described the BofA deal as mirroring the Citi deal, it’s possible that they planned to structure the BofA deal slightly differently, allowing BofA to borrow multiple tranches when losses occurred. The initial term sheet for the BofA transactions leaves open that possibility. It says that BofA could “draw on [the] Federal Reserve loan facility” when the threshold was passed, and it says the Fed’s fee would be based on the undrawn amounts (Fed 2009b). Moreover, Citi’s term sheet refers to “a non-recourse loan,” while BofA’s term sheet refers to “any Federal Reserve loans” (Fed 2008a; Fed 2009b) [italics added].

In negotiations to finalize and execute the term sheet, the USG and BofA subsequently reduced the asset pool from $118 billion to $104 billion (which would have reduced the Fed’s share to $83 billion). Thereafter, the parties were unable to reach agreement on the definition of eligible losses for $42 billion of those assets. As a compromise, they reduced the asset pool by half that amount, $21 billion, resulting in an adjusted size of $83 billion for the asset pool. The Fed’s ultimate loan commitment was $55.8 billion after the $21 billion loss-sharing by BofA, Treasury, and the FDIC, and BofA’s 10% loss buffer (Treasury 2009). Figure 4 shows the evolution of the asset pool size from its original announced size to its size at termination.

The Fed would have charged 20 bps on all undrawn amounts, had the parties finalized the agreement (Fed 2009b). Assuming the Fed’s ultimate maximum exposure was $62 billion, less the 10% BofA loss-sharing (for an ultimate loan size of $55.8 billion), this would have implied an annual fee of $111 million (Treasury 2009).

If BofA had accessed the loan facility, the Fed would have charged a floating rate of three-month Overnight Index Swap (OIS) rate plus 300 bps on the entire amount of the facility (Fed 2009b). Based on the Fed’s commitment fee calculation, that amount would have been the Fed’s ultimate loan size (equal to 90% of the Fed’s overall loss layer, ultimately $55.8 billion—see Figure 5 row [c]) (Treasury 2009). The OIS rate is a market-determined rate representing expectations for the Fed’s policy rate over different tenors—in this case, three months. At the time, the Fed’s “Regulation A”—the regulation it wrote to guide its implementation of Section 13(3) of the Federal Reserve Act—called for Fed emergency liquidity to carry an interest rate “above the highest rate in effect for advances to depository institutions”—a reference to the Fed’s discount window (Fed 2008c, vol. Code of Federal Regulations, title 12, sec. 201.4). At the time, the discount window’s primary credit program charged the Fed’s effective policy rate plus 35 bps; the rarely used secondary credit program was priced at the policy rate plus 85 bps (Fed 2023; FRBSL 2023).

BofA ultimately terminated the Asset Guarantee Program before the parties finalized it. Upon the termination of the arrangement, BofA agreed to compensate the government on a pro-rated basis for its “willingness to bear the risk of loss on the Pool during the period from the date of the term sheet to May 6, 2009” (Treasury 2009). On September 21, 2009, BofA agreed to pay a termination fee of $425 million to the government.

The Fed’s portion of the $425 million fee was $57 million. That included $34 million for its commitment to lend against the pool and $23 million to cover Fed operational expenses. The $34 million represented the pro rata amount of the 20 bps fee for the 110 days in which the negotiations took place from January 16 to May 6; it was charged on the net Fed loan exposure of $55.8 billion (which represented 90% of the Fed’s overall loss layer—see Figure 5) (Treasury 2009). The Citigroup Asset Guarantee Program—on which the BofA ring-fence was designed—did not feature a commitment fee (Arnold [2025]).

The loan facility’s maturity schedule would have aligned with the durations of the asset guarantees: as much as 10 years for residential mortgage-related assets and five years for other assets (Fed 2009a).

On May 6, 2009, BofA expressed interest in terminating the assistance package. BofA and the government had never signed definitive documentation for the arrangement, and BofA had never availed itself of the loan facility or any other component of the Asset Guarantee Program. However, BofA agreed to compensate the USG for the period between the date of the term sheet (January 15) and May 6. BofA and the USG signed a termination agreement on September 21, 2009 (Treasury 2009; Treasury n.d.).

According to the Fed, there were two primary reasons that BofA decided to terminate the Asset Guarantee Program. First, BofA wanted loss coverage for unrealized as well as realized losses, and the lack of protection for mark-to-market losses was unappealing to BofA; while the Fed’s loan facility considered mark-to-market losses eligible for the determination of lending size, the Treasury and FDIC did not—their guarantee was only for realized losses. It is possible that this provision was still under negotiation between BofA and the government parties (or between the government parties internally) when the parties terminated the program. Second, BofA was one of the financial institutions subject to the ongoing Supervisory Capital Assessment Program (SCAP), the stress test exercise that the Fed had started in February 2009—so both parties had agreed to delay negotiations on finalizing the Asset Guarantee Program until after the test was completed in May. The SCAP was a stress test for the largest banks in the country to determine which banks would need more capital to withstand a severe downturn. The result of the SCAP showed that BofA had a $33.9 billion capital shortfall; however, the bank was able to raise $35.9 billion in capital on private markets through asset sales, share conversions, and the issuance of new shares over eight months. Based on the SCAP results, BofA decided the Asset Guarantee Program was not sufficiently valuable to finalize (GAO 2011; Lawson 2021; NYT 2009).

The FRBR never expected to lend through the loan facility. The Fed would not have to lend until BofA made $21.1 billion in losses, but an independent analysis from PIMCO showed that, even under a stress scenario, expected losses on the asset pool did not exceed the $21.1 billion mark above which FRBR would be obligated to lend through the loan facility; see Figure 5 (GAO 2011; Glassman et al. 2009).

Sources: Glassman et al. 2009; author’s analysis.

Sources: Glassman et al. 2009; author’s analysis.

Our research did not uncover any specific impact of the loan facility on monetary policy transmission. As mentioned, the loan was never issued.

In connection with the loss-sharing agreements in the Asset Guarantee Program, BofA was required to maintain a foreclosure-mitigation policy in addition to complying with executive compensation and corporate expenditure policies (Fed 2009a). According to the term sheet, “an executive compensation plan, including bonuses, that rewards long-term performance and profitability, with appropriate limitations, must be submitted to, and approved by” the US government (Fed 2009a, 8–9).

Eric Rosengren, then president of the Federal Reserve Bank of Boston, said to Chairman Bernanke that the Fed should consider replacing BofA’s management team before adopting the ring-fencing strategy (ultimately, the Fed did not pursue this option) (US House Oversight Committee 2009b).