The Rescue of American International Group Module E: Maiden Lane III

Maiden Lane III was one of a set of government interventions arranged to assist AIG. It was announced alongside Maiden Lane II in November 2008 as a restructuring of government financial support (FRBNY n.d.2). These two SPVs were aimed at removing assets from AIG’s balance sheet to address continuing, significant liquidity drains and to improve its capital in the interest of avoiding ratings downgrades (Baxter and Dahlgren 2010, 4). In all, the FRBNY and Treasury would provide $182.3 billion in targeted government support to AIG, including loans, asset purchases, and capital investments (Massad 2012).

According to the legal experts at the Federal Reserve, the FRBNY, and the Treasury at the time, lending to institutions via a special-purpose vehicle “reflected use of the incidental powers conferred on the Reserve Banks” under the Federal Reserve Act (Alvarez, Baxter, Hoyt 2018, 12). Under Section 13(3) of the Federal Reserve Act, the FRBNY loan had to be “secured to the satisfaction of the lending Reserve Bank” (Title 12 U.S.C. 343, 2007). ML III had acquired the CDOs at a substantial discount to their pre-crisis value. The Fed believed the collateral would be “sufficient to repay the loan” once markets improved and the collateral’s value increased (Alvarez, Baxter, Hoyt 2018, 12).

To resolve AIG’s CDS issues, the FRBNY had three options: It could allow AIG to default on the contracts, it could continue to lend to AIG in order to fulfill collateral calls and related obligations, or it could restructure the contracts to stem the financial pressures (GAO 2011, 56). Letting AIG default on its contracts was not a feasible strategy, as it would force AIG into bankruptcy (Geithner 2010). This would perpetuate the disastrous economic consequences that were meant to be avoided through intervention and risk previously committed government funds (Geithner 2010). Continuing to lend to AIG was also not a viable option, as it would have increased AIG’s debt burden at a precarious moment, raising the potential for ratings downgrades and subjecting taxpayers to a potentially open-ended commitment with no additional security (Geithner 2010). Ultimately, the FRBNY settled on restructuring the contracts by purchasing the assets and canceling the insurance (Geithner 2010). This removed the CDS exposure from AIG’s balance sheet, thereby eliminating the possibility of default on the contracts, while stemming the continued collateral calls that were threatening a ratings downgrade (Baxter and Dahlgren 2010, 3-4).

The three options for restructuring AIG’s CDS contracts that were considered included the “as-adopted” structure, the “three-tiered” structure, and the “novation” structure. Each of these structures would use an SPV; however, the interventions taken by the SPV and sources of financing differed.

The as-adopted structure consisted of an SPV jointly funded by the FRBNY and AIG to purchase CDOs. After selling their CDOs to the vehicle, counterparties would cancel their CDS contracts (GAO 2011, 59). This was believed to be the simplest to implement and, among other options, had the greatest chance of attracting widespread participation among counterparties (GAO 2011, 60-61, 65).

The “three-tiered” structure would have provided for the creation of an SPV jointly funded by the FRBNY, AIG, and the counterparties (GAO 2011, 59-60). Counterparties would then sell their CDOs to the vehicle, cancel their CDS contracts, and retain some of the risk underlying the CDOs (GAO 2011, 59-60). Though there was no discussion with counterparties regarding this option, it was rejected because of the lengthy negotiations it would require with counterparties with no guarantee of success (GAO 2011, 61-62).

The “novation” structure would have also provided for the creation of an SPV (GAO 2011, 60). However, rather than canceling the CDS contracts, AIG would novate (transfer) its positions in the contracts to the SPV. CDS protection for counterparties would remain in place but would be paid out by ML III rather than AIG (GAO 2011, 60). The transferal would be cooperatively funded by the FRBNY, AIG, and the collateral previously posted by AIGFP to counterparties (GAO 2011, 60). In addition to retaining their CDO portfolios, counterparties would be required to pay premiums to ML III instead of AIG. They would surrender previously collected collateral to the ML III (GAO 2011, 60). The performance of the SPV and the value of the CDOs would be guaranteed by the FRBNY (GAO 2011, 60). This option was rejected, among other reasons, because of the difficulty in structuring an FRBNY guarantee that would have to be fully collateralized and capped in value (GAO 2011, 61-62).

In terms of policy considerations, the FRBNY wanted to ensure that it would be loaning against assets of value and that its senior loan would be repaid with interest in a timely manner, regardless of the performance of the CDOs (GAO 2011, 62). Speed of execution, both in establishing ML III and throughout the CDO purchase process, was also a key factor in the intervention’s design (GAO 2011, 62). Lengthy initial negotiations with AIG counterparties regarding the terms of ML III funding or the pricing of individual CDOs would have threatened the efficacy of the program (GAO 2011, 61-62). Furthermore, the FRBNY wanted to avoid the establishment of long-term relationships between itself and the counterparties, many of whom were banking organizations over which it exercised supervisory authority (GAO 2011, 61-62).

The Federal Reserve did not possess the authority to purchase the CDOs directly from AIG counterparties in order to restructure the CDS contracts (Title 12 U.S.C. 342, 111). It was, however, able to facilitate the senior loan to ML III, an SPV and independent legal entity that it created for that purpose (FRBNY 2017). Using an SPV to hold troubled assets—in this case, troubled CDOs—allowed the Fed to better manage the collateral and increase the likelihood of repayment (Alvarez, Baxter, Hoyt 2018, 12).

The allocation of residual funds specified in the original agreement was based upon the division of funding contributed to ML III and what FRBNY officials deemed to be a fair return for its loan and AIG’s equity contribution (GAO 2011, 66). Under alternative structures, AIG’s cash flow provision increased in proportion to its equity position, but at a disproportionately higher rate (GAO 2011, 66). To avoid the issue of consolidating ML III onto AIG’s books, advisers to the FRBNY determined it would only need to take over at least 55% of the remaining funds, in turn providing a 45% share for AIG. However, FRBNY officials deemed this would have provided an “extraordinarily high rate of return” on its $5 billion equity contribution, and the FRBNY was also aiming to avoid an equity contribution that was markedly higher (GAO 2011, 65, 66-67). Furthermore, the rating agencies were concerned that limiting AIG’s participation in any residual earnings would have left AIG in a weaker position (GAO 2011, 67). The FRBNY considered rating agencies’ fears along with tax considerations and market perceptions in the creation of the cash flow arrangements (GAO 2011, 67). The FRBNY was not overly concerned with its own stake of the residual earnings, preferring instead to provide stronger protection to its senior loan (GAO 2011, 67). While the split of the residual earnings favored the FRBNY, return on investment was not the FRBNY’s primary concern when structuring such a split (GAO 2011, 67).

As the controlling party of ML III, the FRBNY was tasked with the day-to-day management of ML III’s assets and engaged a number of vendors based on their expertise rather than developing internal departments for each need. However, the FRBNY did increase its internal expertise through targeted hiring in order to assist in decision-making and effectively evaluate recommendations from external vendors. The Investment Support Office department, which managed vendor relations, grew from just a few staff to a sizable business unit once all three SPVs were being managed. The FRBNY retained BlackRock Financial Management Inc. (BlackRock) to act as investment manager FRBNY n.d.2). BlackRock’s previous work with AIG regarding its swap portfolios and underlying CDOs made it an attractive option (SIGTARP 2009, 12).

FRBNY also hired Bank of New York Mellon as administrator and custodian, Deloitte and Touche LLP as external auditor for annual financial statements, and Ernst & Young LLP to perform closing work (Administration Agreement 2008, 2-7; FRBNY n.d.3; ML III 2011, 4-5). Although the FRBNY devoted significant attention to the implications of engaging outside vendors, there were a number of potential conflicts of interests that arose between the FRBNY, advisers, and counterparties, which were dealt with on an ad hoc basis (GAO 2011, 122).

The debt and equity structure of ML III was designed to guarantee that the FRBNY senior loan would be repaid even in times of great financial stress, and that the AIG equity contribution would be sufficient to cover losses and protect the senior loan (GAO 2011, 63-64). A stress test conducted by the FRBNY in November 2008 found that the ML III portfolio would still be worth $27 billion under an extreme stress scenario. While that would represent a loss of 57% of its notional value of $62.1 billion, that would still be sufficient to cover the FRBNY’s $24.3 billion loan (GAO 2011, 64).^FNThis analysis was conducted to measure the impact of different hypothetical stress scenarios, ranging from base, stress, and extreme, on the ML III portfolio. For the extreme stress scenario, the analysis assumed 56% housing price declines nationally and 75% declines in California (GAO 2011, 64).

Under the extreme stress scenario, AIG was still expected to recover $2.4 billion of its $5 billion equity contribution (GAO 2011, 64-65). The FRBNY did not require a higher equity contribution from AIG because it aimed to minimize the use of funds from the Revolving Credit Facility and avoid undermining the company’s position if the $5 billion were lost entirely (GAO 2011, 65). Alternative options for the one-time equity contribution, including a quarterly payment system or a secured loan through ML III, were rejected as too complex (GAO 2011, 65).

In addition to policy and investment concerns, the FRBNY carefully analyzed the impact of ML III on AIG’s long-term stabilization (GAO 2011, 62). It had to consider accounting rules that would have required AIG to consolidate any of the ML III structure onto its books (GAO 2011, 62). Fed officials felt that consolidating ML III on AIG’s books could have “injected volatility into AIG’s operations” (FRBNY n.d.2; GAO 2011, 62). As structured, ML III was therefore consolidated onto the Federal Reserve System’s financial statements (FRBNY n.d.2). Additionally, the FRBNY endeavored to create an intervention that would enable AIG to share in potential gains once federal loans and the company’s equity position had been repaid (GAO 2011, 62).

The interest rates on the FRBNY senior loan and AIG’s equity contribution referenced the one-month LIBOR (GAO 2011, 65). Since LIBOR was also the base rate used for many of the assets in the ML III portfolio, the FRBNY felt this was an appropriate interest rate (GAO 2011, 66). The FRBNY also wanted to leave open the possibility of selling the senior loan in the future and believed that the use of LIBOR and the wide spreads would be attractive to future investors (GAO 2011, 66).

The FRBNY and BlackRock selected only “CDO tranches referencing U.S. dollar-denominated cash bonds” for purchase by ML III FRBNY n.d.2). AIG was thus left with $12.5 billion in “potentially risky” multi-sector CDOs, most of which didn’t meet that eligibility requirement (US COP 2010, 74-75).

A significant number of the CDOs that ML III did not purchase were synthetic. Unlike cash CDOs, which hold portfolios of real assets like bonds or MBS, synthetic CDOs are essentially portfolios of long positions on credit derivatives (GAO 2011, 77). At the end of 2008, the notional value of AIG’s synthetic CDOs totaled $9.8 billion (US COP 2010, 75). In an interview with the GAO, the FRBNY expressed concern that synthetic CDOs “might not have met the Federal Reserve System’s requirement to lend against assets of value, given that they were not backed by actual assets” (GAO 2011, 77). One FRBNY official also cited the complicated nature of unwinding synthetic CDOs as another reason for their exclusion.

Some CDOs were excluded because they were not dollar-denominated. Euro-denominated trades totaling $1.9 billion in notional value (after conversion to USD) were excluded after reticence toward their inclusion from both the FRBNY and the counterparties (GAO 2011, 77).

Another $500 million in assets were excluded because various counterparties no longer held the underlying bonds and could not retrieve them for delivery to ML III (GAO 2011, 77).

Concessions from counterparties would have provided an additional layer of support against potential losses on the FRBNY senior loan (GAO 2011, 68). The FRBNY also argued that the discounts proposed were justified because the counterparties would profit from participation in ML III. In particular, they would be able to release into earnings any valuation reserves booked in connection with the CDS transactions, receive instant liquidity for the CDOs they held, and reduce risk-weighted assets on their balance sheets (GAO 2011, 68). They would also be able to avoid the risk of exposure to AIG on potential future declines in CDO value, as well as eliminate collateral disputes and the associated transaction costs and eliminate the costs of hedge protection on CDS contracts (GAO 2011, 69). Therefore, the combination of risk mitigation for the Fed loan and benefits from participation for the counterparties made concessions an attractive goal for the FRBNY in the creation of the intervention (GAO 2011, 68).

Before approaching the counterparties to negotiate concessions, the FRBNY asked an adviser to create different potential concession strategies (GAO 2011, 69). The adviser created three scenarios, each with specific methods and concession amounts (GAO 2011, 70). The first option would involve a 0.50% annual discount on the CDOs’ notional value until a credit event, up to a maximum of 3.00% (GAO 2011, 70). Under this method, individual counterparty concessions ranged from $2 million to $322 million, with projected total concessions equaling $1.1 billion (GAO 2011, 70). The second method was a flat discount of 2% on CDO notional value, resulting in individual concessions varying from $2 million to $328 million and total concessions equaling $1.3 billion (GAO 2011, 70). The final option was a 50% discount on collateral received leading up to the close of ML III, a total discount of 9.6% of the CDO portfolio (GAO 2011, 70). Under this method, individual concessions ranged from $0 to $2.1 billion and total concessions would equal $6.4 billion (GAO 2011, 70).

According to officials, between November 5 and November 6, the FRBNY had initially contacted 8 of the 16 counterparties with the largest exposures regarding the possibility of concessions (SIGTARP 2009, 15). All respondents expressed concern about the proposal for a number of reasons (SIGTARP 2009, 16). One issue that arose was that the combination of collateral the counterparties had already collected plus fair market value of the CDOs amounted to the par value of the CDOs; therefore, any concessions would result in voluntarily taking a loss (SIGTARP 2009, 16). In addition, counterparties expected that AIG would not default because of the government’s previous actions to avert AIG’s bankruptcy, and they were contractually entitled to the par value of the CDSs (SIGTARP 2009, 16).

The FRBNY would have preferred that AIG take the lead in these negotiations and felt that it was not appropriate to interfere with the sanctity of the CDS contracts while playing the role of creditor throughout the ML III process (YPFS Dahlgren Interview 2018; US COP 2010, 75; SIGTARP 2009, 19). If concessions were pursued, severe time constraints would not have allowed for the lengthy negotiations required to calculate individual haircuts, especially given AIG’s historical inability to negotiate tear-ups of the CDS contracts and the likely negligible size of the discounts in a successful scenario (Baxter 2010). Factors that would have had to be considered in calculating specific concessions included the varying amounts of collateralized and uncollateralized exposure to AIG, the methods of risk management employed previous to the crisis (e.g., some counterparties had bought hedge protection on AIG or AIG CDSs), the differences in credit ratings among the CDOs, and market perceptions on whether the government would continue to provide support (GAO 2011, 80-86). Aside from these complicated calculations, the FRBNY had little bargaining power because it could not threaten an AIG bankruptcy, as it had already begun to rescue AIG and had tens of billions of dollars at risk prior to beginning negotiations (Baxter 2010). Even in a scenario where the FRBNY would be able to threaten bankruptcy, the negative reception to this threat by rating agencies could have led to a downgrade exacerbating stress on AIG (Baxter 2010). An additional suggested strategy involved leveraging the regulatory powers of the Fed to coerce counterparties into accepting concessions (Geithner 2010). However, the FRBNY felt this coercion would negatively affect market confidence in AIG, could lead to potential downgrades, would be an abuse of authority, and would unfairly saddle Fed-supervised firms with the burden of concessions (Baxter 2010; Geithner 2010).

The FRBNY concluded that the correct course of action was to compensate the counterparties at market value. This decision proved to be controversial and led to significant testimony and investigation (Baxter 2010; Geithner 2010). In particular, the strategy would evoke substantial criticism from the Special Investigator General for the Troubled Asset Relief Program (SIGTARP) and the Government Accountability Office (GAO), among others (SIGTARP 2009, 29-30; GAO 2011, 56).

The loan made to ML III by the FRBNY was intended to be repaid from interest and principal payments received from assets if held to maturity, or the proceeds from their sale (GAO 2011, 9). The FRBNY decided not to set a predetermined schedule for selling assets; instead, it indicated a willingness to hold the assets to maturity if warranted (GAO 2011, 9). This afforded it the flexibility to liquidate assets when and if market conditions improved (GAO 2011, 9). If it was determined that liquidation was not the proceeds-maximizing option, ML III would be able to hold these assets to maturity, as the hold-to-maturity proceeds were predicted to be greater than the FRBNY’s senior loan (Baxter and Dahlgren 2010, 6). The assets of ML III had varying maturities and primarily consisted of multi-sector CDOs, which were backed by a combination of corporate bonds, loans, asset-backed securities, or mortgage-backed securities (GAO 2011, 9).

Since ML III’s inception, BlackRock was tasked with the below objective in addition to constantly monitoring market conditions.

“The Manager’s objective is to manage the Collateral in a manner consistent with repayment of the Senior Loan (including principal and accumulated interest), followed by the Equity Contribution Amount (including accumulated preferred distributions representing accrued interest) for as long as the United States Treasury maintains an economic stake in AIG on behalf of the United States taxpayer, while also meeting other obligations in the [payment] waterfall that are senior to the Senior Loan. In meeting the objective, the Manager should strive to maximize sale proceeds” (Investment Management Agreement 2010, 26).

Because of improved market conditions, ML III’s investment objective was revised in April 2012 to begin exploring the sale of assets, and on August 23, 2012, the FRBNY concluded its sale of ML III assets (FRBNY n.d.1).

The FRBNY thought it best to hold the CDO assets on ML III’s balance sheet until market conditions stabilized and any losses from sales could be minimized (FRBNY 2020). Selling the assets at auction would ensure equal opportunity for institutions to benefit from the sales, and for the FRBNY to determine that “the
winning bids represented good value for the public” (FRBNY 2012c). There was no fixed timeframe for the sales, which began in April 2012 and concluded at the end of August 2012 (FRBNY n.d.1).

Knowing that AIG was due to report a substantial loss for the third quarter on November 10, 2008, the government made the decision to announce its financial support restructuring on the same day (US COP 2010, 138; GAO 2011, 53). Credit agencies had notified the FRBNY that they would likely downgrade AIG in the wake of the disappointing earnings announcement absent any restructuring of the government’s rescue, and the potential for ensuing market turmoil led the FRBNY to communicate its plans earlier than it might have otherwise (GAO 2011, 53). At 6:00 a.m. EST on November 10, 2008, the Federal Reserve Board of Governors and the Treasury Department issued a press release that outlined a restructuring of financial support to AIG (BdofGov 2008). This restructuring included purchasing $40 billion of preferred shares in AIG using TARP funds, changes to the terms of the Revolving Credit Facility, and the introduction of ML II and ML III (BdofGov 2008). The release describes these measures as an attempt to “establish a more durable capital structure, resolve liquidity issues, facilitate AIG’s execution of its plan to sell certain of its businesses in an orderly manner, promote market stability, and protect the interests of the U.S. government and taxpayers” (BdofGov 2008).

AIG then filed two 8-K securities filings in December 2008 to formally announce ML III’s creation and implementation (GAO 2011, 94). Although AIG retained responsibility for its own filings, the FRBNY was able to review them and provide suggestions (GAO 2011, 101-102). One comment from the FRBNY regarding the December 24, 2008 8-K securities filings was that the Schedule A attachment, which contained counterparty and CDO deal information, be kept confidential (GAO 2011, 94). The FRBNY’s stated reasoning behind this request was that “the counterparty information was commercially sensitive for the parties involved but did not provide material information to investors, and that disclosure could hurt the ability to sell ML III assets at the highest price, potentially to the detriment of taxpayers and AIG” (GAO 2011, 96). The identities of the counterparties would prove to be particularly consequential, owing to their receiving par value for the CDOs (see Key Design Decision 11). According to the Government Accountability Office, the FRBNY did not believe that deal information regarding ML III would be publicly disclosed and assured counterparties in the transaction that their identities would be kept confidential (GAO 2011, 95). Ultimately, the Securities and Exchange Commission granted confidential treatment for only some parts of the attachment, and the treatment ended up being transitory, as AIG disclosed the entirety of the Schedule A attachment in its amended 8-K filing on January 29, 2010 (GAO 2011, 98-100).

Another example of the FRBNY influencing AIG’s securities filings occurred when the FRBNY’s outside counsel requested that the following language be removed from the December 24, 2008 8-K filing (GAO 2011, 100-101): “As a result of this transaction, the AIGFP counterparties received 100 percent of the par value of the Multi-Sector CDOs sold and the related CDS have been terminated” (GAO 2011, 101). The reasoning behind this change was that it mischaracterized the transaction, since only in combination with previously posted collateral did payments from ML III result in counterparties getting paid par value (GAO 2011, 101). The FRBNY’s involvement with AIG filings was indicative of the unique situation presented by ML III, where the FRBNY assumed the role of both a public institution and private market participant (GAO 2011, 102).

In the ensuing years, while managing ML III assets, the FRBNY was transparent in its reporting and announcements. In particular, the FRBNY provided annual audited statements detailing ML III’s holdings throughout its existence (ML III 2011, 4-5). Also, since ML III was consolidated on the FRBNY’s books, its quarterly fair value (plus normal accounting entries such as interest accruals, senior loan repayments, and expenses) and the outstanding loan balance were disclosed in the weekly H.4.1 release of the Federal Reserve’s balance sheet FRBNY n.d.2).