AIG — Financial Products credit default swap book and 2008 near-collapse

1. Episode summary

American International Group (AIG) was, at the mid-2000s, one of the largest insurance holding companies in the world, carrying a AAA credit rating and a balance sheet that spanned life, property, and casualty lines across more than one hundred countries. Within the group sat AIG Financial Products (AIGFP), a London- and Connecticut-based derivatives unit founded in 1987 to monetise the parent's AAA rating in the swaps market. By 2005 AIGFP had built a portfolio of credit default swaps on super-senior tranches of multi-sector collateralised debt obligations totalling, by year-end 2007, roughly $527 billion notional, of which approximately $78 billion was written on CDOs containing subprime residential mortgage exposure. The contracts contained collateral-posting triggers keyed to AIG's credit rating and to the market price of the referenced CDOs. In late 2005 AIGFP stopped writing new subprime-linked CDS. From mid-2007 onward counterparties — led by Goldman Sachs — demanded collateral against mark-downs; AIG contested the marks. In September 2008 the three major rating agencies downgraded AIG, collateral calls rose from $23.4 billion on 12 September to $32 billion by 15 September, AIG could not fund the shortfall, and on 16 September the Federal Reserve extended an $85 billion secured credit facility in exchange for a 79.9 per cent equity interest. The strategic question was whether AIG had understood the tail-risk economics of super-senior CDS written against housing-cycle collateral on a AAA-funded balance sheet.

2. Sources

Primary:

1. Financial Crisis Inquiry Commission, Final Report of the National Commission on the Causes of the Financial and Economic Crisis in the United States, Chapter 19 "September 2008: The Bailout of AIG", U.S. Government Printing Office, January 2011.
2. Financial Crisis Inquiry Commission, Final Report, Chapter 14 "Late 2007 to Early 2008: Billions in Subprime Losses", January 2011.
3. U.S. Senate Committee on Banking, Housing, and Urban Affairs, hearing transcript American International Group: Examining What Went Wrong, Government Intervention, and Implications for Future Regulation, S. Hrg. 111-291, 5 March 2009.
4. FCIC Staff Memorandum, AIG/Goldman Sachs Collateral Call Timeline (Updated), held at the Federal Reserve Bank of St. Louis FRASER archive, 2010.
5. Testimony of current and former AIG risk officers, AIG Risk Management, FCIC hearing record, 30 June 2010.

Secondary (with justification):

1. Robert L. McDonald and Anna Paulson, "AIG in Hindsight", NBER Working Paper No. 21108, April 2015 — peer-reviewed reconstruction of the AIGFP CDS and securities-lending losses using FCIC and Federal Reserve data; synthesises primary filings into an integrated account.
2. William K. Sjostrom Jr., "The AIG Bailout", Washington and Lee Law Review 66(3), 2009 — legal-academic analysis synthesising SEC filings, credit agreements, and contemporaneous reporting on the CDS contracts and collateral mechanics.
3. Michael Lewis, "The Man Who Crashed the World", Vanity Fair, August 2009 — investigative journalism drawing on interviews with AIGFP insiders including Gene Park; used as a secondary reconstruction of internal decision-making at the unit.
4. Gretchen Morgenson and Louise Story, "The Fall of AIG: The Untold Story", Institutional Investor / New York Times reporting, 2008–2010 — contemporaneous investigative reconstruction of the valuation dispute with Goldman Sachs.

Tertiary (flagged):

1. Wikipedia article "AIG Financial Products" — tertiary; consulted for frame and source-discovery only, not for load-bearing factual claims.

3. OTA narrative

Observe. The observation apparatus inside AIGFP did produce the signal that the portfolio's subprime exposure had become dangerous: in 2005, AIGFP executive Gene Park reviewed the composition of the CDOs being insured, concluded that the subprime share was far higher than the unit's modelling had assumed, and persuaded Joseph Cassano to stop writing new super-senior CDS on subprime-linked CDOs before year-end 2005. That observation was, in narrow terms, correct and timely. What the observation apparatus did not produce was a re-reading of the roughly $78 billion book already on the balance sheet: the collateral-trigger mechanics, the rating-agency linkage, and the mark-to-market exposure on the existing contracts were not reframed as live risks after the stop-writing decision. The FCIC record also documents that the Office of Thrift Supervision, AIG's consolidated regulator, did not independently generate the signal. The first-order observation — that the book contained subprime risk — was made. The second-order observation — that the existing book could force cash collateral under a housing-price decline regardless of whether new business was written — was not made inside AIGFP or at the parent. Observe was partially correct, partially incomplete; it was a transmission step rather than the operative root cause of the 2008 outcome.

Think. The reasoning from the evidence was the dominant root-cause phase in this episode. AIGFP, and the parent risk function that relied on it, treated super-senior CDS as economically equivalent to catastrophe-insurance lines in which actual default losses would be near-zero and in which rating-trigger and mark-to-market collateral-posting obligations were secondary operational details rather than first-order tail risks. The frameworks required to reason correctly about those contracts — collateral mechanics under a rating downgrade, the correlation of a AAA-funded insurer's own rating with the very housing-linked assets its swaps referenced, and the liquidity-not-solvency character of the resulting cash demand — were available in the derivatives and bank treasury literature that AIG's own counterparties were applying to the same contracts. The correct interpretive tools existed; they were not applied to the existing book even after the stop-writing observation in late 2005. This is an Easy-Wrong Think: the reasoning frameworks were accessible to the Archetype peer group of large derivatives-dealer counterparties and were demonstrably in use across the street. The reasoning failure was the root cause, and it was wrong in the easy sense.

Act. Execution during the crisis itself — the 2007–2008 collateral dispute, the attempts to raise capital, the rating-agency engagement, and the counterparty negotiation under duress — was constrained but not incompetent relative to the position AIG had already created. The valuation dispute with Goldman Sachs was contested through normal ISDA-style channels; management engaged the Fed and Treasury in the days before 16 September; securities-lending unwinds proceeded under distressed conditions. Act is not the root-cause phase in this episode: by the time the execution window opened in mid-2007, the size of the book and the rating-linked collateral triggers had already determined the outcome under any continuation of the housing-price decline. Act was the transmission step between a reasoning failure that had compounded over several years and a liquidity event whose magnitude execution could not compress. Act was not a root cause; execution operated under external constraints the prior Think failure had already set.

4. Modality evidence

Direction. A discrete, datable, attributable strategic choice sits in the record: the 1987 founding thesis of AIGFP as a vehicle to monetise the parent's AAA rating in the derivatives market, extended in the early 2000s into super-senior CDS on multi-sector CDOs. The late-2005 Park / Cassano decision to stop writing new subprime-linked super-senior CDS is also a specific, dated, attributable choice, cited in the FCIC final-report chapter on AIG (Ch. 19) and in contemporaneous investigative reporting (Lewis, Vanity Fair, 2009). What the record does not contain is an equivalent dated commitment to reduce, hedge, or restructure the roughly $78 billion of existing subprime-linked super-senior CDS after that 2005 stop-writing decision.

Scoring note (zero-modality rationale): the directional layer described in this subsection is acknowledged in the §4 evidence as present and specific but is not load-bearing for the strategic failure causation of the episode — the operative failure causation mechanism was located in Structure, Capability, Culture rather than in the directional choice itself. Direction is therefore recorded at zero per cent on the rationale of modality acknowledged in narrative but not load-bearing for the strategic value created in the episode. Categorisation under METHODOLOGY-ota-scoring-v4.md §5: modality acknowledged in narrative but not load-bearing.

Structure. AIGFP was organised as a London- and Connecticut-based derivatives subsidiary of the AIG parent, with its own profit-and-loss reporting and a track record that the parent had come to rely on for earnings contribution through the early 2000s. Consolidated supervision sat with the Office of Thrift Supervision rather than with a derivatives-dealer prudential regulator; the FCIC record documents that OTS did not independently surface the collateral-trigger exposure embedded in the existing book. The governance architecture linked AIG parent credit ratings mechanically to AIGFP's collateral-posting obligations through the CDS contract terms, but the parent-level risk function that would have been positioned to re-read that linkage after the 2005 stop-writing decision did not do so in the FCIC record.

Processes. The collateral-posting mechanics on the existing CDS book were tied to two triggers — AIG's own credit rating and the mark-to-market price of the referenced CDOs — each of which had standard operating treatments in dealer treasury functions at counterparty banks. The FCIC collateral-call timeline (FCIC staff memorandum, 2010) records a mid-2007 onset of counterparty mark-to-market collateral calls, contested through ISDA-style dispute channels; by 12 September 2008 the cumulative call stood at $23.4 billion and reached $32 billion by 15 September. The re-review process that would have taken the 2005 observation-signal (the subprime share in the book was higher than AIGFP modelling had assumed) and converted it into a mark-to-market-and-rating-trigger liquidity stress test on the existing book is not documented in the FCIC record or in the McDonald-Paulson reconstruction.

Scoring note (zero-modality rationale): the Processes contribution described in this subsection is classified at the boundary with Capability in the scoring record — the §4 evidence locates the operative driver of the episode's failure causation in Capability rather than in a standalone Processes contribution. Processes is acknowledged in narrative as evidenced but does not carry independent weight in the scoring; weight is borne by Structure, Capability, Culture. Categorisation under METHODOLOGY-ota-scoring-v4.md §5: classification boundary with an adjacent modality.

Capability. AIGFP had the quantitative and derivatives-structuring capability that super-senior CDS origination required, and the parent AIG's actuarial and catastrophe-risk capability was deep in its insurance domain. The capability gap visible in the record is adjacent: the collateral-mechanics-and-liquidity framing that large derivatives-dealer counterparties applied routinely to the same contracts — documented in the Sjostrom Washington and Lee Law Review analysis and in the McDonald-Paulson NBER reconstruction as the frame AIG's counterparties were using — was not the frame AIGFP or the parent risk function applied to the existing book. The FCIC risk-officer testimony (30 June 2010 hearing record) describes the internal treatment of these contracts as catastrophe-insurance-equivalent rather than dealer-book-equivalent.

Culture. The internal cultural pathway visible in the documentary record runs through AIGFP's earnings track record and the deference the parent extended to the unit on that basis. The 2005 Park observation-signal did travel — Cassano accepted it in the narrow sense of halting new writing — which indicates that internal dissent could reach the unit head on a forward-looking question. What the record does not show is the same pathway carrying a re-read of the existing $78 billion book; Lewis's 2009 reconstruction and the FCIC record describe Cassano's posture on counterparty marks during the 2007–2008 dispute as contesting the external marks rather than updating the internal valuation frame. The cultural evidence therefore sits at the boundary between the unit and the parent risk function, in the deference the parent extended to AIGFP's book on the strength of prior earnings.